Amazon (AMZN) never ceases to amaze. The company’s custom chip business has surpassed a $25 billion annual revenue run rate, driven by the fast-growing demand for its Trainium AI accelerator chips and Graviton central processing units (“CPUs”) for cloud computing.

The reason is simple. Amazon Web Services (“AWS”) custom chips are more affordable to use, and its customers want lower-cost chips than what Nvidia (NVDA) offers.

Amazon is already the runaway e-commerce leader in the U.S., with a 40.5% market share. Meanwhile, AWS – which essentially created the cloud-computing industry – is the world’s largest cloud-infrastructure provider, with 28% global market share as of midyear.

But are there enough value-focused custom-chip customers for AWS to take a big enough bite out of Nvidia’s dominant AI-chip market share of roughly 80% to 87%?

The Lower-Cost Appeal of AWS Custom Chips

The most obvious appeal here is affordability. Sometimes, you can get high quality at a fair price, and that seems to be the case with Amazon’s chips.

AI developers who use AWS chips can save 30% to 40%, while still getting excellent performance, according to Dave Brown, AWS vice president of compute and machine learning.

As he told Yahoo Finance, “That’s what our customers are looking for, is to constantly get more compute and more performance, and then super importantly, at a lower price.”

Who wouldn’t want that combination?

Amazon notes that it builds its chips “from the ground up” to accommodate specific workloads. The end products are Amazon’s Trainium AI chips, Graviton cloud-computing chips, and Nitro chips, which provide the storage, networking, and security behind AWS’s cloud infrastructure.

So, what makes Amazon’s custom-built AI chips different from the graphics processing units (“GPUs”) that Nvidia creates? GPUs are designed for general-purpose workloads, not specific workloads. Amazon’s AI chips are custom-designed to handle the specific workloads its customers run.

Now, Amazon is not the first company to build custom chips for its clients to use. Alphabet (GOOGL), for example, was an early pioneer in that area, with its customized tensor processing unit (“TPU”) chips that deliver GPU-like performance for less money.

And money matters when it comes to AI, considering global AI spending is expected to reach nearly $2.6 trillion in 2026, according to research by Gartner. Any money businesses can save on AI costs is a win, with investors growing weary of tech companies’ wild spending sprees – and dinging their stock prices as a result.

Even a modest improvement in AI performance and efficiency can save a company millions of dollars a year. CloudExpat, a cloud-infrastructure optimization company, tested AWS Trainium1 chips against Microsoft (MSFT) Azure’s ND H100 and Nvidia GPUs in 2025 (Google’s TPU was also tested, but for this article, we’re focusing on the comparison between AWS and Nvidia).

Using dozens of highly technical tests, CloudExpat revealed that companies looking to train a GPT-4-scale AI model could save millions in cloud spend by using AWS Trainium chips versus Nvidia GPUs.

And Amazon has continued to improve the efficiency and price-performance of Trainium chips with each new generation. Amazon’s Trainium2 is reportedly four times faster than Trainium1 and cut costs by roughly 35% compared with Nvidia’s H100 chips, according to the Tech Buzz. Amazon says its Trainium3 chips offer up to 40% better price-performance than Trainium2.

And that’s just for AI training. When it comes to actually running AI models, where cost per query can add up quickly, Amazon says its Inferentia2 chips offer up to 50% lower cost per inference than GPU-based chips. When you consider that AWS sees billions of inference queries every day, that makes a massive difference.

Of course, there are other considerations here… For example, adopting AWS Trainium chips to replace Nvidia GPUs would require a company to abandon Nvidia’s very sticky – and familiar – CUDA ecosystem and implement AWS’s Neuron platform. And that’s not free.

But the chip performance itself is likely worth the swap.

A Growing Customer List Has AWS Chips Sold Out for Years

Companies are increasingly noticing the benefits of Amazon’s custom chips and signing up as customers. In February, OpenAI committed to using 2 gigawatts (“GW”) of Trainium capacity to power its frontier models starting in 2027. This development expanded the companies’ existing $38 billion multiyear agreement by $100 billion over eight years.

AI giant Anthropic followed suit in April, committing to spend more than $100 billion over the next 10 years on AWS technologies – primarily its Trainium and Graviton chip capacity. Amazon noted that Anthropic will “secure up to 5 [GW] of capacity to train and power their advanced AI models, including significant Trainium3 capacity.”

But it’s not just the big players investing in Trainium. Smaller startups such as TwelveLabs, Neura Robotics, Odyssey, Poolside, Decart, Karakuri, NetoAI, Splash Music, and Metagenomi Therapeutics (MGX) are among the many companies turning to AWS custom chips as a cheaper alternative to Nvidia.

Due to this demand, Trainium3 chip capacity is practically sold out. And AWS’s Trainium4 chips, which won’t even be released until 2027 or 2028, already have much of their capacity booked.

Even Amazon’s Graviton processors are flying off the shelves. According to Amazon, more than 130,000 customers use Graviton-based servers. And more than half of AWS’ new processing power runs on Graviton chips.

In April, Meta Platforms (META) became one of the world’s largest Graviton customers when it signed a deal to deploy tens of millions of Graviton cores to power its agentic AI workloads.

And global rideshare leader Uber Technologies (UBER) heavily employs AWS Trainium3 and Graviton4 chips to improve its infrastructure and make customer experiences smarter and faster.

Like its Trainium counterparts, Graviton processors are gaining popularity thanks to their significantly better – around 40% – price-performance and 45% savings compared with x86-based CPUs. The Tech Buzz estimated that, based on cost-optimization analyses from cloud-management platforms, “a typical Fortune 500 company running 10,000 instances on AWS could save $8-12 million annually by migrating from Intel-based instances to Graviton3.”

Amazon’s custom chips clearly have tons of momentum, which it rode to the milestone $25 billion-plus annual revenue run rate.

What Amazon Custom Chips’ $25 Billion Annual Run Rate Means for the Company and the Industry

Calling the $25 billion-plus annual revenue run rate impressive would undersell its importance. Amazon reaching this figure is changing the semiconductor and AI-infrastructure industries. Just think about it… AWS went from using someone else’s chips and selling access to those chips through the cloud to creating and selling capacity for its own silicon.

For less than the cost of an Nvidia GPU, Amazon can customize chip hardware to optimize specific types of workloads. That is a massive development.

Just as massive: Amazon is no longer solely dependent on paying companies like Nvidia or Intel (INTC) for their hardware. AWS does still buy – and will likely continue to buy – other companies’ chips to satisfy customer demand. But Amazon’s custom chips now make it a direct competitor in the market.

AWS’s 2026 second-quarter financials prove that its custom chips have had a huge impact on the semiconductor market. The results include:

• $42.2 billion in net sales, a 37% year-over-year increase, which was the fastest segment growth in 18 quarters

• More than $25 billion in annual revenue run rates for its custom-chip and AI businesses, with year-over-year growth in triple digits

• $16.6 billion in operating income, up 64% year over year

• An operating margin of 39.4%, up from 32.9% during the second quarter of 2025

• An annualized revenue run rate of $169 billion

Amazon’s custom-chip success adds even more revenue to the cash-generating machine that is AWS. As profitable as Amazon is beyond technology (remember, Amazon is far and away the No. 1 e-commerce retailer in North America), AWS accounted for roughly 60% of the entire company’s operating profit in 2025.

Where does Amazon’s chip business stand against heavyweights like Nvidia, Advanced Micro Devices (AMD), and Intel? Before comparing, it’s important to note that Amazon CEO Andy Jassy has said that AWS custom chips would double their current $25 billion revenue if the company sold its physical chips – not just chip capacity – to external customers.

In his annual letter to shareholders published in April, Jassy noted:

If we were a standalone chip company, our chips would be generating over $50 billion in annual revenue.

So, when comparing revenue with its competitors, Amazon is right there… even if its chips are not currently generating more than $50 billion in annual revenue. Nvidia remains the clear leader, with its AI-driven Data Center segment generating a company-record $89 billion during its most recent quarter (fiscal second quarter of 2027).

But Amazon is running neck and neck with Advanced Micro Devices and Intel. Advanced Micro Devices’ projected annual data-center run rate of roughly $26.9 billion (based on its actual $6.72 billion second-quarter revenues) sits just ahead of AWS’s $25 billion. And Intel, with a projected $25.04 billion annualized run rate based on its $6.26 billion second-quarter Data Center and AI segment revenue, is virtually tied with AWS.

If you factor in Jassy’s statement that Amazon’s chip business would be valued at roughly $50 billion if it sold its hardware externally, then Amazon blows right past both Advanced Micro Devices and Intel.

Is Amazon’s Stock Undervalued?

If you’re thinking about investing in Amazon, these chip developments are crucial when you consider that AWS generates most of the company’s operating profit every year. And there’s no reason to think the segment won’t keep growing rapidly.

Not only are new chips here or on the way (Graviton5 was released in June, and Trainium4 will arrive in the next year or so), but powerful frameworks and models are, too.

• Amazon Bedrock AgentCore, which was released last fall, is the system AWS uses most to build and orchestrate enterprise AI agents.

• Project Rainier in Indiana, which officially launched last October, is one of the world’s largest distributed AI supercomputing clusters. It currently hosts more than 1 million Trainium processors dedicated to training Anthropic’s frontier Claude models.

• AWS Continuum, released in June, is a new agentic security service that finds and fixes security risks.

• AWS Transform is an agentic-AI service that supports large-scale framework modernizations.

• AWS Context is another new service that automatically builds a knowledge graph from existing data to be used by a company’s agents.

That’s just the tip of the iceberg. And there’s no reason to believe AWS will stop innovating.

Amazon has been a wildly successful company for years. A major reason is the foresight it employs in developing brilliant strategies… like developing its own chips. This has already proved to be a market changer.

That kind of innovation and confidence in its strategies matters to investors. It’s one reason Amazon’s stock feels undervalued at around $260. The average price target for Amazon is roughly $327, giving the stock around 25% upside.

Amazon stock may not get the splashy spikes like Micron Technology (MU) or Sandisk (SNDK), but it has long been a consistent performer. Year to date, it’s up nearly 15%. And over the past five years, it has gained nearly 57% (largely driven by AWS).

Chaikin Analytics, the investment-research platform founded in 2009 by legendary 60-year Wall Street veteran Marc Chaikin, gives Amazon a “bullish” rating in its Chaikin Power Gauge, a 20-factor stock-rating system that scans more than 5,000 stocks and 2,300 exchange-traded funds.

Amazon has a long history of innovation and revolutionizing entire industries. It has done that with Amazon Prime and its e-commerce business, its logistics operations, and its enterprise cloud infrastructure. The company’s development of custom AI chips and processors certainly seems to be the next industry changer.

Regards,

David Engle

Editor’s Note: Whitney Tilson called the rise of Apple, Amazon, and Netflix… as well as the collapse of dozens of companies that went bankrupt. Now the former $200M hedge fund firm manager is stepping forward with what he calls the most important financial warning of his 30-year career. He’s sharing two free stock recommendations  (one to buy, one to sell immediately)  along with details of a new proprietary system fueling his predictions. See it all in his free presentation. 

The post Amazon’s AWS Chips Top $25 Billion Run Rate. Does This Signal Trouble for Nvidia? appeared first on MarketWise.

Nvidia (NVDA) reported strong earnings on Wednesday. In its latest quarter, the artificial intelligence (“AI”) chipmaker grew sales an astonishing 106% year over year to $96.2 billion, while operating income surged 124% to $63.7 billion.

Nvidia puts up numbers each quarter that seem impossible to fathom. And the fact that a $5 trillion company can grow this much is something special.

While any company would love numbers like that, investors are always looking ahead, and some believe things can’t get any rosier for Nvidia than what they’re seeing now. That’s a big reason why we’ve seen the stock plateau in recent months.

The upshot is that Nvidia is losing some levers that have kept its massive profit machine growing at the pace it has sustained over the past few years. At the same time, its willingness to finance massive projects is swelling its own balance sheet, even as hyperscalers are borrowing billions to buy its chips – both of which call into question the ongoing sustainability of Nvidia’s tremendous sales growth.

For example, even as sales have exploded higher, Nvidia’s gross margin has seemed to top out.

Period	2022	2023	2024	2025	Q1 2026	Q2 2026
Sales	$27 billion	$60.9 billion	$130.5 billion	$215.9 billion	$81.6 billion	$96.2 billion
Gross margin	56.9%	72.7%	75%	71.1%	74.9%	75%
Operating margin	15.7%	54.1%	62.4%	60.4%	65.6%	66.2%

Gross margin appears stuck at 75% and seemingly unable to expand above that level. So, future profit growth will rely primarily on sales growth, without the added lift of gross margin expansion.

Nvidia has already warned that surging memory costs are hitting margins, likely pushing them to 71% to 72% in the fourth quarter. Next year, however, they should bounce back to margins of 72% to 73%. This sensitivity to change shows how tough it will be for Nvidia to exceed its historic 75% level.

It’s a similar – but slightly different – situation with operating margins, which continue to grow from already phenomenal levels, albeit more slowly than before. If sales keep growing and Nvidia keeps a hold on costs, it will likely still be able to wring a few more points of operating margin. But the days of 8.1 year-over-year percentage-point gains Nvidia saw in 2024 are gone.

Nvidia’s bleeding-edge products keep it in the lead for now. Still, everyone wants a piece of these fat margins, from established players such as Broadcom (AVGO) to upstarts such as Cerebras Systems (CBRS), which makes wafer-scale chips for AI data centers.

None of this suggests that the near future won’t get better for Nvidia, but the ability for it to get much better has been diminished, since margins can’t expand at yesteryear’s levels anymore. But this is only one part of the concerns suggesting that Nvidia must use new levers, such as more aggressive financing, to keep its sales growth on track.

(Looking for a step-by-step plan to help you use AI in your own investing? Here’s how the N.E.W. System works.)

Nvidia Is Aggressively Financing the AI Boom, Which Comes With Higher Risks

Borrowing behind the AI build-out has surged, especially in 2026. Hyperscalers such as Amazon (AMZN), Microsoft (MSFT), Alphabet (GOOGL), and Meta Platforms (META) have spent nearly all their 2026 cash flow on capital expenditures, most of it on the AI build-out.

To keep that build-out on pace, these hyperscalers have each started borrowing tens of billions of dollars in the past year. With hyperscalers’ AI spending projected to top $1 trillion in 2027 – well beyond their estimated cash flow – they’ll need to borrow even more money.

While that increased financing means the spending spree for Nvidia’s chips won’t keep growing forever, the chipmaker has begun offering more short-term financing on its own balance sheet.

That is, Nvidia is offering more generous payment terms to larger customers, giving these more important clients up to a year to pay. CFO Colette Kress said about 25% of Nvidia’s 2027 sales could be to AI labs (such as OpenAI) where Nvidia is using its own balance sheet.

You can see this change in a metric called accounts receivable, which has soared in recent years, as Nvidia finances more sales, typically for short periods. The average is around 60 days, or about two months. While the figure dipped in the first quarter (not necessarily a bad thing), it rebounded hard in the second quarter, though only to its historical levels.

Period	2022	2023	2024	2025	Q1 2026	Q2 2026
Receivables	$3.8 billion	$10 billion	$23.1 billion	$38.5 billion	$40.7 billion	$63.1 billion
Sales	$27 billion	$60.9 billion	$130.5 billion	$215.9 billion	$81.6 billion	$96.2 billion
Days sales outstanding	51.8	59.9	64.5	65	44.9	59.9

To judge these receivables, analysts look at days sales outstanding (“DSO”), which is the ratio of accounts receivable to a quarter’s average daily sales. DSO shows whether a company may be expanding credit to less worthy customers to keep its sales growth continuing.

This trend isn’t worrisome yet, but it’s worth watching as Nvidia expands its financing, especially to clients with major financing issues such as OpenAI, which is hemorrhaging billions in cash each quarter. Of course, it’s not just OpenAI that could hurt Nvidia, but even established players that rely on OpenAI remaining solvent to continue moving billions in AI spending through the system.

After such a consistent trend for years, a sudden rise in DSO may offer investors an early read on how the business is going. Nvidia is already warning investors that the number should rise – and it’s a sign that it must move more of this financing to its own balance sheet to keep sales rising. If credit problems arise, they will likely appear in rising days sales outstanding figures.

That puts greater credit risk on Nvidia, drawing it even closer to less well-financed players.

More broadly, you can see Nvidia rushing to keep its sales growth on track, as it becomes the “central bank of AI.” It has lined up a $250 billion backstop for OpenAI to construct a massive data center as well as a $350 billion financing arrangement for the facility’s chips.

Nvidia also announced an incomplete $500 billion deal with major investors before it had been finalized. In the end, it’s less a deal than a memorandum of understanding – “a deal to do a deal” – despite the breathless media headlines making it seem like the terms were in place.

Despite this quarter’s excellent numbers, Nvidia is pushing hard to line up financing for its customers and keep all its plates spinning. With each passing quarter, the growth will become harder to sustain, so watch for a rise in DSO as an early warning sign of trouble.

Nvidia’s Stock Looks Undervalued, But Is It Really?

On the surface, Nvidia looks cheap for a company growing at such a brisk pace. In May, various analysts raised stock price targets based on Nvidia’s rising earnings for 2026 and 2027, due to surging AI investment.

For example, Wells Fargo raised its price target from $265 per share to $315 on earnings estimates of $8.45 per share this year and $11.95 per share for calendar year 2027. This level of explosive growth implies a price of less than 19 times next year’s earnings at a share price around $220.

This price-to-earnings (P/E) ratio is well below the stock’s 32 times earnings over the past three years, says Wells Fargo.

In absolute terms, this forward P/E ratio looks quite reasonable for such high growth. But it assumes that growth will continue, even as Nvidia desperately works to provide financing for its clients. Some of Nvidia’s levers for growing profits are becoming less powerful, too.

So this assumption of growth will look fine until one day it suddenly won’t.

It’s less a question of “if” than “when” the Nvidia juggernaut can no longer sustain the high expectations it’s setting now.

Arguably, the world’s most-watched company is trading for what looks like a value. If you believe in the wisdom of the market, you want to be quite sure why it’s putting the company – and many high-growth AI plays – in the bargain bin. They may not be as undervalued as they appear.

Regards,

James Royal, PhD

Editor’s Note: Whitney Tilson — the hedge fund manager CNBC called “The Prophet” — says America has reached its “Ripping Point.” The old financial order is being torn apart, and he believes most investors have no idea what’s coming in the next six months. He’s named the stocks he thinks will be destroyed in the chaos — and the ones he believes will soar. Watch his free presentation while it’s still available.

The post Nvidia’s Blockbuster Earnings: Here’s the Biggest Risk the AI Chipmaker Faces, Even as Sales Soar appeared first on MarketWise.

A $2.6 billion loan just told you more about the AI buildout than any earnings call. The stock already recovered and forgot. The loan document remembers — and it’s the most honest signal in the whole trade.

Twenty years on a trading desk teaches you where to look when you want the truth about a trade.

Twenty years on a trading desk teaches you where to look when you want the truth about a trade.

You can’t use stock prices, they lie all the time. They gap on emotion, they get pinned by options dealers, they run on narrative long after the story has changed.

That’s especially true when you’re evaluating AI infrastructure stocks, where billion-dollar financing decisions often reveal more than earnings calls or analyst price targets ever will.

If you want to know quickly whether a company is actually in trouble, you watch what it costs that company to borrow money.

The lenders behind the money have done the work. They’ve read the contracts. They’ve stress-tested the cash flows. And when they suddenly demand more interest, tighter terms, and a lockbox on the revenue before they’ll hand over a dollar — that’s the smart money repricing risk in real time.

This is exactly what’s happening with one stock at the center of the next AI wave — CoreWeave Inc. (CRWV).

AI Infrastructure Stocks’ Hidden Debt Bomb

In mid-July, CoreWeave — the “neocloud” that rents out Nvidia GPUs to AI companies — went to the leveraged-loan market to raise $2.6 billion.

At first, the loan looked routine.

The structure was a delayed-draw term loan, its fifth-and-a-half such facility, nicknamed DDTL 5.5. The proceeds were earmarked to buy more GPUs and infrastructure to serve a specific set of customer contracts.

According to loan-market reporting from LSEG and Octus, the deal launched on July 16 at a spread of SOFR plus 425 to 450 basis points, offered at 99 cents on the dollar. Standard-issue for this borrower.

Then it stalled.

The lenders investing in the project weren’t convinced. And to get the deal across the line, CoreWeave had to rewrite the terms in the buyers’ favor:

• CRWV agreed to pay a much higher interest rate, making the loan significantly more expensive. The spread widened roughly 100 basis points to SOFR + 550, and the loan was discounted to around 96 cents — a four-point cut. The Financial Times pegged the all-in yield at roughly 9%. That’s a junk-grade cost of capital for a company that, months earlier, had raised GPU-backed money at investment-grade rates.
• The lenders demanded stronger protection. They imposed a cash lockbox — meaning the cash coming in from customer contracts has to go to paying down the debt first, before CoreWeave can touch it — plus a 1.35x debt-service-coverage covenant and full amortization, per Fitch and loan-market coverage. Translation: the lenders wanted a hand on the cash register.

Only after making those concessions was CoreWeave able to secure the financing.  When the deal finally closed around August 10, the company described it as “meaningfully oversubscribed,” meaning more investors ultimately wanted to participate than the company needed.

At first glance, that sounds like a sign of overwhelming confidence, but remember when that demand appeared.

Investors only lined up after CoreWeave caved to the lenders’ demands. In other words, they’re there for the deal,

This was echoed by the independent credit-rating agencies. Fitch rated the loan BB+ and Moody’s rated it Ba2 — both below investment grade.

The deal did get done. But the market forced CoreWeave to pay up, pledge its cash flows, and accept a leash — before it would fund the AI buildout everyone insists is unstoppable.

