The AI Bubble Bust: Is the AI Monster about to Fall on us?

I’ve called the whole AI stock mania a giant Ponzi scheme because it depends on endless new rounds of bond investors at higher and higher interest rates to keep the cycle going because its actual profits off of AI services, in most cases, make these ventures look a lot more like towering zombie corporations that can’t even cover operations without endless new rounds of credit.

(You must exclude the entire portion of their reported profits/earnings that are from gains on their stock investments within the AI industry because those won’t shore any business up if AI stocks are the thing that is falling! They’ll turn overnight from profits to losses. It will be the end of stock “buybacks,” except that they are buybacks within the industry in each other’s stocks that are keeping overall AI stock prices rising so long as they all keep doing it. It’s an incestuous cycle.)

David Trainer, CEO of financial research firm New Constructs, is calling the top dog, Anthropic—even with its 1,300% sales growth, 900% net revenue retention and 85% implied gross margin—“a huge Wall Street grift.”

I’d say he agrees with me.

At the center of the debate over whether the AI bubble is about to bust or boom is Anthropic’s just-released prospectus for its highly anticipated initial public offering.

Besides supposedly having a lot of cash and all those listed plus sides, it’s IPO prospectus shows that it also has …

roughly $80 billion in half-year net losses, hundreds of billions of dollars’ worth of compute commitments and more than 90 pages of risk factors.

And most of those compute commitments are cash-out agreements where Anthropic has to pay the cash even if they never uses the “compute” it has committed to, such as if the company can’t complete its build outs. That is typical of the other AI companies as well.

From either the bull point of view or the bear point of view, the numbers one can latch onto are staggering as Anthropic flirts with accomplishing the largest IPO in history.

Bull or bear?

On the bull side, you hear such arguments as,

“This is a generational moment, and we’ll look back in a decade, and this was the biggest paradigm shift in technology that will shape our lives like nothing else has,” Denny Fish, a portfolio manager at Janus Henderson Investors, told CNBC’s “The Exchange” on Monday….

In its long-awaited prospectus on Monday, Anthropic said its enterprise customer base included more than 11,900 customers with at least $100,000 spend over the prior year, more than 1,900 clients spending $1 million or more and roughly 250 customers spending $10 million or more.

“I think it’s really easy to lose sight of how early it is, and to see a business like Anthropic doing this well this early on in the journey is incredibly impressive,” Erica Brescia, a managing director at Redpoint Ventures, told CNBC’s “The Exchange” on Monday.

On the bear side,

The company lost $80.7 billion in the first six months of this year, after reporting a loss of $12.9 billion over the same stretch in 2025, though most of that was due to the change in fair value of convertible notes and derivative liabilities. Meanwhile, its compute commitments are soaring while competition is picking up from lower-cost open-weight models.

Also, truly odd for a prospectus that intends to sell people on buying the new stock at a fortune of a price but that is legally obligated to lay out the liability risks, the prospectus spends 90 pages telling all potential shareholders how the company’s AI may make Anthropic liable for destroying the entire world or destroying humanity. That is a little bigger liability than you’ll see any non-AI IPO prospectus. It may, in fact, be the only IPO prospectus to come right out and emphasize across many pages that the company might be liable for the entire destruction of mankind. (But then who would be left to sue or left to care?)

Trainer has built a reputation on Wall Street as a notorious bear on IPOs. His firm published a report last month arguing that Anthropic is the “most ridiculous IPO of 2026” and that it will be an “unprecedented test of investor gullibility.”

Having reviewed the prospectus, Trainer told CNBC in an interview on Tuesday that his position hasn’t changed. He said AI will be an “amazing tool for society,” but not so much for businesses, and he’s worried about Anthropic’s competitive position, its “opaque” insider selling schedule, and the fact that its commitments and liabilities dwarf its cash.

The cash is where my claim of the Ponzi scheme comes in. Economist Wolf Richter recently pointed out that many of these AI companies are done for now with needing to issue bonds to raise their huge hoards of necessary cash because they spent the past year doing all of that and now are sitting on huge piles of cash already raised. So, my claim for the Ponzi crash could be premature if they have enough to sit solid for months ahead without needing to raise another round of sucker bond investors to keep the Ponzi cycle going at a time when bond yields are soaring and AI profits still rarely look like a reasonable ROI.

There is a catch, however. Many of these companies have already made massive cash commitments for their buildouts that require full cash-outs, even if they never complete construction or produce any significant services. They still must pay. Many have committed more cash than they have piled. So, the cash piles may buy them some more time initially to avoid the AI bubble bust, BUT those cash piles can be eroded away rapidly, too, if things do go bust.

“All these people have stakes in Anthropic. They want to be able to cash out,” Trainer said. “Having lived through the tech bubble, having lived through the great financial crisis, having been on Wall Street during those times, having seen how the sausage is made, it’s like a slow-moving train wreck for me.”