The major takeaway here?

When a trade is truly bulletproof, the financing is supposed to get cheaper. But this one got more expensive, more restrictive, and harder to place, all at once.

The Stock Already Forgot. The Loan Remembers.

Here’s the part that makes this worth your attention right now.

In late July, fear peaked. The cost of insuring CoreWeave’s debt against default — its credit default swap, or CDS — blew out. Per Bloomberg data cited across market coverage, CoreWeave’s CDS topped roughly 855 basis points, a level that implied around a 50% probability of default within five years on a standard pricing model. The stock bottomed near $60.

Then the mood flipped. The loan cleared. On August 12, CoreWeave reported second-quarter revenue up more than 200% year-over-year, with a backlog it pegged at roughly $104 billion.

The stock ripped — nearly doubling off the lows to around $117 by mid-August before settling back into the high $80s.

The sell-side climbed aboard: Truist moved to Buy with a $165 target, Goldman lifted its target to $139, and Oppenheimer kept an Outperform and a $150 target, arguing AI demand runs about four times available capacity.

It only took the equity crowd about three weeks to get its optimism back. Understandable. But notice what didn’t reverse: CoreWeave still had to accept a 9% all-in cost, a cash lockbox, a maintenance covenant, and full amortization to get that money.

Those terms are baked into a signed credit agreement. They don’t un-widen because the stock bounced. The bond market did its stress test in public, and the price of that test is now a permanent line item in their budget until it’s paid in full.

This is the oldest lesson on the desk. Think of it like this…

The stock market is a voting machine — it forgot the scare in three weeks. The credit market is a weighing machine — it wrote the risk into the contract and moved on.

When those two machines disagree this much, I’ve learned which one to trust.

The Three Key Tells Buried Inside the Deal

Pull the DDTL 5.5 apart and three specific risks explain why the lenders got nervous. Each one is a flashing light that’s still on.

Tell #1: The Durations Don’t Match.

CoreWeave’s problem is the oldest one in finance — borrowing long and getting paid short. This loan runs about five years. But the customer contracts backing it average only about three. And the underlying data-center leases CoreWeave signs with its landlords can run up to 15 years.

So CoreWeave commits to 15-year obligations, funds them with a five-year debt, and covers it with three-year revenue. If those customer contracts don’t renew, the bills keep coming with no guaranteed income behind them.

Fitch flagged exactly this. Visibility gets murky beyond the current contracted window, when the company leans on renewals and re-leasing to sustain growth. If that pattern rings a bell, it should. It’s the same asset-liability mismatch that took down WeWork — long leases, short revenue, and a growth story papering over the gap.

Tell #2: The Customers are Dangerously Concentrated.

The Financial Times reported that the single largest counterparty behind this loan is Anthropic, at roughly 40%, followed by the quant trading firm Jane Street at about 35%, with Midjourney, Hudson River Trading, and Anysphere (the maker of Cursor) making up the remainder. Fitch separately noted that CoreWeave drew about 65% of its first-quarter 2026 revenue from just its top two customers.

Two things stand out. First, that is enormous concentration — a handful of contracts carrying the whole structure. Second, and more pointedly: Anthropic, the 40% anchor, is a privately held company with no public credit rating.

That’s not a knock on Anthropic — it’s one of the best-funded labs in AI, and in August it was being called the hottest upstart in the field. It’s a statement about what the lenders can actually underwrite.

You cannot price the credit risk of a counterparty you cannot rate.

When 40% of your collateral rests on an unrated private company and another 35% on a proprietary trading firm, the margin for error is thin under the best of circumstances.

There’s a footnote worth watching. As bond analyst Rod Dubitsky flagged, Fitch upgraded Jane Street — CoreWeave’s #2 counterparty — right in the middle of the period when this struggling loan was being marketed, citing factors that had been in place for years. Draw your own conclusions, but the timing feels suspicious.

Tell #3: The Lockbox is the Lenders Telling You What They Think.

You don’t demand a cash lockbox from a borrower you trust. You demand it when you want to make sure you get paid before anything else can go wrong. The concession CoreWeave had to grant is itself the verdict.

Not All AI Infrastructure Stocks Are Built the Same

While CoreWeave may be the loudest canary in the coal mine among the AI infrastructure stocks, they’re far from being alone.

The entire AI buildout runs on borrowed money and depreciating collateral.

Neoclouds like CoreWeave and Nebius fund their GPU purchases with debt secured against the chips themselves  —  Nvidia hardware that’s valuable today but could be a generation behind tomorrow.

CoreWeave’s own guidance calls for $31–35 billion of capital spending in 2026 against roughly $6.2 billion of trailing-twelve-month revenue.

Its first-quarter interest expense doubled to $536 million, and it has burned free cash flow every year since 2022. Revenue is expanding, yes — but so is the spending required to produce it. That treadmill only works as long as the cheap money hose keeps spitting out cash.

There’s a circularity here that should make you tilt your head: Nvidia sells the chips, is a major shareholder in CoreWeave, and is part of the capital ecosystem funding the customers who buy those chips.

When the same company is supplier, shareholder, and financier, “demand” gets harder to read at face value.

 That’s why “AI infrastructure” is too broad to be a useful investment category. The companies may be building toward similar goals, but they’re operating under very different financial constraints. The way I see it, the easiest way to think about them is to divide them into two groups:

The self-funders. Amazon, Alphabet, Microsoft, and Meta are pouring hundreds of billions into the buildout too — but largely out of operating cash flow from businesses that already mint money.

Their existing profits subsidize the unprofitable early years of AI. If financing gets expensive, they can shrug.

The debt-funders. CoreWeave, Nebius, Oracle’s cloud arm, and the leveraged neocloud complex finance the same buildout with debt, leases, and project loans against short contracts and depreciating chips.

For this bucket, the cost of money is the business model. When spreads widen, their math breaks first. This isn’t limited to the neoclouds: Oracle — the largest non-financial borrower in the Bloomberg U.S. high-grade index — saw its own CDS climb above 215 basis points this year, up from about 145 at the end of last year, with its long bonds yielding close to 7.8%. Apollo’s chief economist Torsten Slok put a name on the danger: at high enough all-in yields, the AI capex cycle starts to self-throttle not because demand changes, but because financing becomes too expensive.

And the hyperscalers just made the debt-funders’ problem worse.

In mid-July, Meta announced “Meta Compute,” a service to sell bare GPU capacity to third parties — the exact business the neoclouds are built on. When your biggest potential customer becomes your competitor, the “customer concentration” and “insourcing” risks are no longer theoretical.

I call it my AI Destruction List. Big media loves to oversimplify the problem. They say “AI is a bubble” — but that’s too lazy to be useful for anyone, and it’s impossible to time.

The more precise version: the leveraged, debt-funded layer of the AI trade cracks first, and it cracks in the credit market before it cracks in the stock.

Is There Any Bull Case for CoreWeave?

Now that we’ve covered the problems, allow me a moment to argue against myself. After all, a one-sided case is a weak case — and I want you thinking, not panicking.

The bulls have real, current points. The loan did clear, oversubscribed at the wider terms — the demand exists, just at a higher price.

Many of CoreWeave’s contracts are “take-or-pay,” meaning customers owe the money whether or not they use the capacity. That ~$104 billion backlog is enormous (bear in mind it blends remaining performance obligations with other committed amounts, so it’s a scale indicator, not a bank balance).

The company is actively working to diversify the concentration I just flagged — in August it signed a fresh multi-year deal with Hudson River Trading and pushed into sticky government work through a CoreWeave Federal partnership with Leidos.

Its marquee counterparties — Anthropic, Meta, OpenAI, Jane Street — are real, growing, well-capitalized names.

The insider selling that made headlines was executed under pre-arranged 10b5-1 plans and left executives with large stakes.

And most of Wall Street is bullish, with price targets from $139 to $165 against a stock in the high $80s. A CDS spread is a probability, not a verdict — and the one that screamed in July has since calmed as the stock recovered.

Both readings are defensible from the same set of facts depending on which details you choose to focus.

The honest position isn’t “CoreWeave is going to zero.” It’s that the cost of financing the AI buildout went up in a visible, measurable, contractually permanent way — and the market charged the most leveraged players the highest prices.

What is critical to understand before committing your capital to this trade is that ignoring that risk because the stock bounced is exactly the mistake retail makes at every top.

Forewarned is forearmed.

What the Options Say About CoreWeave

People always want the trade. But before we take the trade, we need to know the range of what to expect — and on a name like this, the range tells us the story.

Every day I run a tool I built called the Volatility Visualizer, or the Vol Viz for short. This program plots the expected-move band the options market is pricing in for where a stock could realistically trade over a given window.

To be clear, this isn’t an arbitrary line I drew on a chart. This is hard data, pulled straight out of live options prices — the market’s own, dollar-weighted forecast of how far this thing can move.

Here’s what it’s showing on CoreWeave, with the stock around $88.50: an expected-move band running from roughly $69 on the downside to $111 on the upside — a range more than 40 points wide, better than 20% in either direction.

Sit with that. The options market is pricing a swing of more than a fifth of the company’s value, up or down, on a business worth tens of billions of dollars.

That’s an unusually wide range of expected outcomes. And it echoes the same warning coming from the credit market: investors are demanding a much bigger margin for uncertainty.

The lenders demanded a cash lockbox and 9% money because they couldn’t pin down the risk. The options market is pricing a 40-point range for the same reason.

Two completely different corners of Wall Street — credit and volatility — look at CoreWeave and reach one identical verdict: nobody knows where this settles and they’re actively worried about where it might end.

Why the Expected Move Matters More Than Any Price Target

A sell-side analyst stamps a $165 target on the stock and a headline is born — but that’s one person’s guess about direction.

The expected move is the whole market’s consensus about magnitude, and it comes with probabilities attached. Roughly speaking, the market is pricing about a two-in-three chance the stock finishes inside that ~$69–$111 band over the window.

That single fact should reframe everything you’re looking at. CRWV ripping toward $110 next month? Still “inside expected” — not a breakout, not a miracle, just the top of a range the market already told you about. A decisive break below $69 or above $111 is the only thing that would count as the market being genuinely surprised.

That is how you keep your head in a stock that moves 15% before lunch. You stop flinching at every candle and start asking one question: is this move inside the expected range, or did something just break the market makers’ models?

Because the day CoreWeave slices through the low end of that band on heavy volume is the day the credit-market alarm bells stop being silent.

Five Signals I’m Watching Next

You don’t need to track every loan. Watch these:

The 10-year Treasury. With the long end elevated, the cost of all this debt is set by the bond market. Higher-for-longer is the neocloud’s enemy.

The next neocloud financing. The tell isn’t whether it gets done — it’s the spread and the structure. Wider pricing and more lockboxes mean the squeeze is tightening. Cheaper, cleaner deals mean the fear is fading.

CoreWeave and Oracle CDS. The real-time thermometer. Staying wide is the warning; the more it compresses, the smoother sailing there is ahead.

Contract renewals. The duration mismatch is critical. The first time a big three-year contract comes up and doesn’t renew at full value, the mismatch problems stop being theoretical.

Hyperscaler insourcing. Every “Meta Compute” — every hyperscaler that builds instead of rents — is a direct hit to the neocloud customer base.

The Bottom Line

The biggest mistake investors make with AI infrastructure stocks is assuming they’re all exposed to the same risks.

Financial talking heads love to go on about the AI trade as a contest of technology — whose model is smartest, whose chips are fastest. But at this scale, it’s a contest of capital allocation and financial endurance. The question is no longer whether AI demand is real in the short-term. It’s whether that demand converts to cash fast enough, and lasts long enough, to service the mountain of debt being stacked against depreciating hardware.

The stock market already voted twice — panic in July, euphoria in August. The credit market didn’t flip-flop. It wrote the risk into a contract and left it there: wider spreads, tighter terms, a cash lockbox, and a set of covenants that tell you exactly how the smartest lenders in the room see this. The stock forgot. The loan remembers.

The first crack in the AI trade won’t show up on a chart. It will be buried in the fine print of a loan document.

Always follow the money.

Editor’s Note: Forbes calls $1 billion fund manager Louis Navellier “the king of quants.” Today, he’s stepping forward to reveal why he’s investing $358 million of his own firm’s money in the next stage of Artificial Intelligence… a technological sea-change that could erase millions of jobs, solve humanity’s biggest mysteries, and spark a wave of moneymaking opportunities — both in and outside the stock market. Click here for the details… 

---

I break down the day’s biggest market story every morning on Masters in Trading Live, 11 AM ET. I follow the top stories of the day and show you how the smart money is really positioned — no geometry, no chart-drawing, just evidence of where the money is moving. Join me each morning, drop a “Good Morning” in the chat, and let’s read the tape together. Jonathan Rose is a former CBOE market maker and the founder of Masters in Trading at InvestorPlace. I hosts Masters in Trading Live every weekday at 11 AM ET. Follow me on X at @JRoseTrades

The post CoreWeave Stock: The $2.6 Billion Debt Signal the AI Bubble Is Ignoring appeared first on MarketWise.

It seems like just about everyone is mad about AI data centers right now.

They’re thought of as big, loud nuisances… and it’s hard to find people who actually want them where they live.

At the same time, we’re using artificial intelligence (“AI”) more than ever for searches, coding, writing, problem-solving, and even just chatting. So, demand for data centers is real.

Nonetheless, the AI data-center backlash we’re seeing has become big enough to matter to investors. Towns are pushing back over water, power bills, tax breaks, noise, diesel generators, and the simple reality that nobody wants a windowless server farm dropped into their neighborhood.

That’s a problem for the AI trade because Nvidia (NVDA) stock may be a proxy for AI growth, but chips still need to be installed in hardware which needs to be deployed into buildings, powered by transformers, and receive municipal permits before they become revenue. Nvidia is the top bellwether for AI, but it still relies on data centers for a majority of its chip sales.

It’s also creating a strange side effect in bitcoin mining…

See, while investors have been staring at Nvidia, the bitcoin mining network has been doing something odd.

Crypto-tracking site CoinWarz put bitcoin difficulty at 125.81 trillion today. About a year ago, on August 24, 2025, according to Hashrate Index, the difficulty was at 129.70 trillion… then an all-time high.

In plain English, the work required to mine a block today is roughly the same as it was last summer. That gives miners more time to make a return on their hardware investments without having to upgrade.

So thank you, AI data centers. And thank you to everyone saying they don’t want one in their neighborhood. It’s helping make certain miners a lot of money.

Today, we’ll look further at how the AI-infrastructure build-out and rising bitcoin prices are setting up a profitable scenario in the mining space.

Why Steady Mining Difficulty Is Favorable in a Bitcoin Rally

Bitcoin mining difficulty is the network’s thermostat.

Every two weeks, bitcoin adjusts how hard miners must work to keep blocks arriving about every 10 minutes. When more machines mine for the same reward, bitcoin difficulty rises. Fewer machines, or machines pointed somewhere else, means difficulty stabilizes or falls.

For a miner, that math is brutal and simple. Revenue comes from the bitcoin price, block rewards, and transaction fees. Costs come from power, machines, debt, and maintenance. If bitcoin jumps while difficulty stays flat, the same machines earn more dollars without facing the usual penalty of a harder network.

That’s exactly why the recent bitcoin rally matters. Bitcoin gained 23% in a week– its biggest move since last year’s sell-off – after Treasury-bond buybacks pushed long yields lower, the dollar weakened, and a short squeeze forced buyers back into the market. ETF inflows are the next test, because forced buying burns out quickly, while institutional flows can keep a rally alive.

If bitcoin holds even part of that move higher, miners get operating leverage when the selling price rises, and the production hurdle does not.

AI Pivots Are the New Way Miners Capitulate

In previous bitcoin cycles, miner capitulation looked familiar. The price fell, inefficient miners shut off, mining rigs hit the secondary market, and the network waited for difficulty to adjust down.

This cycle has a different escape hatch. Miners don’t have to disappear. Some have become AI-infrastructure companies.

That sounds like a branding trick until you remember what miners can offer to the AI space: power access, land, cooling, substations, operating teams, and a tolerance for absurd capital intensity. Those are the main things that data-center developers are fighting to secure.

CoinDesk reported in March that public miners had announced more than $70 billion in AI and high-performance-computing (“HPC”) contracts. S&P Global listed Iren, Riot Platforms, Core Scientific, Hive Digital Technologies, Cipher Digital, and TeraWulf among the miners shifting capacity away from pure bitcoin mining, with analysts expecting HPC and AI to drive much of their growth in 2026.

That matters because every megawatt chasing AI is a megawatt that’s not automatically chasing the next bitcoin block. While the machines and customers are different, the scarce input is the same: energized infrastructure.

The result is a softer version of miner capitulation. Instead of “we shut off because bitcoin mining stopped being profitable,” the message becomes “we found a better-paying customer for this power.”

AI Data-Center Backlash Cuts Both Ways

Now for the catch…

The same local revolt slowing the data-center land rush and helping bitcoin miners also threatens the miners trying to become AI landlords.

Governors and lawmakers are moving to force data centers to pay for their own electricity supply, limit water use, disclose more about operations, and win local support, according to the Associated Press.

Microsoft (MSFT) is already working to unwind multimillion-dollar tax breaks on several Atlanta data-center projects as the politics get hotter.

So, the trade is not “buy every miner with AI in the press release.” Please don’t do that. That’s how people end up owning a construction project with a bitcoin ticker.

Look for mining stocks that already control power, have credible tenants, and can survive construction delays while still making money mining bitcoin if their AI customer takes longer than promised to come online.

Global asset manager VanEck has already warned that miners chasing AI face a roughly $50 billion near-term funding gap and that only about a quarter of leased AI and HPC capacity has been delivered so far.

That means there’s still a long way to go.

What Investors Should Watch Next

Chips are no longer the bottleneck in the AI trade. Power is. Permits are. Local sentiment is. The boring parts are now in charge.

For bitcoin miners, that means three numbers are worth watching…

• Bitcoin difficulty: If difficulty stays near last year’s level while bitcoin holds its rally, miner margins improve without any miracle.

• ETF inflows: A short squeeze can light the match (and it really, really did last week), but sustained inflows tell you whether serious capital is returning to bitcoin. They’ve been positive since August 17, which is good, but previous rallies have run for as long as seven months, so time will tell.

• AI data-center approvals: Every delayed hyperscale project makes energized power more valuable. Some miners will monetize that value through AI contracts. Others will benefit because competitors will stop adding hash rate as aggressively.

That’s one thing many bitcoin investors are missing. While the AI data-center backlash is a risk to the AI build-out, it’s also putting a ceiling on one of bitcoin mining’s worst enemies: endless hash-rate growth.

If bitcoin’s price keeps firming while mining difficulty stays flat, it presents a solid opportunity for bitcoin mining stocks to profit from.

For once, the local zoning board is indirectly doing bitcoin miners a favor.

Good investing,

Eric Wade

Editor’s Note: Trillions of dollars — the largest flood of spending in U.S. history — are pouring into AI and energy. And one little-known company sits right in the path of it, collecting a toll on every dollar that flows past. It doesn’t make a single chip or pump a drop of oil. Whitney Tilson calls it “America’s Greatest Retirement Stock” right now — and today, it’s trading at a rare discount. He’s giving away the name for free. Click here…

The post Bitcoin Miners: A Surprising Winner From the Growing AI Data-Center Backlash appeared first on MarketWise.

The United States and Israel are at war with Iran. Ukraine is still fighting Russia. Those conflicts mean big money for global defense contractors and their investors… especially as this year’s global military budgets are expected to surge beyond last year’s nearly $2.9 trillion in world military expenditures.

That makes U.S. defense-technology contractor Lyntris’ (LYNX) downsized August 19 IPO rather interesting.

Lyntris, which manufactures battlefield sensors and designs software for the U.S. and its allies, began trading on the New York Stock Exchange last week after it raised $297.5 million in a downsized IPO at $17.50 per share for 17 million shares. That fell below initial company expectations of 24 million shares priced between $19 and $22 each.

What Caused Lyntris’ Soft IPO?

The short answer is debt and losses. As of the end of June, Lyntris reported $272 million in long-term debt, which severely hampers profitability. In fact, Lyntris plans to use roughly $60 million of its $297.5 million IPO proceeds to help pay off some of that balance.

On top of the debt, Lyntris has absorbed increasing operational costs. During the first half of 2026, the company’s net losses grew to $13 million, up from $9.7 million during the first half of 2025. That 34% growth in operating loss nearly negates the 35% revenue growth the company achieved during the first half of 2026.

Other factors contributed to Lyntris’ soft IPO, which helped bring the stock down more than 14% from its initial IPO price of $17.50 to $15.01 at market close on August 19.

“The market is discounting that this is a capital structure, rather than a venture capital deal,” claimed Josef Schuster, chief executive of Chicago-based IPO index and research firm IPOX Schuster.

That “capital structure” often includes heavy debt backed by private equity rather than a leaner, higher-growth venture-capital structure. In fact, private-equity firm Trive Capital created Lyntris by rolling Accelint and Vitesse Systems, two defense holdings, into one business. And that business, Lyntris, was saddled with $272 million in long-term debt from the get-go.

Beyond that, only 5.7 million of the 17 million shares sold during Lyntris’ IPO were actually offered to the public. The remaining 11.3 million shares, valued at roughly $197.8 million, were sold and cashed out by the company’s private-equity partners. While this is somewhat standard for private equity “roll-up” companies like Lyntris, the cashing out by partners still raises skepticism among investors.

That left Lyntris with $69.5 million in net proceeds… $60 million of which will be used to pay down its debt. That doesn’t exactly fill investors with confidence. Lyntris will have to prove to investors that it can pay off that debt to earn that confidence.

Finally, a combination of recent defense-technology IPOs – Applied Aerospace & Defense (AADX), HawkEye 360 (HAWK), and Aevex (AVEX), to name a few – and a general calming of the defense sector from its early 2026 highs created market sentiment that wasn’t ideal for an IPO.

But it’s not all doom and gloom for Lyntris.

In the first half of 2026, Lyntris generated $241 million in revenue, a 34.6% year-over-year increase. And the company more than doubled its backlog from the first half of last year to $924 million so far in 2026.

Still, investors have major concerns about Lyntris, specifically regarding its client list.

The Lyntris Concentration Risk

That client list I referred to is rather short. In fact, 97% of Lyntris’ revenue in 2025 came from U.S. government contracts – and nearly all of those contracts were with the federal government.

That includes equipment contracts with the Air Force, Navy, Marine Corps, and Space Force. Considering the Pentagon’s fiscal 2026 defense budget is nearly a trillion dollars, those contracts would appear to bode well for Lyntris. But plenty of risk comes when one client ostensibly generates more than 90% of your revenue.

Right now, as the war in Iran continues, Lyntris is a beneficiary. But what will happen (besides oil and gas prices dropping) once the war is over? Its significant backlog will likely sustain the company in the short term.

Plus, it’s not as if Lyntris is only supplying equipment and systems specific to this war. National security will always require the software and technology that Lyntris provides. And while the federal government represents most of the company’s business, Lyntris’ revenue is split across more than 200 active programs within the country. So, there is some measure of diversity.

But is it enough for investors? Because even a slight change in U.S. defense policy could negatively impact Lyntris. Plus, any delay in government contract approval would disrupt the company’s cash flow.

Most limiting, however, is that Lyntris has opted to pursue primarily domestic contracts. During the first half of 2026, non-U.S. contracts accounted for only 7% of the company’s revenue – which is actually up year over year from just 2%. Investors considering Lyntris should watch that number to see if it continues shifting in 2027.

Defense Stocks Outlook: Which Companies Have Outperformed?

The defense-tech industry as a whole is a risky proposition. It’s true that many major defense-contractor stocks – riding the tailwinds of the wars between Russia and Ukraine, Israel and Hamas, and the U.S./Israel and Iran – have risen sharply over the past year.

(Note: Performance data as of Aug. 25, 2026)

Defense Stock	Year-Over-Year Performance
Lockheed Martin (LMT)	24.3%
RTX (RTX)	34.6%
General Dynamics (GD)	17.8%
Elbit Systems (ESLT)	57.5%
GE Aerospace (GE)	31.1%

But that hasn’t fully translated to the small- and mid-cap defense-tech companies.

Defense Stock	Year-Over-Year Performance
Kratos Defense & Security Solutions (KTOS)	-21.3%
AeroVironment (AVAV)	-38.9%
Parsons (PSN)	-39.7%

A couple of outliers are worth noting. VSE (VSEC) has gained 31.8% over the past year, and Ducommun (DCO) has soared by an impressive 93.1% year over year.

As for defense companies that recently went public, like Lyntris, results have been decidedly mixed since their debuts:

Defense Stock	IPO Date	% Change Since Debut
HawkEye 360	May 7, 2026	-36.6%
Applied Aerospace & Defense	June 2, 2026	-30.9%
Aevex	April 17, 2026	-22.3%
Arxis (ARXS)	April 16, 2026	41.5%

Then there are Lyntris and First Breach (FBDT), which debuted within a day of each other. Since its August 20 debut, First Breach stock has plummeted from its opening share price of $12 down to $1.43 at market close on August 25 – a staggering drop of 90.5%. Lyntris, which debuted a day earlier, has dropped around 13.9% since August 19.

By this point, you’ve probably noticed that the performances of these defense stocks are all over the place. We’ve seen the large-cap defense companies make some huge year-over-year gains, with the notable exceptions of Northrop Grumman (NOC), which is surprisingly down 7.7% over the past year, and L3Harris Technologies (LHX), which is down around 3.4%.

The small- to mid-cap companies have run the gamut from Ducommun (up 93.1%) to Parsons (down 39.7%). And the defense companies with recent IPOs have largely been down, with just a couple of exceptions.

What does this say about the overall defense sector? Mostly that geopolitical events do not solely drive it. If they did, every defense stock would be soaring right now, and that’s clearly not the case.

A deeper look at the stocks in the industry reveals that, even as more defense companies take their stocks public and sign government contracts, the “institutional” stalwarts – like Lockheed Martin and RTX – hold a significant advantage over less-established companies.

That’s because these companies possess massive backlogs that span several years (RTX currently has a $289 billion backlog), use cost-plus government contracts that allow the companies to pass unforeseen costs (supply chain, inflationary, etc.) back to the government, and increase earnings-per-share (“EPS”) guidance and stock buybacks to ensure investor returns.

Meanwhile, defense companies stuck in the middle tend to struggle because fixed-price contracts handcuff them and put the burden of extra expenses on them rather than the government. If material prices rise or the supply chain bottlenecks, these companies absorb the added costs, which chokes their profitability.

Additionally, while the big defense companies are procuring huge, profitable contracts, these companies in the middle are receiving research and development funds and securing smaller deals because they lack the track record of handling large contracts. However, they can’t prove they can handle large contracts if no one gives them any.

It’s a chicken-or-egg conundrum known in the industry as the scale catch-22, which looks like this:

That results in less available cash, which means these companies must take on debt to finance expansion… again, a hit to these companies’ profitability.

Is Lyntris a Good Investment Right Now?

Which brings us back to recent IPOs like Lyntris. As a smaller company that is losing money, its IPO was deemed too ambitious by investors, and the company was already in debt the second it went public. And First Breach started hemorrhaging as soon as it debuted.

But that doesn’t mean these stocks are lost causes. Lyntris, after just a few days of public trading, is holding fairly steady amid an iffy defense-tech market. To inspire investor confidence, Lyntris will certainly need to start turning a profit in the second half of 2026 after posting a $13 million loss in the first half.

It will also need to reduce its long-term debt, which sits at more than $270 million. Its net interest expense alone during the first half of 2026 was $16.9 million.

The Falls Church, Virginia-based company’s 2026 revenue is promising, as is its $924 million backlog, which more than doubles its backlog from the end of 2025. But it has a long way to go before it becomes a profitable business and one that investors can rely on for consistent returns. Its soft IPO was just the latest sign.

Regards,

David Engle

Editor’s Note: It’s easy to assume the AI story begins and ends with the big-name chip stocks. But there’s a quieter version of it playing out inside the defense budget – where the government isn’t just regulating AI, it’s starting to buy into the companies that build it. 

A forensic accountant named Joel Litman – someone the Pentagon and FBI have actually consulted, and who called both the 2008 and 2020 crashes – argues this is one slice of a much bigger shift he calls the “Second Declaration.” He believes a handful of companies sitting where AI meets national defense could be repriced when a policy deadline arrives this November. 

Watch his full presentation here → [LINK] 

The post Lyntris’ Soft IPO Raises Red Flags. Is the Defense-Tech Stock Worth the Risk? appeared first on MarketWise.

SpaceX (SPCX) announced on Tuesday that it plans to open a $100 billion spaceport on the Louisiana coast. It’s a huge investment from a company that has offered up ambitious plans, such as going to Mars.

The new facility will be larger than the company’s three existing launch sites, and will include five launch pads, a mission-control unit, offices, a propellant factory, and power plants. SpaceX plans to begin construction in 2027, with the first launch scheduled three years from now.

The Louisiana launch site and SpaceX’s other sites would allow the company to increase its launch cadence, targeting 10,000 flights annually within five years across its various ships.

Such a goal remains ambitious. SpaceX just completed its 100th launch of Falcon 9 in 2026 on August 25. The workhorse rocket flew 165 missions during 2025. Those rockets have been carrying the company’s Starlink satellites, which provide global telecommunications services.

SpaceX also says thousands of those projected flights each year would involve its Starship rocket, which features reusable components that can drastically cut launch costs, “making rocket launches as routine as airplane travel.” Starship is still in its testing phase, however, and it hasn’t reached the high altitude that’s needed to launch some satellites.

Starship is key to SpaceX’s plans for orbital artificial intelligence (“AI”) data centers, putting these facilities in space. Recently, however, the company has pushed back timelines for upcoming Starship launches.

Despite these concerns, SpaceX CEO Elon Musk recently announced on social-media platform X that “SpaceX, in partnership with Nvidia, has designed a space-optimized Vera Rubin NVL72 system for launch to orbit in Q4 next year, with significant scale in 2028.”

That’s a notable speed-up of the timeline for these satellites, which had been scheduled to launch in 2028.

It’s a ton of moving parts for SpaceX before it can really get orbital AI moving, though investors looking to bet on Musk may want to keep an eye on the “Bank of Elon” hidden inside the firm.

Rockets may get a lot of SpaceX’s headlines, but the company highlights AI as its biggest future value driver.

Musk Touts $1 Trillion in Revenue in 2031 or Sooner

While the rocket business and new Louisiana spaceport may be stepping stones to deploying AI, the AI business is the whole ball game at SpaceX. That’s exactly what SpaceX’s prospectus tells you, with more than 90% of what it considers the “largest actionable total addressable market in human history” of $28.5 trillion expected from its proposed AI businesses.

Management pegs the rocket business at just 1.2% of the company’s total opportunity set. That means SpaceX’s valuation lives and dies on AI, and it is the only way for SpaceX to achieve the $1 trillion in revenue Musk has projected.

He has even sped up the timeline for this milestone in the latest earnings call:

It’s probably also worth mentioning that our internal projections for reaching $1 trillion in revenue, not [annual recurring revenue], but revenue, have moved up from 2031 to 2030. Prior to the [initial public offering (“IPO”)], the financial projections we had were reaching $1 trillion in revenue in 2031. We now expect that to be in 2030. There’s a non-zero chance of that being in 2029.

That’s just not going to happen, and it’s another example of Musk making a promise – here a “projection” rather than a promise – that simply cannot be fulfilled. Such a milestone would see SpaceX vastly surpass the 2026 sales of giants such as Amazon (AMZN) and Walmart (WMT). In effect, that means multiplying SpaceX’s annual sales by more than 20 times in four years.

Don’t count on it, despite Musk’s promises of launching rockets at scale in 2028 (which itself is likely another “overpromise, underdeliver” statement).

SpaceX’s IPO prospectus tells you exactly why it’s not going to happen on this timeline:

[M]any of our initiatives described above under “Our Growth Strategies,” including those to develop orbital AI compute at scale, manufacture AI chips at scale, establish a lunar economy, transport humans and cargo to the Moon and Mars, and develop human augmentation systems, involve significant technical complexity, unproven technologies or technologies that do not exist, and such initiatives may not achieve commercial viability….As a result, the timeline for certain of our initiatives involving unproven or new innovations, including our goal of deploying 100 gigawatts of annual compute power to orbit, the establishment of a lunar economy and interplanetary industrialization, and the launch cadence required to achieve these goals may be difficult or impossible to determine.

The realization of these plans is so far out that Musk and the company couldn’t even give a credible timeline three months ago – when it came time to put things in a written legal document that must offer a much more sober analysis of the issue or suffer potentially serious penalties.

So, Musk says one thing when it comes time to make legal statements but plays a different tune when he’s making projections during conference calls or on X.

SpaceX Stock Still Faces Huge Share ‘Overhang’

SpaceX stock has rebounded well from its recent low, as investors sold the stock mercilessly amid waves of lockup expirations from the company’s IPO. Just weeks after the stock debuted, it declined by more than half from its all-time high of $225 per share.

With worries about this “share overhang” having abated – though they are not eliminated – SpaceX stock has gotten its head above the $135 IPO price.

The lockups prevent SpaceX insiders from selling their shares, but SpaceX has already cleared two of those lockup expirations, and a few more remain in the coming months:

• August 6, 2026: Two days after SpaceX’s second-quarter earnings were reported, about 20% of eligible shares were available to be sold.

• August 21, 2026: About 7% of shares became eligible for trading.

• September 10, 2026: About 7% of shares will become eligible for trading.

• September 25, 2026: About 7% of shares will become eligible for trading.

• October 10, 2026: About 7% of shares will become eligible for trading.

• October 25, 2026: About 7% of shares will become eligible for trading.

With just 5% or so of the total shares outstanding issued in the IPO, that first expiration unlocked nearly four times as many shares for sale. Now, about 28% of shares are due to be unlocked for sale over the next two months.

For now, the stock has found its footing, with a relatively lower percentage of shares newly available for trading through October (28% soon-to-be available shares versus about 32% now). The negative expectations from the share overhang should subside even further.

That’s why I believe the stock has further upside during the next few months and could climb back to $165 again, as I outlined. The stock was trading at that level just weeks ago.

This catalyst looks like the next best opportunity for short-term traders to make money on the stock. Long-term fundamental investors still see a stock that’s massively overvalued, run by a CEO with a history of broken promises. So, the stock will stay up only as long as the market wants to believe Musk’s hype in the face of a different reality.

Regards,

James Royal, PhD

Editor’s Note: Elon Musk revived the electric car, pioneered reusable rockets, and beamed down internet to the four corners of the Earth. But his new initiative – “XPANSE” – could be bolder… more transformative… and far more lucrative than all of these innovations combined. And, like Tesla and SpaceX, it could make early investors incredibly wealthy. Click here for the details… 

The post SpaceX Is Planning a $100 Billion Louisiana Spaceport – But AI Is the Real Must-Win appeared first on MarketWise.

Index funds can be a great way to invest in the stock market, and the best index funds allow you to earn attractive returns. The largest and most popular index funds track major stock indexes such as the Standard & Poor’s 500 (“S&P 500”) and the Nasdaq 100 Index.

The S&P 500, Nasdaq 100, and Dow Jones Industrial Average are stock indexes that include America’s largest publicly traded companies and come with a strong long-term track record. The funds based on these stock indexes deliver nearly the same return as the indexes themselves. So, they’re an attractive way to achieve the index’s return with minimal analysis and work.

Top index funds also typically charge low expense ratios, the annual cost of the fund as a percentage of your investment. Top funds might charge $2 to $3 annually for every $10,000 invested in a fund, so you’re not paying a lot for the fund’s advantages. Index funds can charge low fees because they’re passively managed, meaning that they simply replicate the stocks in the index rather than paying analysts to actively find and invest in stocks.

Here are the best index funds for the major indexes, including both exchange-traded funds (“ETFs”) and mutual funds.

(Looking for a step-by-step plan to help you use AI in your own investing? Here’s how the N.E.W. System works.)

Top S&P 500 Index Funds in 2026

Fund (ticker)	Five-year annual returns	Expense ratio	Minimum investment
Fidelity ZERO Large Cap Index (FNILX)	12.9%	0%	None
Vanguard S&P 500 ETF (VOO)	13.1%	0.03%	None
SPDR S&P 500 ETF Trust (SPY)	13.1%	0.095%	None
iShares Core S&P 500 ETF (IVV)	13.1%	0.03%	None
Schwab S&P 500 Index Fund (SWPPX)	13.1%	0.02%	None
Vanguard 500 Index Fund Admiral Shares (VFIAX)	13.1%	0.04%	$3,000
Fidelity 500 Index Fund (FXAIX)	13.2%	0.015%	None

These funds track the S&P 500, which includes around 500 of the largest stocks traded on American exchanges, such as Nvidia (NVDA) and Apple (AAPL). So, it shouldn’t be surprising that these funds have basically the same average annual return of 13.1% over the previous five years. Therefore, the key differentiator here is the fund’s expense ratio, which ranges from 0% to 0.095%.

While ETFs are usually available to buy and sell wherever you can trade stocks, not all brokerages give you access to the entire mutual fund universe. Mutual funds often charge a minimum initial investment, too.

Three of these S&P 500 funds are ETFs, while four are mutual funds. Just one fund – the Vanguard mutual fund – charges a minimum investment, while its counterpart – the Vanguard S&P 500 ETF – charges a lower expense ratio and does not have a minimum investment. So, it makes sense to buy the Vanguard ETF or another lower-cost ETF unless you can’t access them.

Some of these funds also appear on this year’s list of the best ETFs.

Top Nasdaq Funds in 2026

Fund (ticker)	5-year annual returns	Expense ratio	Minimum investment
Invesco QQQ Trust (QQQ)	14.9%	0.18%	None
Invesco Nasdaq 100 ETF (QQQM)	14.9%	0.15%	None
Victory Nasdaq-100 Index Fund   (USNQX)	14.6%	0.42%	$3,000
Fidelity Nasdaq Composite Index ETF (ONEQ)	13.3%	0.21%	None

These funds track the Nasdaq, with three following the Nasdaq 100, which includes the 100 largest non-financial firms on that exchange. The other fund tracks the Nasdaq Composite, which measures the performance of all stocks trading on that Nasdaq exchange, including Microsoft (MSFT) and Amazon (AMZN).

While the Invesco QQQ Trust has long been the most popular ETF, the recently introduced Invesco Nasdaq 100 ETF has delivered the same performance at a lower expense ratio.

Meanwhile, the Victory Nasdaq-100 index fund is a mutual fund tracking the Nasdaq 100 and requires a $3,000 minimum investment. The Fidelity fund here tracks the Nasdaq Composite rather than the Nasdaq 100, giving it broader exposure and less concentration in the largest tech stocks.

The Nasdaq Exchange includes the world’s largest tech companies, such as the recently debuted SpaceX (SPCX). (CEO Elon Musk is working on a way to disrupt the global financial system, here’s how.)

Top Dow Jones Industrials Funds in 2026

Fund (ticker)	Five-year annual returns	Expense ratio	Minimum investment
SPDR Dow Jones Industrial Average ETF Trust (DIA)	10.6%	0.16%	None
Invesco Dow Jones Industrial Average Dividend ETF (DJD)	11.9%	0.07%	None

Fewer funds track the Dow Jones Industrial Average and its 30 stocks, including Goldman Sachs (GS) and Caterpillar (CAT). The SPDR Dow Jones Industrial Average ETF is the key name to watch here. Meanwhile, the Invesco ETF here focuses only on dividend-paying stocks in the Dow Jones Industrials.

Investors looking for more dividend-paying investments should review the best dividend ETFs.

Which Index Fund Is Best for You?

In general, if you’re searching for an index fund for a major index such as the S&P 500 or Nasdaq 100, you have several fund choices. Since funds tracking the same index deliver very similar performance, the best pick is often the one with the lowest expense ratio.

Any of the funds on these lists are cheap or reasonably priced. The S&P 500 funds here offer the lowest expense ratios overall, but even funds in other categories are still fairly priced.

If you go with the Fidelity ZERO fund above, which charges no expense ratio, it’s important to know it does so by avoiding the S&P branding and the related fees. The fund still tracks the index, but Fidelity can’t market it as an S&P 500 fund.

In general, unless you need to buy a mutual fund – say, you’re investing in a retirement account – an ETF is often the better choice. They tend to have a lower expense ratio and don’t require a minimum initial investment, unlike mutual funds. Plus, they trade throughout the market day, unlike mutual funds, which are priced and traded only at the end of the day.

As mentioned above, not all mutual funds are available at all brokers. For example, the Fidelity fund with no expense ratio is not available at many brokers. However, any of the ETFs above should be available at any broker that offers stock trading.

Reasons to Invest in an Index Fund

The key advantage of an index fund is that investors can earn strong returns without spending much time or energy analyzing stocks.

Investors in an S&P 500 Index fund get exposure to the 500 or so stocks in that index, and the funds closely track the index’s performance, which has been about 10% over long periods and even more recently. Each share of the index fund lets investors own a small slice of all the companies in the index, and it charges one low management fee that is deducted seamlessly from the account.

Index funds offer immediate diversification, lowering the risk of investing in an individual stock. Diversification is the idea that your money is spread out across many different investments, so that the performance of one stock doesn’t hurt your overall performance too much.

Of course, there’s a flip side to diversification, too. An index fund’s performance is just the weighted average performance of all the stocks in it. So, it’s necessarily an average. In other words, some stocks will perform much better than the average.

If you can find outperforming stocks, your overall return can do drastically better than an index fund’s average. So, it can be worth your time and money to search out high-flying stocks.

Regards,

James Royal, PhD

Editor’s Note: Whitney Tilson called the rise of Apple, Amazon, and Netflix… as well as the collapse of dozens of companies that went bankrupt. Now the former $200M hedge fund firm manager is stepping forward with what he calls the most important financial warning of his 30-year career. He’s sharing two free stock recommendations  (one to buy, one to sell immediately) along with details of a new proprietary system fueling his predictions. See it all in his free presentation.

The post Best Index Funds: Top Funds For the S&P 500, Nasdaq, and Dow Jones appeared first on MarketWise.

Anthropic is on the verge of acquiring Decart AI, a company that increases the efficiency of artificial intelligence (“AI”) chips, as Anthropic seeks to lower operating costs in a brutally competitive market. Anthropic is racing toward what its investors insist will be a $2 trillion initial public offering (“IPO”), and the company needs a compelling story showing how it can get on a path to profitability.

The deal would be worth about $6 billion, according to Bloomberg, making it the AI lab’s largest acquisition ever. Decart provides optimization software that helps squeeze more performance from chips, lowering the cost of training AI models. The startup also builds “world models,” a kind of simulation that mimics the real world and helps AI models navigate it effectively.

Besides the potential to lower costs, the acquisition of Decart could allow Anthropic to serve more users on existing hardware. That may prove useful as AI labs say they’re capacity-constrained by the fast-rising demand for services and the relatively slow build-out of AI data centers.

Anthropic and its key American rival, OpenAI, face rapidly increasing cost pressures, as low-cost Chinese AI model companies such as DeepSeek and Moonshot churn out high-quality models that may cost just 5% to 10% of what their American rivals charge. As a result, Chinese models have been rapidly growing their market share.