That would be like this:

As I wrote in my article about the Everything Bubble Bust, don’t expect the AI Bubble Bust to happen quickly. Remember that I have pointed out a number of times over past months that this bubble will break like the dot-com bust, and that bubble took 2.5 years to fully plumb its bottom, 3 years to really show any signs of starting to rise off its bottom, and it had some big rallies along the way down, and took ten years to recover.

The companies that have already raised huge piles of cash through bond offerings will likely be the companies that go down in the final year of that train wreck but remain alive. Those who have not already raised all of the cash they need and especially those who must soon go back to the bond market, may likely be the first. From there the long slide with all of its legal battles over who gets paid out begins.

Roger McNamee, an early investor in both Google and Facebook, told CNBC on Tuesday that the Anthropic IPO is a “very dangerous thing for investors,” adding that the industry still has a lot to prove.

“We should view this S-1 as speculative fiction because it’s a great story, it’s a really great story, but it’s hundreds of pages, and those hundreds of pages are not there to clarify the story,” McNamee, co-founder of Elevation Partners, told CNBC’s “Squawk on the Street.” “And based on the feedback I get on [X], a lot of people who disagree with me are sitting there going, ‘I love this story,’ and I go, ‘Look, God bless, it’s your money.’”

The possible road to ruin

Yup. It’s your money, and along the way to just the IPO, many steps remain where things can go wrong:

On its roadshow, the company will have to woo prospective investors, gauge demand, set the IPO offer price and allocate shares. Once all of that is complete, it will finally list on the Nasdaq under the ticker symbol “ANT.”

And yet …

“We can talk about the risks, but as we sit here right now, and based on what we know right now, Anthropic is still the main AI company, and it’s a pure play,” Luria said. “So the IPO should go well because investors will want exposure to the pure play.”

Is it a pure play or a bleeding-edge company that will set the new paradigm for the destruction of humanity as its prospectus lays out for the possible risks. How fast will the cash get consumed if along the way to the IPO or even after the IPO, its AI gets out of the sandbox and destroys global monetary systems? If Anthropic didn’t think that was a real risk, they wouldn’t spend so many pages on that kind of thing, and banks wouldn’t be rushing to close the risk like its Y2K all over again for them!

What if data-center heat in small communities where its build-outs are supposed to happen keeps rising?

The public’s souring position on data centers poses a major risk for Anthropic, which relies on an ever-expanding portfolio of state-of-the-art computing facilities to train and maintain its prized foundation models….

“Opposition to data center development has also become more organized and coordinated across jurisdictions and our public profile may draw particular attention to projects associated with us,” the filing said.

Lawmakers in 16 states are already considering moratoriums. Remember that, even if the buildout never happens because the public doors slam close, the PAY OUT must still happen to all those who are in bed with Anthropic; and this is typical of the other AI developers.

And then what about those massive power-grid shortfalls during a time of an even more massive war on energy:

Anthropic’s future success depends not only on building giant data centers, but also on the electrical and related power infrastructure needed to juice those computing facilities.

The lack of “sufficient electrical power” across the nation and world poses big risks to Anthropic and its rivals, the filing notes. Even if energy becomes more readily available through “on-site natural gas-fired generation,” that power could become more expensive with so much demand, the filing said.

There are many ways this train can wreck, and it’s complicated because there are many huge companies to go down, all with their own different exposures, including to each other. So, again, don’t expect a quick trip to the bottom.

In fact, the AI bubble may bust just because the stock market in general goes down. Anthropic wouldn’t be the first IPO to get flushed entirely away because the stock market chose to generally crash just as the IPO hit launch date and an outgoing tide lowered all boats.

Anthropic needs to keep racing and needs the rest of the business world to keep up. If companies curb their AI spending, Anthropic could feel the burn.

That is the Ponzi-scheme aspect where endless cycles of high-momentum investment and equally rapid sales growth (to justify that growing mountain of capital investment) need to keep happening to keep the thing from falling as it likely will if the momentum stops. It is not an actual Ponzi scheme, of course, as in something that is a planned scam from the beginning, but it suffers from the same risk of needing seemingly endless new capital pouring in to maintain the development long enough to hit operational profitability FROM AI ACTIVITY large enough to make a sensible return on investment without relying on that capital fundraising to pay out demands for a return on investment or cash-out demands on those commitments. Again, profitability from stock purchases in its business partners doesn’t help if stocks start falling. That is just a double loop on the Ponzi side.

What if the rest of the business world doesn’t keep up?

So, what happens to Anthropic and the whole AI bubble if the rest of the world doesn’t keep up and other stocks keep going down as has been somewhat the case in today’s market where breadth as been, at times, looking like only a few high-tech stocks were holding the averages positive.

With that, let’s take a quick look under the hood of the overall market since a few seriously critical indicators were brought to light in the news:

The pace at which investors are borrowing money to invest in stocks has cooled substantially, raising concern that the stock market may be due for a fall.

This is a transition that is totally typical of major stock-market busts: You can see in the following graph what happens when the rate of growth in money borrowing for stock investing hits a high peak and then starts to fall. Look at how suddenly the growth rate of borrowing falls downhill and how the stock market falls right after it, especially whenever the peak is at or above present levels and falls by the amount it just has...