OpenAI has responded by slashing prices on one of its newest AI models by up to 80%, a move that threatens to heat up a simmering price war in the industry. OpenAI’s move shows AI labs lack pricing power and threatens the company’s ability to ever go public.

Anthropic feels the price pressure as well. As the AI industry’s dynamics rapidly shift and costs become ever more important, Anthropic is rushing its IPO across the finish line. Reports suggest that the company is still targeting an October debut, and it has been touting its sales growth in recent months to keep investors warm for the offering.

Media Pumping Anthropic $2 Trillion IPO Rumor

Have you ever gone to a restaurant and paid for a meal before you even knew what you’re being served? Of course not, but that’s what a few investors are doing for the Anthropic IPO, trying to attach a $2 trillion price tag to the offering before the public knows what they’re getting.

Unfortunately, the Financial Times is helping them do it. The pink paper quoted a half-dozen Anthropic investors who said the AI lab could hit a $2 trillion valuation for an IPO. That would make Anthropic the largest IPO of all time, surpassing SpaceX’s (SPCX) recent public offering.

Sharp investors should be highly dubious and consider this primarily as an approach to pump the stock of an upcoming IPO:

Anthropic’s backers say booming demand for the lab’s advanced AI models and tools justifies their lofty expectations. Investors expect the Claude maker’s annualised revenue to be between US$100 billion and US$120 billion by the end of 2026 — using the start-up’s preferred measure, which infers full-year sales from recent performance — up by more than 10 times over the course of 2026.

Incredibly, the Financial Times goes on to quote one investor who posits a $3 trillion valuation on Anthropic.

Let’s unpack some of what’s happening here.

• The $2 trillion IPO valuation is insiders’ price target. Insiders are trying to normalize a $2 trillion figure for the investing public, so it looks reasonable. A $3 trillion valuation, also tossed around, helps make $2 trillion look less extreme.

• The public is being fed this info by highly interested insiders.This valuation data is being fed to the media by insiders who have an investment interest of millions to billions in an Anthropic IPO, so they want to use the largest number possible.

• No detailed financial information is public yet. Anthropic’s financial info remains high-level in this FT article, so the public can’t clearly know what it would get for $2 trillion. How much money is Anthropic losing, and what is its trend toward profitability?

• Annualized revenue is not defined. While insiders provide a target for annualized revenue, the formula isn’t clearly defined. A guess might be that it’s 12 times the last month’s revenue, but that is not really clear. This lack of clarity works well for those who want the investing public to be ready to buy the IPO stock at any price.

• Annualized revenue is not revenue.Annualizing revenue is a forecast of what revenue would look like over the course of a year. So it’s not a statement of realized revenue, as the figures on financial statements filed by publicly traded companies are. For a high-growth company, annualized revenue will generally look much higher than actual revenue.

In short, the Financial Times is offering readers the best possible image of Anthropic – full of make-up, doused with perfume, and with a photo filter on. The point of this kind of “media plant” is to get the investing public excited about AI as “the wave of the future,” so that it uncritically buys whatever the big investors are selling BEFORE they see the financials.

And if OpenAI’s financials are anything to go by, Anthropic could be burning cash at a fearsome rate.

Billionaire Elon Musk used a similar strategy with the SpaceX IPO. Musk publicized a $1.5 trillion IPO figure in late 2025 and early 2026 before detailed financials were made public – indeed before even filing the IPO paperwork. The pre-IPO figure rose to $1.75 trillion, and company officials even briefly worked $2 trillion into the conversation before the IPO.

While SpaceX went up briefly in the wake of the IPO, it has been mostly down since then. The pre-IPO jawboning helped the company price the stock at a massive overvaluation.

So, investors should carefully review Anthropic’s financials when they’re released to see what they’d be getting for their money. Insiders are using the media to normalize a huge valuation before the public is forced to see Anthropic’s actual – quite likely terrible – financial figures.

Regards,

James Royal, PhD

Editor’s Note: Could a major AI company’s “silent IPO” end up overshadowing both SpaceX and Anthropic?  

According to Marc Chaikin — the 60-year Wall street legend who called Nvidia before it shot up 45,000% — a tech firm that been called “the unseen winner of the AI race” could soon break itself up into three separate companies.  

If it does, the next Netflix… the next Tesla… and the next Amazon are all poised to spin off just from this one stock. That means, if you simply make one trade today, you could could automatically get your hands on shares of all these potential spinoff IPOs with no more work on your part. 

Click to see how you can get in on this “silent IPO” before it’s too late… 

The post Anthropic in Talks to Buy Decart AI in Push to $2 Trillion IPO While Financials Remain Murky appeared first on MarketWise.

Shares of Circle Internet Group (CRCL) and Coinbase (COIN) both moved higher on Monday after the Treasury Department proposed new rules for the GENIUS Act, the stablecoin law enacted in July 2025. The proposal will be open to public comment for 60 days, with a final rule expected before January 2027.

Although stablecoins won’t suddenly become mainstream simply because the Treasury Department published its proposal, something more important has already happened.

Stablecoins are getting boring.

For most of crypto’s history, that would have sounded like an insult. Now, it may be stablecoins’ investment case.

The original crypto pitch involved overthrowing banks, replacing currencies, and cutting out financial intermediaries. Stablecoins are turning into something much less dramatic – a form of payments infrastructure. Mastercard (MA) is building around them. Payments app MoneyGram uses them. Businesses can settle payments at midnight on a Sunday instead of waiting for the banking system to open.

In short, stablecoins are becoming part of our ordinary world.

The GENIUS Act’s Impact on Stablecoins and Crypto Stocks

That leaves investors with a more useful question than whether crypto is “the future.” The question now is: If stablecoins become ordinary financial plumbing, who gets paid for it?

Circle and Coinbase have two very different answers…

A stablecoin such as Circle’s USDC is a digital dollar that can move across blockchain networks at any hour. The useful part of that idea has little to do with a person buying a cup of coffee… Electronic payment networks such as Mastercard and Visa (V) already handle those types of transactions just fine.

The bigger opportunity is where the existing system gets expensive or slow. Again, with traditional finance, those drawbacks happen with things like cross-border payments, remittances, corporate treasury, and settlement outside banking hours.

On stablecoin rails, the payment and settlement can become the same event rather than a message followed by money moving a day or two later. The hope is to fix these inefficiencies.

That gets us to Circle.

Circle issues USDC and holds reserves against those coins. It earns interest on the reserves it holds. In the second quarter, reserve income was $667.7 million, or 95.2% of Circle’s total revenue. Average USDC in circulation reached about $76.5 billion, up from $61 billion a year earlier.

The attraction is obvious. If more businesses and consumers use USDC, Circle gets more dollars to hold in reserve and potentially more reserve income.

But there’s a catch.

Circle’s economics depend heavily on interest rates. Lower rates mean lower returns on those reserves. Circle’s own filing estimates that a one-percentage-point decline in rates, if USDC circulation remains constant, would cut annual reserve income by roughly $737 million.

Then there’s Coinbase.

Coinbase is fascinating because it sits on the other side of Circle’s economics. It distributes USDC and gets paid handsomely for doing it.

Circle paid Coinbase $324.6 million in distribution costs in the second quarter alone. Coinbase says its customers held a total of $20 billion in USDC in Coinbase Products, on average, during the quarter – more than 30% of the total supply of USDC.

Coinbase captured roughly half of all USDC economics (meaning the yield earned by the investments USDC holds and who benefits from it) over the past year.

That gives investors two very different stocks tied to the same trend.

Circle makes more money when USDC circulation and reserve income grow… while Coinbase gets a large piece of the economics for putting USDC in front of customers.

Stablecoins also give Coinbase something it badly wants: revenue that doesn’t require customers to trade bitcoin constantly. Subscription and services revenue reached $555 million last quarter, or 48% of net revenue.

The Next Test for Stablecoins Is Real-World Use

There’s plenty of room for stablecoins to disappoint, however.

Most stablecoin activity still has little to do with buying goods or paying workers. The figures I have been watching put actual payments at roughly $390 billion out of about $35 trillion in stablecoin-network activity over a recent 12-month period (annualized for 2025 in December). Call it 1%. Trading, arbitrage, and machines moving liquidity still dominate the industry.

I keep coming back to the other side of that number, though.

The payments portion is tiny, but it more than doubled from 2024 to 2025.

The GENIUS Act will not make stablecoins mainstream by itself. The Treasury cannot force people to use USDC. What stablecoin regulation can do is remove one of the primary reasons banks, payment companies, and large businesses have hesitated to build around it.

That creates competition, too. Circle and Coinbase will not have this market to themselves.

But the investment question has become much more concrete. We no longer have to ask whether blockchain will someday find a use beyond crypto trading.

Money is already moving over these networks, and we’ll soon get to find out who exactly stands to gain when stablecoins become part of the main financial plumbing.

Good investing,

Eric Wade

Editor’s Note: Whitney Tilson … hedge fund legend, Harvard grad, and the investor CNBC called “The Prophet” … just stepped forward not just as a financial expert, but as a father and an American. He says we’re entering the most chaotic six months this country has seen in a generation. The signs are already everywhere. And he believes most people have no idea what’s coming for their money. His warning (and the one move he says can make all the difference) is in his free presentation. 

The post Circle and Coinbase Are Turning ‘Boring’ Stablecoins Into Great News for Investors appeared first on MarketWise.

Neocloud stock CoreWeave (CRWV) has sharply risen by about 54% from its one-month low of $60.55, making now a good time for investors to consider heading to the sidelines after the latest rally.

The artificial-intelligence (“AI”) data-center company is both debt-heavy and relies on continual financing to stay afloat. Worse, the people who know the company best – executives and other insiders – have been unloading the stock as fast as they can.

In short, if the AI bubble bursts, there are reasons to think that CoreWeave could go to zero.

If you look at the surface of CoreWeave’s results, you might think that nothing’s the matter. In fact, the company’s rapidly increasing sales in recent quarters would seem to nullify any bearish fears. But a closer look shows why I said CoreWeave is one of five AI stocks that could go bust.

Period	Sales growth	Operating margin	Interest expense	Net margin
Q1 2025/2026	112%	-2.7%/-6.9%	$264/$536*	-32.1%/-35.5%
Q2 2025/2026	112%	1.6% / -1.9%	$267/$640*	-23.9%/-24.3%

On the top line, CoreWeave grew sales 112% year over year in both the first and second quarters of 2026. For any normal company, such rapid sales growth would quickly translate into much higher operating margins, but the improvement has been less obvious here.

We didn’t see higher operating margins in the first quarter, as they fell from negative 2.7% last year to negative 6.9% in the first quarter of this year. Second-quarter operating margin also fell, from 1.6% last year to negative 1.9% this year. However, operating margin improved sequentially from the first quarter to the second quarter, so it’s important to see whether this trend continues.

While operating margin is a key financial factor, CoreWeave’s heavy use of leverage makes the net margin especially important. Despite CoreWeave’s stellar sales growth, it’s taking on significant interest expense to finance that growth. As a result, net margin is not improving much, especially in the second quarter.

For example, CoreWeave paid $267 million in interest expense during the second quarter a year ago. In the second quarter this year, interest costs ballooned to $640 million. So, despite CoreWeave’s 112% sales growth, net margin actually worsened year over year, from negative 23.9% to negative 24.3%.

While CoreWeave reports those fast-growing sales figures in its press releases, it leaves out the fast-growing liabilities that finance them. To see that, you need to turn to Form 10-Q, where you can see just how much the company relies on liabilities.

Of the company’s $77.1 billion in assets as of June 30, 2026, about $72 billion are financed by liabilities. That leaves a meager $5 billion in shareholders’ equity supporting that pile of assets. In other words, CoreWeave is massively leveraged. If the value of those assets is written down by 6.5% or more, the company will have zero or negative shareholders’ equity.

Zero or negative shareholders’ equity might be fine for a company that has a strong history of turning a profit, but that’s not CoreWeave. It must keep financing its assets and paying interest.

With the company’s market capitalization of about $51.4 billion, investors are willing to pay about 9.1 times the book value of the company. Put another way, the market is saying the company’s assets are 9.1 times as valuable as their book value.

Despite huge growth, this valuation doesn’t make sense given the lack of profitability.

CoreWeave Insiders Are Selling Aggressively

But this analysis is not the only thing telling you CoreWeave’s valuation doesn’t add up. Key company insiders have been telling you all year that CoreWeave is overvalued, selling hundreds of millions of dollars in stock as fast as they can. The company’s recent addition to the Nasdaq 100 Index gave the stock another boost that insiders could sell into.

Before we run through some numbers, CoreWeave bulls will point out (rightly) that insiders may sell for many reasons. They sell to diversify their own portfolios, because they need cash elsewhere, and because they’re raising money to pay taxes on stock grants. At the same time, they typically buy only when they believe the stock is poised to go up.

Or as the old investment saying goes, “insiders sell for many reasons, but they buy for only one.”

But you rarely see the depth and breadth of selling that’s evident at CoreWeave. It’s the kind of thing that you should be terrified of if you own the stock. Insiders have been selling at a furious pace since August 2025 – when the company’s lockup on insider selling expired as part of its initial public offering. They’ve been selling every month since then, too. Plus, there has not been a single insider buy over that time, according to MarketBeat.

Let’s look at selling from the company’s CEO, Michael Intrator, over the past few months.

Insider	July (Sales/$)	June (Sales/$)	May (Sales/$)
CEO Michael Intrator	7 sales / $75.4 million	10 sales / $174.5 million	9 sales / $136.2 million

Intrator sold shares in seven separate batches in July for a total of $75.4 million, according to MarketBeat. That followed 10 sales for $174.5 million in June and nine sales for $136.2 million in May.

The CEO’s sales pattern goes back to August 2025.

But it’s not just the CEO selling shares. Rather, it’s a “Who’s Who” of CoreWeave that has been getting rid of shares at a brisk pace:

• Brannin McBee, chief development officer

• Nitin Agrawal, CFO

• Kristen McVeety, general counsel

• Brian Venturo, co-founder, director

• Jack Cogen, director

• Chen Goldberg, senior vice president of engineering

• Jeff Baker, chief accounting officer

• Sachin Jain, COO

• Karen Boone, director

The amounts of stock they’ve sold differ, of course, from relatively small sales of a few shares to gargantuan amounts, such as the $49.4 million from Brannin McBee in September 2025 or the $105.3 million from Jack Cogen that same month.

All those sales and literally zero insider buys. Nothing is illegal, but as an investor (or potential investor) in CoreWeave, it’s an unambiguous signal that’s quite valuable.

Insiders selling CoreWeave stock this aggressively tells you almost everything you need to know about it as an investment. Tack on a business that can’t even budge its net profit margin despite 112% sales growth, and you have a stock that could plummet if the AI bubble bursts.

Regards,

James Royal, PhD

Editor’s Note: Whitney Tilson — the hedge fund manager CNBC called “The Prophet” — says America has reached its “Ripping Point.” The old financial order is being torn apart, and he believes most investors have no idea what’s coming in the next six months. He’s named the stocks he thinks will be destroyed in the chaos — and the ones he believes will soar. Watch his free presentation while it’s still available.

The post Despite CoreWeave’s Recent Rebound, Here’s How the Neocloud Stock Could Go to Zero if the AI Bubble Pops appeared first on MarketWise.

The best dividend stocks offer solid payouts that investors can rely on for years. They help investors build wealth over time, offer a growing dividend, and are backed by the safety of a company with a solid track record. With a sizable portion of their returns paid out in cash, these top stocks are a great choice for retirees or anyone seeking income.

Top dividend stocks can offer investors the following benefits:

• Strong, long-term returns, with attractive capital gains and not just a high payout

• A high dividend, with yields higher than 2%, much greater than the S&P 500’s yield

• A growing dividend, with a payout that grows at least at the target rate of inflation

• A solid business, a larger company that has survived and thrived, often for decades

• Lower volatility, meaning the stock fluctuates less than the average stock

The best dividend stocks give you the chance to build wealth while offering a growing stream of dividends for decades, giving you the cash to spend however you like. Certain investors prefer dividend stocks with high yields and some growth over time, while others prefer a lower yield today but higher growth potential over time.

This list of 11 dividend stocks offers options for both approaches.

Best Dividend Stocks: How They Compare

Stock (Ticker)	Dividend yield	Five-year annualized dividend growth	Five-year annualized return
Verizon Communications (VZ)	5.9%	2.0%	2.0%
Edison International (EIX)	5.0%	5.7%	8.6%
Kimco Realty (KIM)	4.4%	8.6%	5.7%
Fidelity National Financial (FNF)	4.3%	8.4%	4.5%
Lamar Advertising (LAMR)	4.1%	19.9%	10.9%
Paychex (PAYX)	3.8%	11.9%	4.2%
CareTrust REIT (CTRE)	3.7%	6.0%	15.0%
Public Storage (PSA)	3.7%	8.5%	4.7%
NextEra Energy (NEE)	2.9%	10.7%	2.8%
McDonald’s (MCD)	2.7%	7.3%	5.0%
Coca-Cola (KO)	2.4%	4.5%	11.1%

Some of these stocks are also included in MarketWise’s list of best retirement stocks for 2026.

1. Verizon Communications (VZ)

Everyone knows Verizon as one of the major U.S. telecom players, but maybe less well known is that it pays a massive dividend of nearly 6%. The trade-off for that high current yield is slow dividend growth, but investors can decide for themselves how they want to play the dividend game. The stock has not had a great five-year run, but its higher current yield gives investors plenty of reason to stick around and wait for the stock to catch a bid.

Market cap: $201 billion

Dividend yield: 5.9%

2. Edison International (EIX)

Electrical utilities can be great dividend picks because they enjoy stable, growing demand for their key product. They also tend to increase prices year after year, and their stock tends to be relatively stable because of these advantages. For California-based Edison, that has resulted in an attractive dividend and a history of raising it quickly.

Market cap: $27 billion

Dividend yield: 5.0%

3. Kimco Realty (KIM)

Kimco Realty is a real estate investment trust (“REIT”) that owns grocery-anchored shopping centers and mixed-use properties in major metro areas. Kimco’s “must-visit” properties generate reliable rents that grow over time, and as a REIT, it must pay out most of its income as dividends.

The company’s meaty current yield doesn’t come at the expense of strong dividend growth over the past five years either.

Market cap: $16 billion

Dividend yield: 4.4%

4. Fidelity National Financial (FNF)

Fidelity National Financial may not be a familiar name for many investors, but if you’ve ever bought or sold a house, there’s a good chance you’ve used the company for its title insurance. It’s the leading player in this oligopolistic industry, and it has delivered excellent returns over time, even if the past five years have been ho-hum.

It offers a high current yield, and management hasn’t been stingy about increasing the payout over the past half-decade, giving investors reason to hold the stock for a while.

Market cap: $13 billion

Dividend yield: 4.3%

5. Lamar Advertising (LAMR)

It seems like artificial intelligence (“AI”) or tech is disrupting nearly every business these days, so it’s refreshing when you find a business such as Lamar’s – billboard advertising – that seems largely unaffected. The company has some 360,000 billboards in the U.S. and Canada that advertise local and national brands to travelers.

The company has grown its meaty dividend nearly 20% annually over the past five years, on the way to putting up nearly 11% annualized returns.

Market cap: $16 billion

Dividend yield: 4.1%

6. Paychex (PAYX)

Paychex offers human resources management and payroll processing for small- and medium-sized businesses, among other related services. It has delivered tremendous returns for investors over time, even if the past five years have been a bit mediocre.

But that may be just a brief pause before another period of exceptional performance. Today’s investors can lock in a strong dividend and look for signs that management will raise that payout aggressively moving forward.

Market cap: $43 billion

Dividend yield: 3.8%

7. CareTrust REIT (CTRE)

CareTrust may not be a household name like others on this list, but it has delivered solid returns to investors, with a high yield and a growing payout over time.

As a REIT, it must make sizable dividend payments, and its business of leasing out healthcare facilities on long-term leases with annual rent escalators ensures the stable revenue that’s a prerequisite for a top dividend stock.

Market cap: $9 billion

Dividend yield: 3.7%

8. Public Storage (PSA)

Public Storage is the behemoth in the self-storage space. As a real estate investment trust, it has to pay out most of its income as dividends. It offers a high current yield that has grown at a brisk 8.5% annually over the past five years, meaning you’re getting the best of both worlds.

The REIT has redesigned its executive incentive system to help drive stock outperformance, a move that should help keep the dividend AND share price growing.

Market cap: $61 billion

Dividend yield: 3.7%

9. NextEra Energy (NEE)

NextEra is one of the largest North American electric utilities, with about 80 gigawatts of generation and storage capacity. NextEra is almost the definition of a “must-have” asset, with a product that will always be in demand, even while offering strong upside as AI-driven power demand pushes prices higher.

The stock’s above-average dividend has been growing quickly over the past half-decade, and recent price weakness may be a long-term opportunity.

Market cap: $179 billion

Dividend yield: 2.9%

10. McDonald’s (MCD)

McDonald’s may be famous for its Big Mac, but the firm makes much of its money from franchise fees and leasing out the real estate that sits beneath its thousands of restaurants.

As one of the most popular consumer businesses, the Golden Arches has shown notable durability, helping drive a solid dividend that’s grown at a fast 7.3% over the past five years.

Market cap: $193 billion

Dividend yield: 2.7%

11. Coca-Cola (KO)

While everyone knows the company’s namesake drink, Coca-Cola is a drink distributor that can deliver whatever consumers want – sodas, water, sports drinks, and juices. Coca-Cola has proven that it can thrive over decades, while increasing its dividend for 64 straight years.

Although its yield is on the lower side of our list, it has grown at a reasonable 4.5% annualized clip, while the stock has returned about 11% annually on average over the past five years.

Market cap: $377 billion

Dividend yield: 2.4%

Another Top Dividend Stock to Consider for Retirement Investors

The best dividend stocks offer strong returns over time that investors can rely on year after year, and Wall Street veteran Whitney Tilson has spent decades hunting for a business so durable that an investor could build an entire retirement portfolio around it.

Now, he says he has found a business that could deliver even bigger, more durable returns than the legendary business of Berkshire Hathaway and keep delivering for decades.

Whitney calls it “America’s Greatest Retirement Stock.”

Most investors have never heard of this company, and it only has around 100 employees. The company doesn’t exactly sit in a major metropolitan area either, doing business on a stretch of West Texas and New Mexico bigger than the size of Rhode Island.

That might make you think of oil, but this company doesn’t drill wells and doesn’t own a data center or generate any power. Instead, it owns the ground beneath one of the largest industrial build-outs in modern American history.

And every single time someone wants to use that ground, this company collects a payment.

Whitney recently took a helicopter out to see it for himself… Click here to watch his exclusive investigation.

Regards,

James Royal, PhD

The post 11 Best Dividend Stocks for 2026: High Yields and High Growth appeared first on MarketWise.

It is looking increasingly likely that SpaceX (SPCX) will acquire carmaker Tesla (TSLA), with even SpaceX and Tesla CEO Elon Musk not denying the possibility. While the move could make some sense for operational and business reasons, Musk’s insatiable ego is the driving factor.

But as pessimistic as I was about the rationale for the merger, I still failed to account for a key reason for Musk to merge the two firms: He stands to earn megabucks in performance bonuses from Tesla as part of the $1 trillion compensation package he signed with the carmaker last year.

For investors in Tesla and SpaceX, the ultimate question is really: Will SpaceX vastly overpay for Tesla, hurting SpaceX investors for the benefit of Tesla investors? Or will SpaceX offer a fair price for Tesla, thereby reducing the stock price of the massively inflated Tesla and benefiting SpaceX investors?

The best prediction is probably to look at Musk’s compensation package and determine the circumstances in which he benefits most.

Here’s How Elon Musk Benefits in a SpaceX-Tesla Merger

Musk benefits in the event of a takeover of Tesla, due to his 2025 CEO Performance Award Agreement with Tesla. The agreement laid out 12 different awards that Musk could earn for achieving various operational goals, such as delivering 20 million Tesla vehicles, or by market capitalization goals such as Tesla being valued at $2 trillion or $2.5 trillion, for example.

For each of the 12 levels, Musk could earn more than 35.3 million shares of Tesla, which are currently worth about $11.5 billion at a stock price of $326.

But achieving the operational goals is not necessary in the event of a change of control, such as if SpaceX acquires Tesla. In this case, all that needs to happen is for Tesla’s market cap to exceed one or more of the thresholds for Musk to achieve at least some of the shares.

At a share price of $500 – where Tesla’s market cap is about $2 trillion – each 35.3-million-share tranche is worth nearly $18 billion. So, the bigger the size of the deal, the larger his payday could become, as he earns more shares at a higher valuation.

Because Musk owns supervoting stock in SpaceX that gives him effective control of the company, he has a strong ability to decide whether SpaceX acquires Tesla and at what price. So, he has both the incentive to overpay for Tesla and the means to make a merger occur.

Given his strong voting position in SpaceX, he can even overpay for Tesla using SpaceX stock while still maintaining effective control of the newly merged firm, as a Wall Street Journal article shows. Even if SpaceX pays at the top of the compensation range for Tesla – a whopping $8.5 trillion – Musk would still hold control of the merged company, with about 56% of the vote.

At that level, he’d add more than $800 billion to his personal wealth, soaking SpaceX investors by massively overpaying for Tesla stock. Again, there’s that conflict of interest.

This level of self-dealing – It may all be too much for even the most ardent Musk fans, despite all the opportunity inside the company: rocket ships, Starlink satellites, Optimus robots, Terafab, as well as the hidden financial business inside SpaceX.

Tesla Explores Divesting Its China Business

A SpaceX-Tesla merger might require Tesla to separate its Chinese business in some form, and Tesla has been exploring that possibility, according to the Wall Street Journal. While Musk called the article “fake news,” a divestiture would prove problematic, given Tesla’s enormous valuation.

The concern here is that SpaceX is a significant defense contractor, with the federal government making up 20.9% of its 2025 sales. With SpaceX units such as top-secret satellites, the U.S. would want some separation between the company and China. At the same time, China would take a close look at a SpaceX merger if it led to a U.S. contractor controlling Tesla factories.

Tesla executives have been told to prepare for the Chinese unit to be hived off, whether that’s a spinoff, a sale, or even a closure. A sale or spinoff could prove particularly problematic because it would cause a standalone Chinese unit to be valued separately from Tesla as a whole.

A standalone valuation of the China unit will prove problematic because Tesla is so overvalued to start with. Tesla might actually be worth 80%, 90%, or maybe 95% less than today’s value, about $1.3 trillion, but Tesla enjoys a Musk premium, a price boost because Musk is at the helm.

But a strategic acquirer such as another car company will not purchase Tesla China at a price that’s 5 to 20 times higher than its actually worth, as the stock market is doing. So, a sale of the China unit at something near fair value necessarily means that the remaining non-China units are being priced by the market at an even more massive overvaluation, given their assets.

In this case, a SpaceX acquisition of Tesla near its current price implies a huge overpayment for the remaining Tesla assets. In other words, SpaceX investors get hosed, while Tesla investors enjoy a huge windfall by selling their overinflated assets – one reason Tesla is too risky to short.

A spinoff of Tesla China into a separate publicly traded company isn’t likely to solve the problem either. If the market is not willing to award the spinoff a Musk premium, it’s not going to fetch the kind of price that it’s currently doing inside Tesla today, where it’s likely 5 to 20 times overvalued. In this case, SpaceX would again vastly overpay for the remaining Tesla assets, soaking investors.

Given the huge difficulties created by Tesla’s overvaluation, the company has been exploring other ways of “firewalling” the Tesla China unit from the rest of the business.

Why SpaceX Is Likely to Acquire Tesla

Besides the personal payday for Musk, Musk may seek to combine SpaceX and Tesla for a variety of other reasons:

• Musk has a history of merging companies to paper over bad news: Tesla bought the debt-laden SolarCity in 2016, of which Musk owned 20%. The move bailed out Musk’s personal investment in the renewable energy company, which his cousins managed.

• A merger puts Musk in total control:Musk runs the whole ship at SpaceX through his supervoting stock, and a purchase of Tesla may mean that he can simply ignore investors who want to challenge his decisions or compensation, for instance.

• A SpaceX-Tesla merger stokes Musk’s ego:A merger would put the Musk-helmed company among the world’s largest companies by market capitalization.

Investors should expect to see more of Musk and his team laying the groundwork for a potential merger in the coming quarters, much as he did in the last earnings call:

Well, as you can tell from all the many collaborations on so many fronts with SpaceX, there’s more and more overlap, especially with Terafab, that’s really going to be a gigantic project. Obviously, we can’t talk about combining companies and that kind of thing on an earnings call. It’s got to be done with the appropriate process.

This kind of non-denial shows that Musk is likely to be actively pursuing a merger, and it’s only a matter of time before we get the details. As we’ve seen in Musk’s recent deals, he’s going to come out of it just fine. The real question is which investor group – SpaceX’s or Tesla’s – gets worked over harder. It looks like Musk’s money is riding on SpaceX to pay up for Tesla.

Regards,

James Royal, PhD

Editor’s Note: Elon Musk revived the electric car, pioneered reusable rockets, and beamed down internet to the four corners of the Earth. But his new initiative – “XPANSE” – could be bolder… more transformative… and far more lucrative than all of these innovations combined. And, like Tesla and SpaceX, it could make early investors incredibly wealthy. Click here for the details… 

The post Tesla-SpaceX Merger: Here’s How Elon Musk’s Big Potential Payday Could Lead to an Overpay for Investors appeared first on MarketWise.

In a somewhat hush-hush deal reported by Bloomberg on August 4, Anthropic signed a six-year, $10 billion agreement with Volta Infrastructure, a Singapore-based AI neocloud startup that, according to its website, “is the upstream enabling layer for AI infrastructure in high-growth markets,” which “control[s] the physical prerequisites that make hyperscale deployment possible.”

With this deal, Anthropic receives exclusive access to 133 gross megawatts (“MW”) of power capacity from Volta at Bitdeer Technologies’ (BTDR) Tydal data-center campus in Norway. This ensures Anthropic has plenty of Norwegian hydropower and wind energy to train its next-generation Claude AI models.

But the Anthropic news is only one layer of a much bigger deal between Bitdeer and Volta.

On August 4, Bitdeer signed a 16-year lease and services agreement with Volta worth roughly $4.7 billion in scheduled payments through its initial term. The deal could potentially reach a total of $8 billion through a one-time eight-year lease extension, per Bitdeer.

Bitdeer expects to deliver the 133 gross MW capacity through four data halls (which are secure areas housed inside data centers) split into two phases… The first is slated for completion by December 31 and the second by March 31, 2027.

The data halls – which are being converted from a bitcoin-mining campus – will be outfitted with PowerRack infrastructure from Dell Technologies (DELL) using Nvidia (NVDA) advanced Vera Rubin AI-accelerator chips.

It’s all a bit complicated, so here’s a quick recap of each deal:

The Volta-Bitdeer Deal: $4.7 Billion Over 16 Years, With a Path to $8 Billion

• Volta and Bitdeer agreed to a 16-year physical infrastructure colocation lease and services agreement in which Volta pays $4.7 billion in total contracted revenue to Bitdeer, with options that could push the total value to $8 billion.
• Bitdeer supplies 133 gross MW of renewable power to Volta through four data halls in Tydal, Norway.
• Bitdeer takes on the physical conversion of the bitcoin-mining facilities into data-center structures and runs the operations (such as power management and cooling) for each facility.
• Volta deploys Dell and Nvidia AI hardware within the facilities.

The Anthropic-Volta Deal: $10 Billion for Six Years of Tydal Capacity

• Anthropic and Volta agreed to a six-year contract that involves Anthropic paying Volta $10 billion to secure direct cloud access to the computing capacity from the Tydal data centers, which it will use to train its Claude models.
• Volta is effectively the neocloud layer between the physical site (owned by Bitdeer) and the data-center equipment from Dell and Nvidia, which Volta will install and manage.

These are important deals for a couple of reasons. For one, they demonstrate how AI companies are increasingly looking outside the box (in this case, to Norway) to secure the power they need to train and run their constantly evolving models.

The deals also illustrate the push that AI companies are making to use the newest generations of chips… in this case, Nvidia’s much-anticipated Vera Rubin architecture. They also show how traditional bitcoin-mining companies like Bitdeer are finding other revenue streams – in this case, converting its crypto sites into AI cloud data halls – during a time when the crypto market as a whole is trying to find its footing after losing roughly 50% of its value since its peak last October.

Investors noticed Bitdeer’s strategy shift after its deal with Volta (and Volta’s deal with Anthropic) and pushed the stock up as high as roughly 14%, to $12.93, during August 4 trading.

The JPMorgan $1.3 Billion Credit Backstop That Was Key to the Bitdeer-Volta Deal

Inconspicuously listed as the 10th bullet point, way down within Bitdeer’s August 4 press release announcing the deal with Volta, was this:

Institutional-grade Credit Support: Volta’s obligations are anticipated to be backed by Letters of Credit arranged by affiliates of J.P. Morgan and another top-tier global financial institution, totaling approximately $1.3 billion and subject to customary conditions.

This is critical. JPMorgan Chase’s (JPM) credit support was instrumental in facilitating the deal. And this $1.3 billion credit backstop essentially secures Volta’s payment obligations and helps protect Bitdeer against potential credit defaults by Volta, a company still in its infancy.

But it’s JPMorgan’s credit-decoupling structure that made it all possible. With JPMorgan and another top-tier bank putting together the $1.3 billion in letters of credit, Bitdeer didn’t have to underwrite a massive loan for a new startup (Volta) on an equally massive 16-year lease. Instead, it underwrote for two globally elite financial institutions.

Put simply, JPMorgan’s letters of credit “decoupled” Volta’s basically non-existent credit profile from the deal. This allowed Bitdeer to secure a major AI data-center lease without taking on Volta’s potential credit risk.

While this credit separation paved the way for Bitdeer’s pivot from crypto miner to high-margin data-center landlord, the fact remains that Bitdeer still needs up to $500 million to fund infrastructure costs and meet the aggressive deadlines. Bitdeer plans to finance through debt and a dilutive $1 billion stock offering.

Even as it flies somewhat under the radar, JPMorgan’s backing in this deal could very well represent a pivotal moment of the AI boom. By providing the $1.3 billion letters-of-credit backstop, JPMorgan told the world that Wall Street is ready to assume the risk inherent in brokering deals between startup AI companies and data-center operators.

In other words, AI-startup financing may have just transitioned from high-risk venture into institutional-grade backing.

This deal also opens the door for other bitcoin miners to pivot from volatile crypto mining toward AI infrastructure (as Bitdeer did) without taking on significant lending risks. And, of course, JPMorgan’s role in this deal encourages new AI startups to seek substantial data-center leases if they can secure bank-grade credit.

What This Deal Means for Bitdeer, Dell, Nvidia, and Anthropic

Besides providing the AI hardware and infrastructure for the data centers in Norway, Dell CEO Michael Dell and Nvidia backed Volta as it raised $300 million in venture funding, which valued the company at $2.4 billion. So, Dell and Nvidia are – literally – invested in Volta’s deal with Bitdeer.

But all parties involved stand to benefit. Here’s how.

Bitdeer (BTDR): $4.7 Billion Guaranteed, but a “Very Bearish” Rating

Bitdeer may gain the most from its deal with Volta, as the company almost instantly morphed from a struggling bitcoin-mining business into a landlord of four potentially high-margin AI data halls in Norway.

The company’s strategic decision to convert its energy-intensive bitcoin-mining campuses into data centers is already paying off, as the 16-year lease with Volta guarantees the company at least $4.7 billion – with the potential of that sum growing to $8 billion if the lease is extended.

Bitdeer could certainly use it. While its total revenue was strong during the second quarter of fiscal 2026 ($228.8 million, a year-over-year increase of 47%), the company suffered a year-over-year gross loss of $8.5 million and a net loss of $92.3 million – both significantly worse numbers than 2025’s, though improvements from the previous quarter. Additionally, its generally accepted accounting principles (“GAAP”) earnings per share (“EPS”) of negative $0.37 missed consensus expectations.

Bitdeer stock has been on a gradual downward slide since the start of the year, dropping nearly 25% as of August 11.

Chaikin Analytics, the investment-research platform founded in 2009 by legendary 60-year Wall Street veteran Marc Chaikin, rates Bitdeer “very bearish” in its Chaikin Power Gauge, a 20-factor stock-rating system that scans more than 5,000 stocks and 2,300 exchange-traded funds (“ETFs”).

Despite the bearishness, Bitdeer could rebound as it establishes itself more as a data-center landlord rather than a shaky crypto company.

Nvidia (NVDA): Backing the Customer That Buys Its Chips

It’s not as if the chip bellwether was relying on these Bitdeer/Volta/Anthropic deals to turn its fortunes around. It is, after all, the most valuable company in the world. But these types of deals certainly don’t hurt, especially as Nvidia rolls out its next-generation Vera Rubin AI architecture on a large scale.

The Bitdeer/Volta data-center campus is simply a showcase for Nvidia’s latest and greatest. But by backing – both financially and with its state-of-the-art technology – a specialized startup like Volta, Nvidia diversifies and expands its already sizeable footprint. And it will likely result in more startups lining up at Nvidia’s door, hoping for the same level of support.

Unsurprisingly, Nvidia gets a “very bullish” rating from the Chaikin Power Gauge.

Dell (DELL): PowerRack and Liquid Cooling Inside All Four Tydal Halls

Most people know Dell as a long-standing, successful, and reputable business within the consumer-computing industry. But the company has proven itself to be a major player in the AI world. These deals further solidify that sentiment.

The Bitdeer/Volta deal places Dell’s enterprise hardware, PowerRack server systems, and liquid cooling technology into the new data halls in Norway – another major milestone for Dell’s growing AI business.

And it’s that AI business that has pushed Dell stock up 245% since the start of 2026 (as of August 11) and prompted a “very bullish” Chaikin Power Gauge rating.

Anthropic: 133 MW and Less Reliance on Amazon

The Volta deal is huge for Anthropic on a couple of fronts. First and foremost, it gets 133 gross MW of power that it desperately needs to run the infrastructure that will train and deploy its next-generation Claude AI models and Claude Code.

Second, it secures this infrastructure using renewable, clean energy at a large scale. By tapping into Norway’s nearly carbon-free hydropower, Anthropic not only complies with strict European environmental regulations, but it also avoids the overly congested U.S. power grid and its yearslong connectivity wait times.

And this transaction also lessens Anthropic’s dependence on Amazon (AMZN), its main cloud provider. Expanding your supplier network, especially geographically, is never a bad idea in the AI world. Doing so will help Anthropic avoid bottlenecks, reduce geopolitical risk, and even improve performance.

The Risks: $500 Million Unfunded and a Chip That Just Started Shipping

While these agreements certainly have upsides for all involved parties, there are some risky elements… especially for Anthropic and Bitdeer.

Let’s start with Bitdeer. Per the agreement, the company must finish roughly $500 million of physical build-out on the Tydal data-center campus to meet both phases of delivery – the first by the end of 2026, the second by the end of next March.

The problem is, as of August 12, Bitdeer has not yet closed financing for those funds. If Bitdeer fails to meet those delivery deadlines, it faces potentially steep penalties and even the possibility of contract termination.

For Anthropic, the primary risk is the heavy reliance on a brand-new cloud infrastructure company (Volta) as the only safety net between itself and the hardware to be provided by Dell and Nvidia.

Speaking of hardware, that’s no sure thing either, considering large-scale production of Nvidia’s Vera Rubin platform only began in late May. Deploying the new hardware on the expedited timeline required by Bitdeer’s delivery deadlines undoubtedly introduces a layer of uncertainty.

Do the potential upsides outweigh the risks of these deals? That all depends on the level of risk investors are willing to take on. Sure, there’s a possibility that the Volta deal blows up for Anthropic. After all, it’s putting a lot of faith in a company that has only existed since January – a company that will be responsible for installing the hardware and managing the cloud-computing-power operation of a massive new data-center campus.

But that deal also secures much-needed computing capacity for Anthropic’s newest Claude models. And it expands the company’s global footprint.

For Bitdeer, this deal opens a brand-new revenue stream as a data-center landlord – one that should be far steadier and more lucrative than its bitcoin-mining business. We should know by next spring whether these deals proved successful not only for Bitdeer and Anthropic, but also for Dell and Nvidia.

Regards,

David Engle

Editor’s Note: Marc Chaikin, the founder of Chaikin Analytics, built an award-winning system that flagged Nvidia as a BUY before it soared as high as 45,000%. It also turned “bearish” on software stocks two months before they crashed this year. Now, Marc’s warning that a  “jump to lightspeed” has taken place behind the doors of a Silicon Valley AI lab – and says the repercussions are about to cleave the market in half this summer. This 60-year Wall Street legend has a FREE Hotlist of stocks to buy and an urgent Hitlist of stocks to sell now. Click for the full story, including stock names and tickers here… 

The post Bitdeer’s $4.7 Billion Data-Center Lease to Cloud Startup Tied to Anthropic AI Deal: Here’s Who Benefits appeared first on MarketWise.

Intel (INTC) raised $20 billion from a new stock offering this week, allowing it to turbocharge the company’s AI-infrastructure investments.

Initially, Intel announced in a brief press release on Monday that it aimed to raise $15 billion from the new stock offering. Investors didn’t react very positively to that news.

Between the dilution of existing shares and companies’ uninhibited spending on AI, many investors are skeptical of and, frankly, fatigued by this type of development. Not surprisingly, Intel’s stock, which closed at $101.65 on Friday, August 7, fell during pre-market trading, opening Monday morning at $98.26.

This unexpected announcement also raised some questions.

Why the surprise public stock offering? What does it mean for Intel and its investors? And is this offering a good buying opportunity at a lower stock price, or is it an AI warning sign?

Why Intel Sold $20 Billion of Common Stock

If you guessed artificial intelligence (“AI”), you are correct! What decisions aren’t driven by AI these days? If you haven’t paid much attention to Intel in the past few years, let’s get you caught up.

As I wrote in a May 1 article about Intel’s fast-rising stock, much of Intel’s recent success can be attributed to the revival of Intel Foundry Services.

Here’s an excerpt from that article:

In 2022 and 2023, Intel Foundry Services (“IFS”) earned external foundry revenue of $895 million and $952 million, respectively.

IFS revenue then exploded to $17.5 billion in 2024 and increased to $17.8 billion last year. The pattern continued into 2026’s first quarter, with Intel’s foundry business earning $5.4 billion, a nearly $1 billion jump from the previous quarter.

What sparked such a drastic revenue surge?

• Intel began manufacturing its own products, including the Core Ultra Series 3 and Xeon 6 processors, rather than having them manufactured by Taiwan Semiconductor.

• The AI data-center boom pushed demand for Intel’s Xeon processors to unprecedented levels.

• Intel’s application-specific integrated circuit (“ASIC”) custom chip business increased by more than 50% thanks to demand from cloud customers…

And what might be the best

AI demand is obviously still quite strong, so Intel is doubling down on its AI investments.

In its press release, Intel stated that:

Progress in emerging areas including physical AI, purpose-built silicon, advanced packaging and external wafers represent significant growth opportunities for Intel.

Thus, Intel decided to offer the common stock for sale and put the proceeds toward capital expenditures (“capex”), working capital, and growth opportunities. In other words, Intel is looking to grow its AI-infrastructure business significantly. And this offering would greatly expand the available capital Intel needs to do so.

That capital comes without Intel increasing its debt, which is important considering, in July, Intel upped its 2026 capex projection to more than $20 billion.

Other key details of the transaction:

• The offering’s share price came in at $95, meaning 210,526,315 new shares were made available for purchase.

• The offering’s underwriters will have a 30-day option (called a “greenshoe”) to purchase up to a total of $2.25 billion in additional shares of common stock at the public offering price, minus underwriting discounts.

• This would add another 31,578,947 shares of stock to the market, resulting in a total of around 242 million new shares of Intel common stock, based on the $95 share price.

• Wall Street titans JPMorgan Securities, Goldman Sachs, Morgan Stanley, and Citigroup are serving as joint book-running managers for the offering, reflecting the magnitude of this transaction.

As I mentioned earlier, Intel stock took a hit upon the initial news of the transaction, falling more than 4% on August 10. The reason? Concerns about dilution for existing Intel shareholders… a common worry whenever a company makes this much stock available for public sale.

And it’s a legitimate concern. Issuing $20 billion in new equity will, indeed, dilute Intel’s investors. But the other side of the argument is also valid. Intel stands to gain up to $20 billion in needed capital from the stock sale, giving the company tons of liquidity as it continues to invest in future growth. And it does so without increasing its debt.

The dip in stock price on August 10 may be a blip on the radar, but it’s worth monitoring for investors – especially those who own a sizable stake in the company.

The Federal Government’s 10% Stake in Intel Faces Dilution

About a year ago, the United States government, with the backing of President Donald Trump and his administration, purchased 433.3 million shares of Intel at $20.47 per share. That gave the federal government 9.9% ownership of Intel.

Whether coincidental or not, Intel has since taken off, seeing its stock price increase by nearly five times since the federal government’s investment last August. Intel and its investors – especially the government – have made a lot of money during that span.

Now, with this $20 billion stock offering, the government’s 10% stake is about to shrink. Based on the math used earlier, and the fact that the government did not participate in this sale, that 10% became roughly 9.5%.

Also, as part of the administration’s deal with Intel, the government received a five-year warrant at $20 per share for an additional 5% of Intel common shares, which, according to Intel, is “exercisable only if Intel ceases to own at least 51% of the foundry business.”

Because the August 2025 deal locked in 240.5 million warrants at $20 per share, the growing pool of total outstanding shares automatically dilutes the ownership of those warrants. In other words, the government’s 5% of additional shares from that warrant will also dip closer to 4.64%.

Unfortunately for the government, as a passive minority shareholder, it doesn’t hold any veto or voting power. So, there’s nothing Trump or his Cabinet can do to stop the offering. Ultimately, the government stayed on the sidelines.

On Monday evening, the only official comments from the administration came from White House National Economic Council Director Kevin Hassett, who told CNBC that it’s ultimately up to President Trump when the administration sells its stake – which it does not appear ready to do just yet. So, that’s also something to watch.

What the $20 Billion Offering Means for Intel Investors

We should learn the details of Intel’s stock offering after the sale closes on August 12, as the Wall Street banks running the transaction build an order book. At that point, we should have a better idea of how much Intel’s stock will be impacted.

But you can probably count on an Intel stock slide in the immediate aftermath of the transaction.

Why’s that? Dilution, for one. Then there are discounts. Underwriters often price newly available shares at a small discount (generally 2% to 5%) below the current trading price to entice institutional investors to spend millions or billions on the new shares. Sure enough, Intel’s stock was priced at $95, a discount of 2.6% from its August 10 ​close.

Finally, simple supply and demand negatively impact the stock’s price when millions of new shares are added without an instant surge in buyer demand. But considering Intel’s sale did see a big swing in buyer demand, this shouldn’t be much of a factor.

We don’t have to look back too far for similar examples:

• In 2020, as Tesla (TSLA) stock was soaring, Musk and company took full advantage of the high price by rolling out multiple multibillion-dollar capital raises. One example was a huge $5 billion “at-the-market” stock offering, which immediately pulled Tesla stock back, resulting in a loss of roughly 4.7% by market close that day.

• Earlier this year, Oracle (ORCL) announced it was raising nearly $40 billion in capital through mixed debt and equity offerings to bankroll its AI-infrastructure and cloud data-center build-out. The company’s shares plunged more than 12% in after-hours trading as investors lost their collective minds over the amount of money needed to support Oracle’s AI expansion.

But what about long-term impact? This is where things get more complicated. A short-term reaction – or overreaction – is one thing, because it’s typically based on impulsive investor and analyst emotion and sentiment, whether it’s sheer panic or total exuberance.

Long-term impact is generally driven by more rational and strategic thinking. It takes time to analyze and unpack how a company has performed after certain transactions have completed – in Intel’s case, the $20 billion public stock offering.

If you’re considering investing in Intel, you’ll want to see whether this transaction leads to real revenue and profit, whether there’s significant growth and expansion of its foundry facilities to handle demand for its AI hardware, and whether Intel can establish strong cash flow.

During the second quarter of fiscal year 2026, Intel operations generated an impressive $7 billion in cash flow. However, the company’s free cash flow (“FCF”) came in at negative $8.42 billion – compared with last year’s negative $1.1 billion – because of rising capex.

Intel is projecting another year of negative FCF, which would extend its streak to five straight years… the company accumulated a negative FCF of roughly $44 billion between 2022 and 2025, fueled by constant investments in AI and foundry expansions.

Here’s the thing: Intel is probably doing the right thing by increasing its capex to fund its AI infrastructure and meet customer demand. And doing so by diluting its existing shareholders was preferable to growing its debt by the billions, from Intel’s perspective.

The problem is, AI expenditures and financing are simply out of control.

Intel is seemingly doing everything right. Just look at the highlights from its July quarterly earnings report.

• $16.1 billion total revenue, well above the projected $14.4 billion.

• Data-center and AI revenue of $6.3 billion, a massive 59% year-over-year gain.

• Non-generally accepted accounting principles (“GAAP”) earnings per share (“EPS”) of $0.42, doubling expectations.

• A 40.4% gross margin, up from 27.5% year over year.

By any standard, that’s a highly successful quarter. Until you look at the company’s second-quarter net loss of $11 billion, up from $2.9 billion the year before.

All the positives are being negated by capex that appears to be unsustainable over the long term. And it shows no signs of slowing.

Intel Chief Financial Officer David Zinsner, who had already raised 2026 capex guidance from $18 billion to more than $20 billion, noted that 2027 capex would be “significantly above” that threshold.

So the questions investors must ask themselves are, “Will all of this pay off?” And, “If so, when?”

That’s yet to be determined. In the meantime, investors can take some comfort in knowing Intel is raising capital through equity rather than taking on more debt.

Which begs the question… Does this $20 billion stock offering make Intel a buy, or is this a warning that AI expenditures are simply too overwhelming for the company?

From my view, it’s another ominous warning of the prohibitive costs of financing the AI boom. Sure, the stock offering makes some sense from a bottom-line perspective. But when successful companies – and Intel has been riding a yearlong wave of success – start looking for different avenues to help them raise money to pay their seemingly endless string of capex bills, it’s worrisome.

The Bottom Line on Intel Stock

Intel has momentum. As I mentioned earlier, its stock price has nearly quintupled over the past year, clearly a positive sign. And, if Intel manages this stock offering the right way and makes smart investments with this capital, there’s no reason to think the company won’t continue along the right path, and its stock along with it.

But there’s no escaping the daunting shadow of AI capex that hovers over everything Intel does. The company has already burned billions in cash to expand its capacity, making some investors and analysts tired of the nonstop AI-related expenses that grow more exorbitant by the day.

It’s resulting in too many reputable companies with negative FCF despite growing revenues. And there’s no sure answer as to whether these massive investments will ultimately pay off. That’s just the uncertainty of the AI economy.

Intel is truly straddling the buy/sell fence right now. It’s possible the investments made using the capital from the $20 billion stock offering push Intel into “elite global foundry” territory alongside Taiwan Semiconductor Manufacturing (TSM). That makes Intel look undervalued.

But there’s also a possibility that Intel’s AI investments don’t pay off the way it envisions and the company never even remotely approaches anything close to that status. (Let’s face it, TSM has no peer and probably never will.) That, combined with shareholder dilution, is not looked upon favorably by investors.

The AI-expenditure crisis certainly isn’t unique to Intel. It spans the entire industry. But the spotlight is on Intel now, after its public stock offering. And the investing world will be watching its every move to see where it goes from here.

Regards,

David Engle

Editor’s Note: Whitney Tilson called the rise of Apple, Amazon, and Netflix… as well as the collapse of dozens of companies that went bankrupt. Now the former $200M hedge fund firm manager is stepping forward with what he calls the most important financial warning of his 30-year career. He’s sharing two free stock recommendations  (one to buy, one to sell immediately)  along with details of a new proprietary system fueling his predictions. See it all in his free presentation. 

The post Does Intel’s $20 Billion Stock Sale Signal a Buying Opportunity or a Warning Sign? appeared first on MarketWise.

Strategy (MSTR) continues to sell bitcoins for cash, plowing the proceeds into shares of its own STRC Series A preferred stock, as part of the firm’s recently announced “digital credit” plan.

In the week ending August 9, Strategy made several key moves:

• Sold common stock worth $653.1 million, at an average price of about $99.17.

• Sold 1,690 bitcoins for $108.6 million, at an average price of $64,262.

• Repurchased 1,152,020 shares of STRC Series A preferred stock for a total price of $108.6 million, or about $94.27 per share.

Strategy’s moves effectively exchanged its 1,690 bitcoins for more than 1.15 million shares of the preferred stock. The cash proceeds from the common stock went to the company’s USD Reserve, now at $4.65 billion, which backstops the payment of dividends and other interest.

After the sale, Strategy owns 840,447 bitcoins at an average purchase price of $75,385. It is the world’s largest bitcoin treasury firm, owning about 4% of the cryptocurrency’s total issuance.

The intent of the digital credit plan is to increase the number of bitcoins per common share by closely managing Strategy’s common stock, STRC preferred stock, and its bitcoin holdings. The recent sales bolster the firm’s balance sheet, raising investors’ confidence that Strategy can meet the payments on its preferred stock and other indebtedness without needing to raise cash in a fire sale, which might occur while bitcoin prices are low or at a similarly inopportune time.

Strategy’s focus on its STRC preferreds has worked out well for holders so far. The shares have bounced back significantly from a major June downdraft, after investors’ confidence was rattled. The digital credit plan may lead to further gains in the preferreds in the short term as well.

Strategy’s Digital Credit Plan: How it Works

Strategy’s recent portfolio moves are part of what executive chair Michael Saylor calls the company’s digital credit plan. This plan refocuses the company’s fundraising around its STRC Series A preferred stock, with Strategy managing its dividend yield and price, so that it can increase the number of bitcoins it owns per share of its common stock.

Saylor says that the STRC preferreds are already the “most liquid and largest preferred stock in the world.” By carefully managing the preferred, Strategy can raise cash more cheaply.

“We expect to keep laser focus on it,” said Saylor in the second-quarter earnings call. “We’re going to do everything we can to make it the most appealing credit in the entire digital credit space.”

This approach means that Strategy will keep a tight rein on its Series A preferreds so that they stay popular with investors while also offering a reliable source of funds for buying bitcoins.

In a real sense, then, Strategy is acting like a central bank of bitcoin, using some of the same tools as an actual central bank, such as adjusting interest rates and liquidity.

As I explained in this recent piece on Strategy’s new gambit, Strategy has the following tools:

• Strategy can use dollars to repurchase preferreds if they drift too low, or it can issue them for more dollars if the price reaches a favorable range.

• Strategy can adjust the yield on this preferred to raise its price, for example, if it wants to raise cash or if bitcoin is surging and its preferred stock enjoys investors’ confidence.

• By creating a greater market for this preferred, Strategy creates more liquidity in the financial instrument itself, meaning it can more easily issue the preferred and likely lower the cost of funding over time as investors come to trust the stock.

So, Strategy will take advantage of pricing in each of its markets – bitcoins, common stock, and preferred stock – for the purpose of raising the number of bitcoins per common share. To do so, it has recently raised cash to strengthen its balance sheet, giving it a more defensive posture.

At the same time, it has the potential to go on offense when the time is right to buy bitcoins, ideally giving it more ability to accumulate bitcoins if the STRC preferred becomes highly trusted.

So, Strategy’s digital credit plan requires a longer-term approach to acquiring bitcoins rather than the short-term “squeeze” of the bitcoin market that some traders may have first envisioned.

Is There Upside in Strategy’s STRC Preferred Stock?

A key part of Strategy’s approach is ensuring that its STRC preferred stock trades close to its par value of $100. At this price, it will likely issue more shares to fund its bitcoin purchases. Strategy’s moves here may offer a short-term opportunity to profit for its preferred investors.

Strategy can help STRC reach its $100 reference price in a few ways, including maintaining investors’ confidence that it will reliably pay its high dividend, adjusting the dividend to attract investors, and buying back the preferred stock in order to push up its trading price.

In fact, Strategy has recently made all three of these moves:

• Management’s focus on the preferred share price and a more resilient cash-heavy balance sheet have helped restore confidence in its preferred stock.

• Strategy raised the payout on its STRC preferred stock to a 12% annual rate in June.

• Strategy authorized a $1 billion buyback for the preferred series and has already begun deploying the cash.

The result? The STRC preferreds are up solidly from their 52-week low of $71.25 in June to around $95 today. Plus, there may be even more upside as the company looks to return the stock to par.

Saylor effectively promised to return the preferreds to par in this time frame, almost down to the day, in the company’s last earnings call, though without making any explicit promise to do so.

“When we came public with STRC, it took us about 70 trading days to reach par,” said Saylor. “Some people forget this, but when we did the IPO, we IPO’d at $90. It took us 70 days to get to par…. If we were to trace forward 70 trading days from when STRC fell out of our trading range, that would put us around September 8.”

Saylor noted on the call that if the $1 billion authorization doesn’t work,

We have ample additional capital in order to cure the dislocation. We’re currently in a discovery process to figure out how much capital will be required on our part in order to return STRC to par. We’re certain that we will return STRC to par.

So, expect further upside for the STRC preferred series for now. Then tack on that heavy dividend – which looks safe at least in the short term – while you wait for shares to move higher.

However, Strategy will have a tough balancing act down the road, since it will want to lower the shares’ steep 12% coupon. But if it lowers the payout, will the stock remain near par, or will Strategy be forced to buy back shares in order to keep the price in management’s target range of $99 to $100?

Strategy’s refined digital credit plan sees the company playing all sides of its markets to keep the number of bitcoins per common share rising over time. We’ll watch how well it does.

Regards,

James Royal, PhD

Editor’s Note: Every generation or so, the way money moves gets a fundamental upgrade. The people who see it coming have the chance to get extraordinarily wealthy. Everyone else watches from the sidelines. Luke Lango says that moment is here again, and Elon Musk is behind the upgrade in an amazing way. He’s revealing exactly what to buy, including one free pick, in this presentation.

The post Strategy’s New Bitcoin-Backed ‘Digital Credit’ Plan: Does the Juiced Preferred Stock Have Upside? appeared first on MarketWise.

OpenAI CEO Sam Altman is aiming for a $1 trillion valuation for the initial public offering (“IPO”) of the artificial-intelligence (“AI”) model company. But the increasingly poor economics of AI and the company’s terrible financials mean that even an already delayed IPO may never arrive.

OpenAI is facing white-hot competition, with the company now severely marking down the price of high-end AI models created just weeks before. Altman and his public relations team tried to balance that news with the fact that OpenAI’s customers now comprise more than 1 billion users and more than two million businesses. Instead, investors need to pay attention to OpenAI’s lack of pricing power.

In late July, Altman announced the following price cuts on two recently adopted models and an upgrade on a third model.

AI model	Feature	Price cut
GPT‑5.6 Luna	Fastest and most affordable model	-80%
GPT‑5.6 Terra	Balanced model for everyday work	-20%
GPT‑5.6 Sol	Now offers Fast mode	None

As part of the announcement, OpenAI explained that GPT-5.6 had become more efficient to run, so the company said: “We’re passing those gains on to customers with lower prices…”

It’s a sales pitch that feels a lot like a seedy car salesman trying to “get you in a new car today.” But businesses aren’t usually trying to lower prices – their profit driver – unless they need to. And that’s even more true for a company’s latest and greatest models, as is the case at OpenAI.

That’s the tough situation OpenAI finds itself in today, as Chinese AI rivals such as Kimi K3 and DeepSeek models make heavy incursions into OpenAI’s user base. They’re releasing high-power AI models that do most of what OpenAI’s (and Anthropic’s) models do, but they charge a fraction of the price, sometimes just a few percentage points of what higher-end models charge.

It’s a price war that OpenAI simply can’t afford, one way or the other. OpenAI is already burning cash furiously and must be propped up by its customers. At the same time, it needs revenue to make good on an estimated $1.4 trillion in spending commitments in the coming years.

Either way, the company may well go broke. Many firms that are counting on it for revenue – Oracle (ORCL), Cerebras Systems (CBRS), and CoreWeave (CRWV) – may be badly hurt.

So, OpenAI must raise money to continue operating, but investors need to see a credible path to profitability before Altman’s IPO goal of $1 trillion can ever come to fruition. This impasse means that OpenAI has tentatively delayed its IPO until 2027, but it may be “delayed” permanently.

OpenAI’s Cash Burn Is Probably Getting Worse

The unprecedented scale of OpenAI’s ability to lose money is something to behold. OpenAI reported an operating loss of $20.9 billion in 2025 and estimates for the first quarter of 2026 peg its operating loss at a further $7 billion. At this rate, OpenAI could easily lose $30 billion in 2026 and possibly even $35 billion to $37 billion. We’ll have to wait for leaked financials for a better idea.

But a price war won’t likely improve profitability, even if it does increase sales for the moment.

In fact, we’re seeing OpenAI’s sales continue to soar, though the company’s executives provided only very high-level financials. CFO Sarah Friar told employees that OpenAI’s July annualized recurring revenue had exceeded the company’s total for the second quarter, according to CNBC.

While that feat may actually be less impressive than it sounds at first, it does indicate a ramp in sales at the end of the quarter. But it may well be one that the company can’t afford.

The interesting detail here is what leaders attributed the growth to: the company’s enterprise agent ChatGPT Work, its AI coding tool Codex, and its GPT-5.6 series of models.

That is, the same GPT model that OpenAI put on sale at a severe discount at the end of July.

For normal companies, sales growth translates into scaling profitability. In OpenAI’s case, higher sales translate into even greater losses, and we have plenty of signs that OpenAI is burning more cash than ever, as it’s forced to lower prices to keep getting business.

So, higher sales are not good news when you’re losing more money in absolute terms.

The inability to price AI models so that they’re profitable shows that AI models are mostly fungible. Businesses are shifting to the lowest-cost model to carry out their work, and some are reserving the high-cost models for their best and highest uses. But Chinese AI models have been showing that they, too, can move upmarket, leaving less room for OpenAI here, too.

For example, among AI startups using open-source software, about 80% are using Chinese AI models, according to Andreessen Horowitz.

While OpenAI struggles to get itself in shape for an IPO, China’s DeepSeek is planning its own public offering by the end of 2026 or early next year.

OpenAI’s Next Funding Round

All this means that OpenAI is likely to be burning even more cash at an even faster clip than before. The upshot then is that it will need to raise another funding round sooner than some may have thought. OpenAI last raised $122 billion in March, valuing the company at $852 billion.

OpenAI is under significant constraints, however. Venture-stage companies do not like to raise money at a lower valuation than the prior funding round. It indicates that something’s not right and dings the company’s reputation. That reputational worry is even greater in the field of AI, which is being hyped as much as any technology ever has been. If OpenAI can’t raise its valuation in the next funding round, it will look like there’s something wrong. (And there is…)

OpenAI needs to raise money at ever-higher valuations to keep the market’s confidence up and so that it can retain the potential to raise even more money in the future.

Of course, Altman wants to line up funding before he needs it, too. Increasingly larger losses in future quarters will hurt investors’ confidence to put up more cash. At the same time, ready cash that can subsidize its ongoing losses means OpenAI is not immediately subject to the market’s whims, giving Altman some maneuvering room if funding conditions tighten for a while.

Like the last funding round in March, OpenAI’s next round is likely to be funded by mega-cap companies that need the AI party to continue. Last time out, Amazon (AMZN), Nvidia (NVDA), and SoftBank (SFTBY) put up $110 billion of the $122 billion raised by OpenAI. Investors may need to raise cash from the debt markets themselves to plow into the money-losing OpenAI.

OpenAI is a linchpin in the AI economy, since it’s promised to spend literally more than $1 trillion across some of the biggest AI-related firms. If it’s unable to meet these commitments – and it must raise the cash to do so – the dominoes will begin to fall across the entire AI sector.

Confidence is key in these situations, and investors are already feeling skittish after a poor debut from SpaceX (SPCX) and massive declines in memory stocks such as Micron Technology (MU). The inability of OpenAI to raise at a higher valuation could be a sign that the Teflon stock market for AI stocks may be about to turn. So, keep a close eye on OpenAI’s ability to raise funding.

Regards,

James Royal, PhD

Editor’s Note: One company quietly beat Apple, Amazon, and the S&P 500 — combined. Over the long run it has returned more than 8,300%… enough to turn $10,000 into $830,000. Yet most investors have never heard of it. Former fund manager Whitney Tilson says right now it’s trading at a rare discount — and he’s revealing the name and ticker, free of charge. Click here for the details…

OpenAI IPO Delay FAQ

Why is OpenAI’s IPO delayed?

OpenAI has tentatively delayed its IPO until 2027. Sam Altman is holding out for a $1 trillion valuation, but investors need to see a credible path to profitability before that can happen, and OpenAI hasn’t demonstrated one.

When will OpenAI’s IPO happen?

OpenAI has tentatively set 2027 as its new timeline, after already delaying the offering once before. Given OpenAI’s mounting losses and the pricing pressure it’s facing from competitors, that date isn’t set in stone.

Could OpenAI’s IPO never happen at all?

It’s possible. OpenAI’s already-delayed IPO may never arrive at all, given the increasingly poor economics of AI and the company’s own terrible financials. What looks like a delay to 2027 could turn out to be permanent.

Why is OpenAI cutting the price of its AI models?

In late July, OpenAI cut prices on GPT-5.6 Luna by 80% and Terra by 20%, just weeks after launching them. OpenAI said it was passing along efficiency gains, but the cuts come as Chinese AI rivals like DeepSeek and Kimi K3 offer similar models for a fraction of the cost.

How much money is OpenAI losing?

OpenAI reported a $20.9 billion operating loss in 2025, with estimates putting its first-quarter 2026 loss at another $7 billion. At that pace, OpenAI could lose $30 billion to $37 billion in 2026.

How much has OpenAI raised in funding, and at what valuation?

OpenAI raised $122 billion in March at an $852 billion valuation, with Amazon, Nvidia, and SoftBank contributing $110 billion of that total. OpenAI will likely need to raise again soon, and doing so at a lower valuation than its last round would signal to the market that something is wrong.

Which stocks could be hurt if OpenAI’s IPO stalls or fails?

Companies that depend on OpenAI for revenue, including Oracle, Cerebras Systems, and CoreWeave, could be badly hurt. OpenAI has promised more than $1 trillion in spending commitments across the AI industry, and if it can’t raise the cash to meet them, the dominoes could begin to fall across the entire AI sector.

The post OpenAI IPO Delay: Just the Start? Price Cuts Hint It May Not Happen at All appeared first on MarketWise.

Companies don’t give away $200 million businesses.

Except Boeing just did. On purpose. And if you understand why, you understand where the next wave of AI money is headed.

On Monday, the Boeing-Archer deal was announced. Boeing (BA) handed three of its subsidiaries — Wisk Aero, SkyGrid, and Insitu — to Archer Aviation (ACHR). Three businesses. Two decades of investment. One of them generating over $200 million a year in defense revenue across 35 countries, with drones flown by the U.S. Navy.

The price? Zero dollars.

Boeing took stock instead. A 19.75% stake in Archer, plus options to buy more over the next four years — all laid out in a regulatory filing, not a marketing deck.

The market heard it. ACHR stock ripped 20% by mid-morning.

So why would Boeing do this? Simple. I spent 28 years on trading floors in Chicago, and I watched this exact trade a hundred times: when a market maker holds a position he can’t run properly, he doesn’t dump it at the bid — he swaps it to the guy who can run it and keeps a piece of the upside. That’s what Boeing did. CEO Kelly Ortberg has been shedding everything that isn’t commercial planes, defense, or space since he took over — he sold the Jeppesen flight-planning unit for $10.55 billion last year. But this time, instead of taking cash and walking away, Boeing kept a nearly 20% seat at the table and licensed back Wisk’s autonomy tech for its own future aircraft. Sell the headache, keep the upside. That’s not desperation. That’s a pro’s trade.

But here’s the part most investors will miss: the deal announcement gave Wall Street something more valuable than a stock pop.

It gave Wall Street new language.

“Physical AI” Just Entered the Chat

Archer didn’t call this a drone acquisition. It didn’t call it an eVTOL consolidation.

It called it an “end-to-end physical AI platform for aerospace and defense.”

Physical AI. Remember that phrase. You’re going to hear it a thousand times over the next year.

Here’s the plain-English version. The first wave of AI lived in a data center. It wrote your emails. It answered your questions. It made Nvidia the most valuable company in history.

Physical AI is what happens when that intelligence gets a body. Aircraft that fly themselves. Drones that navigate without a pilot. Air traffic systems that manage thousands of autonomous vehicles at once.

That’s what Archer just bought:

• Wisk Aero — six generations of autonomous electric aircraft, more than 1,700 test flights
• SkyGrid — the air traffic management software that keeps autonomous aircraft from hitting each other
• Insitu — military drones flown worldwide, including by the U.S. Navy, with 3,500+ systems built and $200M+ in annual revenue

Feed all of it — nearly two million combined flight hours of data — into Archer’s AI foundation model, called ZEE. That’s the platform.

Whether Archer executes on that vision is a different question. But the theme just got a name, a poster child, and a Boeing endorsement. Themes with all three don’t stay quiet.

The Tape Told You Everything at 10 A.M.

I don’t care what a press release says. I care what the money does.

So here’s what the money did on Monday.

ACHR jumped 20% — fine, that’s the deal stock, it’s supposed to move. Part of that pop was mechanical anyway: nearly 15% of Archer’s float was sold short, and shorts covering into a takeover headline is rocket fuel that has nothing to do with conviction.

The real tell was everything around ACHR:

• AeroVironment (AVAV) — up 9%
• Kratos Defense (KTOS) — up nearly 6%
• Red Cat Holdings (RCAT) — up 6%
• Planet Labs (PL) — up 5%

None of those companies were in the deal. Not one.

That’s called a sympathy move, and it’s one of the most honest signals in markets. Nobody issued a press release telling AVAV to rally 9%. Real buyers showed up because the deal re-priced what an autonomous-systems business is worth. Insitu — a drone maker doing $200 million a year — just got valued inside a “physical AI” wrapper, and every fund manager holding a comparable business did the same math before lunch.

Meanwhile Boeing itself? Up less than 1%. The market shrugged at the seller and chased the theme.

That’s what the tape can’t hide, folks. The money went straight to the drone complex.

Follow the Money: 3 Ways to Play Physical AI Stocks

I’m not going to hand you a “buy this now” list. What I’ll do is show you where the money is pointing, so you can do your own work.

The pure comp: AeroVironment (AVAV). If you want to know what Insitu is, look at AVAV. Military drones, loitering munitions, U.S. defense contracts. It led Monday’s sympathy rally for a reason — it’s the closest public company to the business Boeing just handed Archer. When the theme gets a re-rating, the cleanest comp gets it first.

The autonomy arms dealer: Kratos (KTOS). Kratos builds tactical drones and the autonomy software that flies them. If “physical AI” becomes the label funds put on their defense-tech sleeve, Kratos sits in the middle of it.

The small-cap torque: Red Cat (RCAT). Higher risk, higher beta. Red Cat’s Teal drones won a U.S. Army short-range reconnaissance program, and the stock moves violently when the drone theme catches a bid. This is the name that goes up the most when the theme is hot — and down the most when it isn’t. Size accordingly.

And a word on the obvious one everybody will mention: Palantir (PLTR) is the software layer of battlefield AI, but it’s already priced like everyone knows it. The fresh money in a new theme usually flows to the names that haven’t been discovered yet.

Read the Fine Print Before Chasing the Boeing-Archer Deal

Two things to keep in your back pocket before you chase anything.

The deal isn’t done. It’s expected to close by the end of 2026, pending antitrust review. Between now and then, Archer is a story stock with a $200 million revenue business attached to a press release, not a balance sheet.

Archer paid in paper. This was an all-stock deal. Archer just issued roughly 20% of itself to Boeing. Dilution is real, and existing shareholders paid for these assets whether they realize it or not.

None of that kills the theme. It just means the theme and the ticker are two different trades. You can believe in physical AI without believing ACHR at any price.

Physical AI Stocks: The Bottom Line on the New Theme

Twenty-eight years of watching order flow taught me one thing above everything else: nothing matters until the money moves.

On Monday, the money moved. Boeing swapped $200 million of annual revenue for equity in a startup. The drone complex re-rated in hours on a deal it wasn’t even part of. And Wall Street got handed a brand-new theme with a Fortune 50 company’s signature on it.

Chatbots were the first act. Physical AI — machines that fly, see, and decide — is the sequel.

The smart money already bought its ticket. The question is whether you’re watching the screen or the popcorn line.

Editor’s Note: Every generation or so, the way money moves gets a fundamental upgrade. The people who see it coming have the chance to get extraordinarily wealthy. Everyone else watches from the sidelines. Luke Lango says that moment is here again, and Elon Musk is behind the upgrade in an amazing way. He’s revealing exactly what to buy, including one free pick, in this presentation.

The Boeing-Archer Deal FAQ

Why did Boeing sell Wisk, SkyGrid, and Insitu?

Boeing is streamlining under CEO Kelly Ortberg to focus on commercial aircraft, defense, and space. Rather than selling the three autonomy-focused units for cash, Boeing traded them for a 19.75% stake in Archer — keeping upside exposure to the technology while removing the units from its own cost structure. Boeing also retains licensed access to Wisk’s autonomous flight systems.

What is physical AI?

Physical AI refers to artificial intelligence embedded in machines that operate in the real world — autonomous aircraft, drones, robots, and the systems that manage them — rather than software that only generates text or images. The Boeing-Archer deal made it a Wall Street buzzword by branding the combined Wisk, SkyGrid, and Insitu businesses an “end-to-end physical AI platform.”

Why did Archer Aviation stock go up?

Archer (ACHR) surged roughly 20% after announcing it would acquire three Boeing subsidiaries — including profitable defense drone maker Insitu — in an all-stock deal that gives Boeing a 19.75% stake. Heavy short interest (about 15% of the float) amplified the move as short sellers covered.

What did Boeing get in the Archer deal?

Boeing received a 19.75% equity stake in Archer’s Class A shares, options to purchase additional shares over four years, and continued access to Wisk’s autonomous flight technology for its own current and future aircraft. Boeing took no cash.

What are the best physical AI stocks?

There’s no official “physical AI” index yet, but the names the market treated as beneficiaries on deal day were AeroVironment (AVAV), Kratos Defense (KTOS), Red Cat Holdings (RCAT), and Planet Labs (PL) — all up 5–9% on the news despite having no role in the transaction. Always do your own research; sympathy rallies can fade as fast as they appear.

Is the Boeing-Archer deal complete?

No. The companies signed definitive agreements on August 10, 2026, but the transaction is expected to close by the end of 2026 and remains subject to antitrust review.

Want to see how a 28-year floor trader reads moves like this in real time? Jonathan Rose goes live every weekday at 11 a.m. ET on Masters in Trading Live — breaking down the tape, unusual options activity, and trades like these as they happen. It’s free. Search “Masters in Trading Live” on YouTube and pull up a chair.

The post Boeing-Archer Deal: Why Give Away a $200 Million Business — on Purpose? appeared first on MarketWise.

Since the artificial-intelligence (“AI”) boom began, the industry has faced its share of bottlenecks. Power continues to be a major choke point, as America’s grids can no longer accommodate the immense energy needs of AI data centers. Thermal management has also been a significant issue, with data-center servers and equipment requiring proper cooling to prevent overheating.

But perhaps the biggest AI bottleneck to date has been memory. And that shortage has been building for months.

During Micron Technology’s (MU) first-quarter earnings call on December 17, CEO Sanjay Mehrotra stated:

Over the last few months, our customers’ AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage. We believe that the aggregate industry supply will remain substantially short of the demand for the foreseeable future…

Micron is working hard to support our customers’ demand during this time, and we expect to grow our [dynamic random-access memory (“DRAM”)] and NAND bit shipments approximately 20% in calendar [year] 2026. Despite significant efforts, we are disappointed to be unable to meet demand from our customers, across all market segments.

After Micron’s second-quarter earnings were released on March 18, Mehrotra told CNBC:

We are only able to supply, for our key customers in the midterm, about 50% to two-thirds of their requirements.

We’re now into August, and the supply of memory hasn’t improved. The world’s leading memory manufacturers – Micron, SK Hynix (SKHY), and Samsung Electronics (005930.KS) – have sold out their entire DRAM and high-bandwidth memory (“HBM”) not just through the rest of this year… but through all of 2027.

That should be good news for these companies, right? Sellouts mean that all of your product has been purchased, with backlogs promising millions or billions in future revenue. So, why has Micron stock sunk from its all-time closing high of $1,213.56 on June 25 to its August 3 closing price of $829.50? That’s a nearly 32% drop. In the same time frame, Samsung, which trades on the Korea Exchange, has fallen roughly 33%. And since SK Hynix went public on the Nasdaq on July 10, its price is down 15% through August 3.

What should investors make of this broad decline in the top memory stocks?

Why the 2027 Memory Sellout Matters

As bad as 2026 has been for memory shortages, 2027 will likely be much worse, at least according to SK Hynix CEO Kwak Noh-jung, who told Reuters that 2027 will likely be “the worst year in the industry’s history from the supply perspective.” He also projected that customer demand will exceed supply capacity “even beyond 2030.”

Here’s why: Memory manufacturers have been encouraging customers to sign long-term (three-to-five-year) contracts, “using an advance payment deposit model where customers essentially pre-pay for some of the memory capacity,” according to TechPowerUp.

With all of 2027’s memory allocation already being sold out, the only customers who will receive any memory supply are those who signed these agreements. Everyone else looks to be out of luck. There literally isn’t enough DRAM available. Those businesses that didn’t sign contracts with the memory manufacturers will be without the memory they need in 2027 and forced to wait until the 2028 supply is made available to order.

Even for those customers that did sign contracts, they’re still not guaranteed the memory supply in 2027. That’s because most of the memory supply will go to AI data centers, hyperscalers, and cloud-service providers. For example, Samsung has five-year agreements in place to supply the world’s five largest data-center operators with the memory they need… and that amounts to between 60% and 70% of Samsung’s memory production capacity.

It’s undoubtedly a smart business move by Samsung because it locks in steady revenue for the next five years. But who suffers the most from these deals? Everyday consumers.

With up to 70% of all memory production going everywhere but into consumer-electronics items, simple math tells us that only 30% of the memory supply remains for everyday products like phones, computers, tablets, and gaming consoles, to name a few.

And when supply is that limited, Economics 101 informs us prices rise. In fact, if you’ve shopped recently for some of these products, you’re already seeing the ripple effect.

So far, this year:

• Steam PC prices have increased between $240 and $300.

• Xbox and PlayStation game console prices have jumped between $100 and $150.

• Apple (AAPL) iPads are between $150 and $200 more expensive, and MacBook prices have soared by $100 to $300.

• Computer and server makers Dell Technologies (DELL), HP (HPQ), Acer, Asus, and Lenovo have also raised prices between 15% and 20%.

On top of the price hikes, global smartphone shipments are projected by IDC’s Worldwide Quarterly Mobile Phone Tracker to finish 2026 down nearly 14% from last year. That would represent the largest annual decline in the history of smartphones. IDC also projects a 1.1% decline in 2027.

All of this trickles down to other types of memory as well. With DRAM silicon wafers in such short supply, electronics manufacturers are turning toward alternative memory options, like NAND flash and solid-state drives (“SSDs”). And that higher demand is pushing those prices up.

With all that in mind, it’s fair to ask why memory manufacturers’ stocks continue to fall.

Why Memory Stock Prices Are Falling Despite Record Sales

There isn’t one clear answer as to why memory stocks have been sinking even as memory manufacturers have completely sold out their supply through next year. Multiple factors are contributing to the stocks’ declines.

For one, as the market looks ahead, there are serious concerns about how sustainable this nonstop AI spending is, even if most of the companies spending are flush with cash. Investors cast a wary eye on hyperscalers and other tech companies and wonder whether this head-spinning spending will slow.

Second, the valuation of stocks like Micron and SK Hynix has been bolstered by huge price surges. While those increases may have been warranted based on the seemingly endless demand for their products as well as the subsequent sold-out inventory, the massive growth may already be baked into their stock prices. That leaves Micron and SK Hynix stocks with a paper-thin margin for error if demand for memory wanes even slightly.

Finally, this is just how the market works. Before the recent memory-stock dip, Micron, for example, experienced a huge move higher. For example, investors could have bought Micron shares for $367.85 on April 1. By June 25, the stock’s price rose to $1,213.56. That’s an increase of 230%… in less than three months. After an otherworldly surge like that, Micron was bound to fall back to Earth – which it has done since, falling roughly 27% (as of August 6) since its late-June high.

That’s simply a product of investors selling high and taking their profit, which brings the stock price back down. The market also tends to correct itself when stocks get ahead of themselves (see SpaceX). This volatility is likely the new norm for memory stocks, which will probably continue to see periods of big gains followed by pullbacks. This isn’t a reflection on the products or the demand, however. That remains real.

How New High-Bandwidth Flash Memory Impacts the AI Build-Out

A brand-new technology has entered the mix. On August 3, flash memory giant Sandisk (SNDK) announced that it, along with SK Hynix, had released the High-Bandwidth Flash (“HBF”) technical specification through the Open Compute Project.

Sandisk and SK Hynix were joined in this venture by Google and AI startup Tenstorrent to create this specification, which, according to Sandisk’s press release, “provides companies and developers designing AI inference systems and accelerators with a common technical framework for incorporating HBF technology where larger, near-compute memory capacity and higher bandwidth are needed to improve power and performance metrics and help reduce total cost of ownership.”

Near-compute memory capacity is essentially HBM placed close to computing cores, which is what AI inference systems need to process today’s AI workloads. The new HBF technology accomplishes this by merging high bandwidth with high capacity.

HBF basically bridges the gap between fast HBM and slow SSDs by stacking NAND flash memory, which provides high storage capacities at speeds that rival HBM. This technology has the potential to loosen AI inference bottlenecks and reduce costs, as HBF is substantially cheaper than HBM because it removes expensive DRAM from the equation.

Plus, HBF could potentially alleviate the AI data-center power problem. Most data centers today use HBM built with DRAM, which consumes power while constantly refreshing itself. HBF, using NAND flash memory, doesn’t require those refresh cycles. That means less power is used during AI inference, which could help lower costs.

Keep in mind, however, that HBF is just in its blueprint stage. It’s not a buyable product yet. Sandisk and SK Hynix expect HBF memory samples later this year, with actual AI-inference products possibly available in early 2027. This is a development worth watching, as it could play a large role later next year if it’s deployed at a commercial scale.

What to Watch for Going Into 2027

We know that all of Micron’s, SK Hynix’s, and Samsung’s memory inventory is sold out through 2027. So, what should investors watch for in the memory industry and its stocks going into next year?

We’ve alluded to some of these already, but they all bear repeating.

• The influx of HBM and DRAM memory alternatives: If data-center operators and consumer-electronics companies want memory, they don’t have any other choice but to look for alternatives to HBM and DRAM since there isn’t any available. Keep a close eye on different types of memory, such as NAND flash memory, high-speed graphics memory (GDDR6 / GDDR7), enterprise SSDs, and the emerging HBF memory.

• The companies making those memory alternatives: SK Hynix is still one to watch, as it manufactures high-speed graphics memory, enterprise-grade NAND flash, and data-center SSDs, as well as the sold-out HBM and DRAM. The South Korean company, as mentioned earlier, was also instrumental in the development of HBF. Samsung also produces the same memory types as SK Hynix and is now developing its own HBF solutions. And Micron is one of the world’s leading makers of high-speed graphics memory, NAND flash wafers, and enterprise SSDs.

Besides those “Big Three,” however, there are other major memory players. The largest being Sandisk, the longtime leader in advanced flash memory that partnered with SK Hynix in the development of HBF technology. In fact, Sandisk’s stock has seen some of the biggest growth in the history of the American stock market. A year ago, Sandisk was trading for roughly $42 per share. As of August 6? $1,258.58 per share. That’s a mind-boggling 2,890% increase in one year.

And that’s actually down from its June 25 closing high of $2,335 (which means, at that point, Sandisk was up 4,841% year over year). This backs up my earlier point regarding the volatility of the memory sector. After its absolutely massive surge, the market pulled back as investors sold off.

While NAND flash memory is a strong alternative, it’s also important to keep in mind that Sandisk, Micron, and Samsung have also sold out their full-year 2026 NAND flash capacity, and SK Hynix expects its inventory to sell out by the end of this month.

• The memory-making wild cards: You may not know much about Asian memory manufacturers Kioxia or CXMT. But they’re worth monitoring.

Kioxia, which trades on the Tokyo Stock Exchange, is actually a spinoff from Toshiba, the Japanese electronics giant that invented NAND flash memory. So, it’s no surprise that Kioxia is a global leader in the manufacturing of NAND flash memory as well as SSDs.

ChangXin Memory Technologies (“CXMT”) is a Chinese manufacturer of DRAM chips for smartphones, computers, servers, and other consumer-electronics products.

This isn’t a recommendation to invest in Kioxia or CXMT. Rather, it’s an alert to watch these Asian tech giants because they could very well eat into the profits of Micron, SK Hynix, and Samsung.

CXMT, for example, is quickly ramping up its DRAM production. In fact, it currently holds a significant 8% of the global DRAM market. And CXMT’s late-July IPO on Shanghai’s STAR Market closed around 466% above its offer price, ensuring it has plenty of financial heft (roughly $8.6 billion) behind it to keep expanding its production.

How Risky Are Memory Stocks?

The massive gains have certainly been enticing. Anyone who bought Micron or Sandisk stock at this time last year must like what they see in their portfolios right now. However, as lucrative as the memory industry has been, some risks come with investing in it.

As I mentioned, the memory industry is volatile. Since late 2022 and the start of the AI data-center boom, it has been riding a massive growth wave that hasn’t fully crashed down yet. One only needs to look at the recent stock growth of all the companies discussed in this article.

Until recently, that is. Consider this recent downturn part of the boom-and-bust cycle.

Back in May, William de Gale, a portfolio manager at BlueBox Asset Management, told CNBC that the industry typically experiences “enormous ups and downs” and that, “in the long run, it’s a pretty dreadful industry.”

While de Gale’s view is pretty grim, his point is valid. Investors in the memory industry better buckle their seat belts, because they’re in for a wild ride.

Consider the five-year betas, which measure stock volatility compared with the overall market, for Micron (2.21), SK Hynix (2.41), and  Samsung (1.56), all of which are roughly 1.5 to 2.4 times more volatile than the market.

Investing in memory stocks through 2027 may produce big gains simply because of supply and demand. But what happens in 2028, when more memory is expected to be available? Again, supply and demand come into play. As more product becomes available, prices should drop, which would result in tighter profit margins… and more than likely lead to a pullback in memory stocks.

Is that a guarantee? Of course not. Nothing investment-related ever is. And who knows what may unfold between now and 2028? But simple economics has stood the test of time for a reason. So there’s a strong chance memory stocks suffer as more capacity is manufactured.

Similarly, what happens if hyperscalers and AI companies tighten the reins on their spending? If you’re considering investing in memory stocks, pay close attention to what Amazon (AMZN), Alphabet (GOOGL), Meta Platforms (META), and Microsoft (MSFT) are spending on AI infrastructure over the next year or so. If there’s a downturn in spending, you can bet memory stocks and their investors will feel the consequences.

And don’t forget about companies like CXMT and Kioxia as they expand production and threaten to take a larger bite out of the global memory pie.

The bottom line? Memory stocks are a fascinating play. Despite this recent downturn, which we can likely attribute to simple valuation and market correction driven by selldowns, the industry is as strong as it’s ever been. And with AI’s demand for memory not expected to die down anytime soon, its stocks could bounce back soon.

Regards,

David Engle

Editor’s Note: What ever happened to the AI stock boom? Even AI darlings like Nvidia have essentially gone nowhere since summer 2025. Our friend and colleague at InvestorPlace, Louis Navellier, may have the answer. According to Louis, the AI industry is quietly “staging” ahead of the next great AI breakthrough… a new class of AI he calls “Superintelligence… but better.” How will it trigger a $100 trillion reset of the AI markets. How will the launch of this tech send some stocks to zero, and others soaring? And why does Louis say: Don’t buy or sell an AI stock in 2026 until you see what’s coming next? Go here for the full story (and Louis’ #1 pick).  

The post Memory Capacity Is Sold Out Through 2027. Why Have Micron, SK Hynix, and Samsung Shares Tumbled? appeared first on MarketWise.

Major technology sectors – semiconductors, software, and artificial intelligence (“AI”) – are some of the stock market’s hottest investments right now. With major tech companies investing in AI at record levels, many related stocks have surged to all-time highs over the past year.

For investors looking to ride this investment wave, it can be challenging to pick individual tech stocks. An easier way to get diversified exposure to these companies and find winners is to invest in some of the best tech exchange-traded funds (“ETFs”), often for a low cost.

The best tech ETFs in chips, software, and AI have returned 30%, 40%, and even 50% annually over the past three years – without investors needing to analyze and trade stocks.

While tech darlings such as Nvidia (NVDA) and Broadcom (AVGO) have raced higher for years, the big winners aren’t all trillion-dollar mega-cap stocks. In fact, even smaller chip companies and lesser-known software and AI companies can offer huge returns in the investment boom.

So, the best tech ETFs offer an accessible way for investors of all skill levels to invest in a portfolio of winning stocks across the semiconductor, software, space, and AI sectors.

Top Tech ETFs to Invest in for 2026

Fund (ticker)	One-year return	Three-year annualized return	Expense ratio
VanEck Semiconductor ETF (SMH)	99.0%	55.3%	0.35%
Invesco PHLX Semiconductor ETF (SOXQ)	116.7%	49.0%	0.19%
State Street SPDR S&P Kensho Final Frontiers ETF (ROKT)	69.8%	40.2%	0.45%
Franklin Intelligent Machines ETF (IQM)	37.1%	32.0%	0.50%
Procure Space ETF (UFO)	54.3%	35.2%	0.75%
Invesco AI and Next Gen Software ETF (IGPT)	83.6%	39.3%	0.56%
Amplify Bloomberg AI Equal Weight ETF (AIVC)	99.5%	44.0%	0.59%

How MarketWise Selected These Funds

MarketWise chose its best tech funds based on the following criteria:

• Funds with exposure to the semiconductor, software, AI industries, and others
• Funds with at least three years of strong returns
• A reasonable expense ratio
• No leveraged or inverse funds

1. VanEck Semiconductor ETF (SMH)

This fund tracks the MVIS U.S. Listed Semiconductor 25 Index, which includes 25 semiconductor production and equipment companies. This fund also makes our list of best semiconductor ETFs.

This VanEck fund is the market’s largest semiconductor-focused ETF, and it charges a fair 0.35% expense ratio. Its returns were notably strong over the past year and three years, and even its five-year returns bested the largest chip-focused funds, too.

Assets under management: $70.5 billion

Top holdings: Nvidia, Taiwan Semiconductor Manufacturing (TSM), Broadcom, Micron Technology (MU), Advanced Micro Devices (AMD)

2. Invesco PHLX Semiconductor ETF (SOXQ)

This passively managed Invesco fund tracks the PHLX Semiconductor Sector Index, which includes the 30 largest semiconductor companies listed on U.S. exchanges. The fund’s low expense ratio of 0.19% means that more money stays in your pocket rather than going to the fund manager.

This chip fund recorded the best one-year returns on our list as well as the second-best annualized returns here over the past three years. It’s a testament to the fact that you don’t need a high-cost fund to achieve attractive total returns.

Assets under management: $2.9 billion

Top holdings: Nvidia, Broadcom, Micron Technology, Applied Materials (AMAT), Marvell Technology (MRVL)

3. State Street SPDR S&P Kensho Final Frontiers ETF (ROKT)

This passive fund tracks the S&P Kensho Final Frontiers Index, which includes companies driving the exploration of space and the deep seas.

The fund has had a tremendous run over the past year and three years, and yet it still charges a not-too-costly expense ratio of 0.45%, putting it among the lower half of expense ratios on the list, despite those stellar returns. This State Street fund also ranks in our list of best space ETFs.

Assets under management: $209.2 million

Top holdings: Oceaneering International (OII), Ducommun (DCO), RTX (RTX), Forum Energy Technologies (FET), Heico (HEI)

4. Franklin Intelligent Machines ETF (IQM)

This Franklin actively managed fund invests in stocks that fit the theme of intelligent machines, where AI is driving the “transformation of products, software, systems and machinery as well as product design, manufacture, logistics, distribution and maintenance….,” according to the prospectus.

Despite being an actively managed fund, it doesn’t charge a steep expense ratio, at just 0.5%. And it has delivered some outstanding annual returns over the past year and three years.

Assets under management: $108.7 million

Top holdings: Nvidia, Taiwan Semiconductor Manufacturing, Celestica (CLS), Broadcom, SanDisk (SNDK)

5. Procure Space ETF (UFO)

This passively managed fund tracks the S-Network Space Index, which includes companies with a majority of their sales coming from space-related sectors, including satellite-based telecoms, space technology and hardware, as well as rocket and satellite manufacturing, among others. The fund’s above-average expense ratio of 0.75% has been worth the extra returns you’re getting, at least so far.

This fund’s portfolio is more tightly focused on telecom names such as Viasat and Sirius XM, as well as rocket companies such as SpaceX (SPCX) and Rocket Lab (RKLB). So, the fund may be a better pick for investors with that kind of investment angle on space.

Assets under management: $609.7 million

Top holdings: Viasat (VSAT), Sirius XM (SIRI), Trimble (TRMB), Garmin (GRMN), AST SpaceMobile (ASTS)

6. Invesco AI and Next Gen Software ETF (IGPT)

This passively managed Invesco ETF tracks the STOXX World AC NexGen Software Development Index, which includes “companies with significant exposure to technologies or products that contribute to future software development through direct revenue,” per the prospectus.

The fund’s one-year performance at more than 80% is nothing short of amazing, while its three-year annualized performance approaches 40% – all for a reasonable expense ratio.

Assets under management: $1.2 billion

Top holdings: Advanced Micro Devices, Alphabet (GOOGL), Nvidia, Meta Platforms (META), Micron Technology

7. Amplify Bloomberg AI Equal Weight ETF (AIVC)

This passively managed Amplify fund tracks the Bloomberg AI Value Chain Equal Weight Total Return Index, which includes cloud computing, semiconductor, and hardware companies that form the basis of AI technologies. The fund uses an equal-weight approach, where stocks have a broadly similar position size in the portfolio.

The fund’s price has effectively doubled over the past year, while its 44% annualized returns over the past three years would have generated some significant wealth. This fund’s returns have been among the best on this list of superstars, and it charges a not-too-high expense ratio.

Assets under management: $120.9 million

Top holdings: Alibaba (BABA), Atlassian (TEAM), Hewlett Packard Enterprise (HPE), Microsoft (MSFT), Snowflake (SNOW)

Are Tech Stocks a Good Long-Term Investment?

Tech stocks have long been among the market’s most popular investments. The best stocks offer scalability – that is, their profits can grow much faster than their sales. As they grow, their profits often soar as their operating margin improves.

Semiconductor stocks have performed very strongly in the past few years, as Big Tech companies, such as Microsoft and Alphabet, have poured money into AI investments. However, much of this high return has occurred in the past year, with two funds in the list up about 100% or more.

This level of performance will be hard-pressed to continue over time, even if semiconductors do post attractive long-term returns. It’s effectively impossible for a stock or fund to put up this level of return over extended periods. So, investors should expect returns to come back to more normal levels, say 20% annually over time. Even this more manageable level will build huge wealth over time.

For investors who like the prospects of the top tech stocks, it’s important to invest cautiously, especially after the phenomenal returns over the past year and three years. Take a long-term view on the market and invest regularly over time. Doing so helps reduce your risk of buying at a relatively high point, and it still can deliver attractive long-term returns.

Regards,

James Royal, PhD

Editor’s Note: One company quietly beat Apple, Amazon, and the S&P 500 — combined. Over the long run it has returned more than 8,300%… enough to turn $10,000 into $830,000. Yet most investors have never heard of it. Former fund manager Whitney Tilson says right now it’s trading at a rare discount — and he’s revealing the name and ticker, free of charge. Click here for the details…

The post 7 Top Tech ETFs to Invest in Semiconductor, AI, Software, and Space Stocks appeared first on MarketWise.

In Tolkien lore, the “palantíri” were indestructible dark crystals used to “see” across vast distances telepathically. The stones themselves weren’t necessarily dangerous. But the person who controlled them could be a major threat… Alex Karp, who named his company after these stones, seems to wield this massive power. And Wall Street is hanging on his every word.

This past Monday, Aug. 3, Palantir Technologies (PLTR) reported an “otherworldly” quarter as revenue accelerated to its fastest pace ever. At the same time, U.S. commercial sales grew 149% year-over-year, and management lifted full-year guidance by roughly half a billion dollars.

The next morning PLTR stock rose about 30%, its largest single-day move in years, and closed back above a technical line it had failed to hold all year. So, is Palantir stock a buy? Karp’s conference call seemed to answer with a simple idea: in the age of interchangeable frontier AI models, the value goes to whoever holds the stone.

And right now, that’s Palantir.

Palantir’s Q2 2026 Earnings, by the Numbers

Palantir posted second-quarter revenue of $1.935 billion, up 93% year-over-year, its highest growth rate as a public company. Adjusted earnings came in at 41 cents per share against consensus near 35 cents, the ninth straight beat. GAAP earnings also reached 41 cents, a sign the profits were real rather than a product of adjustments.

The U.S. business is what carried the quarter. U.S. commercial revenue surged 149% to $764 million. U.S. government revenue grew 90% to $809 million. Combined, U.S. revenue reached $1.573 billion, up 115%, now more than 81% of the total.

The forward-looking metrics have changed how analysts are modeling the company.

Net dollar retention climbed to 157%, up 700 basis points from the prior quarter, meaning existing customers spent 57% more than they did a year earlier. U.S. commercial total contract value bookings hit $2.132 billion, up 153% and nearly $800 million above the prior record. Adjusted operating income was $1.194 billion, a 62% margin. Adjusted free cash flow reached $1.22 billion, a 63% margin. The company’s Rule of 40 score, which adds revenue growth to profit margin, reached 155%, well above the 40 threshold considered elite.

Management raised full-year 2026 revenue guidance to a range of $8.15 billion to $8.158 billion, roughly 82% growth at the midpoint, up from $7.65 billion to $7.66 billion. The company also guided U.S. commercial revenue above $3.424 billion, at least 134% growth. “This quarter was otherworldly,” Karp said in the release.

Palantir’s “SaaSmageddon” and Why Better AI Makes PLTR More Valuable

On the other side of the AI boom, one particular “anti-trend” dominated the discourse for much of 2026: “SaaSmageddon.” The idea is simple: if AI agents can do the same work as certain software, then enterprises will opt for the AI over a per-seat subscription model.

By SaaStr’s measure, software’s forward price-to-earnings (P/E) multiple collapsed from 84.1 times at the 2020-2022 peak to 22.7 times by March 2026, the first time the sector traded below the S&P 500 on that basis. Salesforce (CRM) fell about 30% and Workday (WDAY) fell 33% over that same stretch.

Initially, Palantir itself was part of the SaaSmageddon thesis, with investors worried the company would be eaten alive by frontier AI. But Karp turned that criticism of his company on its head, going to the media to argue that raw model access is the commodity, and value only goes to the layer that connects AI models to proprietary data and workflows.

This layer is what Palantir calls its “Ontology,” which is wrapped in its Artificial Intelligence Platform (AIP) and staffed with engineers who deploy at client sites. Take its partnership with Nvidia (NVDA): Palantir runs Nemotron models air-gapped in customer-controlled environments. And, as Karp notes, more and more enterprises are refusing to become “vassals” of the large-language labs.

Bottom line: The cheaper AI models become, the more valuable the plumbing (the layer that governs and connects AI to enterprise operations) becomes. If you needed evidence of that, look no further than the 157% net dollar retention metric Palantir posted this quarter. If SaaSmageddon were affecting Palantir, customers would not be expanding their usage. This marks the first quarter with hard evidence against that thesis.

Is Palantir Stock Still a Buy After Its 30% Pop?

Palantir’s recent earnings surge repaired some of the technical damage to the stock, but PLTR is still down ~13% as of this writing. Let’s talk about this a bit before getting to the meaty bits.

PLTR stock peaked around $207.52 in November 2025, before collapsing throughout 2026 and bottoming near $106 in June. That’s a ~49% drawdown from Palantir’s highs. Heading into earnings, shares were still close to a 30% shave YTD. Palantir stock formed a death cross when it fell below its 200-day moving average (MA) in February. After earnings, the stock is back above the 200-day MA (around $153). Nevertheless, it’s still below its old high, so there’s more room for PLTR to run.

After the recent quarter, several analysts lifted (or affirmed) their Palantir stock price targets. Citi’s Tyler Radke lifted his target to $245 from $200, Mizuho to $215, and D.A. Davidson to $200, while Bank of America reiterated its Street-high $255 target. Deutsche Bank’s Brad Zelnick upgraded the stock to “Buy” at $200, calling Palantir “a time traveler, having already arrived in the AI future others are still aspiring towards.” Per TipRanks, the consensus is a “Moderate Buy” across 21 analysts, with a pre-surge average target near $187 to $189.

On the road to $200+, Palantir’s nitrous boost, if you will, is the sovereign AI argument. That is, if enterprises and governments increasingly reject frontier AI labs in favor of AI they own and control, then it is Palantir that’s best-positioned to be the operating layer for such demand.

Still, the risks involved with investing in Palantir stock cannot be overstated.

The first big risk is valuation.

Even after the drawdown and even on raised guidance, PLTR trades near 43 to 50 times forward sales, against a software peer group closer to five times enterprise-value-to-revenue. RBC’s Rishi Jaluria kept an “Underperform” rating with a $90 target, writing that PLTR “remains the most expensive stock across our RBC All-SaaS universe at ~40x EV/CY26” and that “we struggle to underwrite a scenario that justifies current valuation levels.” Morningstar’s Mark Giarelli pegs fair value at $153, below where the stock closed after the pop. A move from a 77-times forward P/E down to a still-rich 40 times, with no change in estimates, would roughly halve the stock.

The second big risk is political contagion.

On Aug. 5, the Centre for International Corporate Tax Accountability and Research published an analysis, commissioned by European public-service unions, estimating Palantir’s global effective tax rate at 1.4% on pre-tax profit of about $1.66 billion, with under $21.7 million in global net corporate tax and zero U.S. federal tax in 2025, achieved by booking European revenues in the U.S. parent. Unison’s Andrea Egan said “systems that enable tax to be shirked on an industrial scale clearly have to change.” But Palantir calls this sort of transfer pricing “entirely standard practice.”

That’s not all, however.

The European backlash to Palantir is pulsating and tangible. Germany’s domestic intelligence service, the BfV, picked French firm ChapsVision over Palantir, the German army cut ties entirely, France’s Thales is courting Belgium, the Netherlands and others, and Palantir is suing London’s Mayor’s Office for Policing and Crime (MOPAC) after losing a Metropolitan Police contract. All the while, a £330 million NHS contract faces parliamentary pressure. As such, Palantir’s international revenue slipped to roughly 19% of the total in Q2, from 26% a year earlier. For a company billing itself as selling sovereignty, there’s an irony in being the sovereign no European nation actually wants.

So again, is PLTR stock a buy? Going only off Palantir’s growth, it’s an album with a story told in two sides:

The A-Side: An American growth machine and sovereign AI beneficiary.

The B-Side: A European trust deficit.

The verdict: Palantir delivered the quarter the bulls were waiting for, and the fundamental case is stronger than at any point in its history. But the valuation now prices in flawless execution, and the European resistance caps the global ambition the multiple assumes. PLTR is a buy for investors who can hold through 20%-plus swings and who are buying the operating layer rather than the price. For everyone else, the layers around Palantir are cheaper

3 Palantir Alternatives to Buy

If data and workflow layers increase in value the more frontier models are commoditized, then there are several stocks to buy that own those layers at a fraction of Palantir stock’s multiple. Here they are, in no particular order:

PLTR Alternative No. 1: Snowflake (SNOW)

Snowflake is the proprietary-data layer. Its fiscal Q4 2026 product revenue grew 30% to $1.23 billion, remaining performance obligations jumped 42% to $9.77 billion, and net revenue retention held at 125%. More than 9,100 accounts use its AI weekly, and consumption pricing means revenue scales as AI workloads grow. It trades near 12 to 14 times forward sales, rich for software but well below Palantir. The catch is GAAP losses and a forward P/E near 124. Clean, governed data is the fuel every enterprise AI system needs, and Snowflake stores it.

PLTR Alternative No. 2: ServiceNow (NOW)

ServiceNow is the workflow and agent-orchestration layer, what CEO Bill McDermott calls the “AI control tower.” Q2 2026 subscription revenue grew 24.5% to $3.877 billion, AI annual contract value crossed $1 billion for the first time, and agentic AI production deployments grew ninefold in nine months. The stock fell about 37% in 2026 on the same SaaSmageddon fear, leaving it near 5.8 times enterprise-value-to-revenue, a discount that looks misplaced against accelerating growth. As enterprises deploy fleets of agents, someone has to govern them. ServiceNow is building to be that someone.

PLTR Alternative No. 3: MongoDB (MDB)

MongoDB is the operational data backbone of AI-native applications. Fiscal Q4 2026 revenue grew 27% to $695.1 million, Atlas grew 29%, and the company posted its first GAAP-profitable quarter. Its document model plus integrated vector search and Voyage AI embeddings puts it under the applications developers are building on top of models. It trades far below Palantir’s revenue multiple.

The Bottom Line on Palantir Stock

Palantir’s earnings, much like the palantíri seeing-stones, do not lie. The company is more valuable than ever to enterprises, especially those who wish to free themselves from frontier model vassalage.

Yet, the road ahead for Palantir stockholders isn’t clear of wheel-breaking stones. Both the valuation question and the resolution to the European backlash still encircle the company. For the remainder of 2026, Alex Karp needs to prove that the company selling sovereignty is sovereign in its own right.

Editor’s Note: What ever happened to the AI stock boom? Even AI darlings like Nvidia have essentially gone nowhere since summer 2025. Our friend and colleague at InvestorPlace, Louis Navellier, may have the answer. According to Louis, the AI industry is quietly “staging” ahead of the next great AI breakthrough… a new class of AI he calls “Superintelligence… but better.” How will it trigger a $100 trillion reset of the AI markets. How will the launch of this tech send some stocks to zero, and others soaring? And why does Louis say: Don’t buy or sell an AI stock in 2026 until you see what’s coming next? Go here for the full story (and Louis’ #1 pick).  

Is Palantir Stock a Buy? Your Questions Answered (FAQ)

Is Palantir stock overvalued? By conventional measures, yes. At roughly 43 to 50 times forward sales, it is among the most expensive large-cap software stocks, and bears like RBC see fair value near $90. The counter is that Palantir grows several times faster than peers with 60%-plus margins, which no traditional software comparison captures.

Is it too late to buy Palantir stock? It is too late to buy it cheap. It closed near $162 after the pop, below its $207 high but far above its June low near $106. Investors buying now are paying for years of execution, and the technical reclaim of the 200-day moving average argues the trend has turned.

What are the best alternatives to Palantir? For exposure to the same “own the layer” thesis at lower multiples: Snowflake for data, ServiceNow for workflow and agent governance, and MongoDB for the operational backbone.

Will Palantir stock keep going up in 2026? Analysts lean yes, with a median target near $200, but the range runs from $90 to $255, which tells you how uncertain the answer is. The upside needs continued triple-digit U.S. commercial growth. The downside is a valuation reset or a widening European backlash.

The post Is Palantir Stock a Buy After Earnings? The Bull Case, Plus 3 Alternatives appeared first on MarketWise.