14 September 2026
Heard In AI

Ed Zitron's 2027 forecast: OpenAI runs out of cash, and the losses spread

On The Diary of a CEO, critic Ed Zitron laid out a sequence he expects to start with OpenAI failing to raise its next round and end in ordinary retirement accounts. He traces the chain from a delayed stock-market listing to SoftBank's paper holdings, cloud growth forecasts and the concentrated US indexes — while Amazon's own filings and Andy Jassy's shareholder letter offer a different account of why the spending is happening.

A briefing reports one development when it happens. We correct or clarify it later; a new development gets a new briefing. How our formats work

Asked what happens when the AI bubble pops, Ed Zitron did not start with chatbots or chips. He started with a calendar. OpenAI, he said, had been expected to join the stock market this year; the listing has been pushed to next year, and he summarised what the company's chief financial officer, Sarah Friar, had since offered as timing: they would do it earlier than 2027, or in 2027. His verdict on that answer: "Great answer there."

Zitron writes the newsletter Where's Your Ed At and argues that the AI boom is financially hollow. Speaking on The Diary of a CEO, he set out a forecast with a specific shape: sometime in 2027, he thinks, "things are going to start running out of steam," beginning with OpenAI running out of cash. Everything after that is his projection of what follows, not a documented outcome.

Why the listing matters so much

Going public, as the host explained for listeners, means joining the stock market — the moment when investors who backed a private company can finally sell the equity they bought. Zitron's point is that the prospect of that moment is itself a fundraising tool: companies flirt with going public because investors need to picture getting their money back.

OpenAI, he argues, cannot stop raising. He said the company raised $122 billion this year, that most of it is already spent, and that it will need "at least $100 billion a year just to survive." OpenAI's own February 27, 2026 announcement describes part of that: $110 billion in new investment at a $730 billion pre-money valuation, with $30 billion each from SoftBank and NVIDIA and $50 billion from Amazon, alongside commercial infrastructure deals. Zitron put the company's last round at an $865 billion valuation.

The trap, in his telling, is arithmetic about that number. He said OpenAI wanted to list at a $1 trillion valuation and that advisers, according to reporting by Mike Isaac of the New York Times, warned against it. If the company cannot go public, it has to raise privately again — and Zitron thinks it would struggle to match its last valuation, leaving a "flat" round. For a company that has to keep telling investors the number goes up, he said, "they can't do that."

He pointed to one deal as evidence of how badly the money is needed: Amazon, he said, sent $35 billion that was meant to be contingent on OpenAI going public early. Amazon's filing for the quarter ended June 30, 2026 records $28.7 billion invested in OpenAI over the first half, including $13.7 billion in the second quarter, with the remaining $21.3 billion of the commitment funded after the quarter closed. The same filing records $10 billion invested in Anthropic nonvoting preferred stock in the quarter.

And Zitron expects a competitor to reach the exit first. Anthropic, he said, is likely to go public before OpenAI, which would make OpenAI's own listing "borderline impossible" — because Anthropic, in his description, is another unprofitable, unsustainable AI lab, "but a better business that's growing faster." He does not exempt it from the same ending: "They're eventually going to face perdition too."

The banana tree

Underneath the forecast sits Zitron's older argument. In his 2023 essay The Rot Economy, he described an investment culture that rewards expanding revenue, valuation and market share above sustainable businesses or useful products. On the podcast he extended it into what he calls the Rotcom bubble: the big platforms, he says, are out of hyper-growth ideas, "but the market doesn't think so."

His sequence runs from record 2021 earnings and cheap money into a hangover, when growth slowed. Then, he said, "suddenly they started buying GPUs" — and markets read the purchase as proof of a new business. Stocks ran up hundreds of percent, he argued, on expenditures rather than disclosed AI revenues: "everyone just gave them credit in advance."

The host pushed back with an image. Imagine a desert island with 10,000 people on it, he said, and someone announces they have found a banana tree. People will stampede toward it and claw each other to pieces — irrational decisions, but made by hungry people around something real. He thinks the underlying technology has more long-term promise than Zitron does.

Zitron took the analogy as confirmation rather than rebuttal. "That is the Rotcom bubble," he said. "They don't have a new thing and they're desperate." His counter to the promise argument was about fuel rather than potential: keeping the labs improving requires tens of billions a year on training, which requires the cloud companies, venture firms, private credit lenders and NVIDIA to keep circulating money toward them. "Once the money stops flowing, the progress stops."

He also complains that the companies will not show their work. Across the large platforms, he said, AI revenues are not disclosed; when a figure appears it is an "annualized run rate" that is never defined and can mean a month multiplied by twelve, or thirteen, or the last four weeks multiplied by thirteen. "When they have good news, they'll tell you," he said. "And when they don't tell you something, well, that actually speaks volumes."

How he thinks the losses travel

The contagion argument depends on who is tied to OpenAI. SoftBank, one of the largest companies on the Japanese stock market, holds roughly $100 billion of OpenAI stock on paper, by Zitron's count, and stays solvent by liquidating holdings — selling shares or borrowing against them. Without a listing, he said, it cannot do that, and while SoftBank probably would not run out of money, "we're going to see one of the largest holding companies in the world become much smaller."

Next come the suppliers. Zitron said sell-side analysts expect OpenAI and Anthropic to spend more than $400 billion with Microsoft, Google and Amazon over the next three and a half years — around 30% of projected cloud growth from two companies that must first be given the money. If that spending does not happen, he expects Amazon, Google and Microsoft to restate guidance: "we don't think we're going to grow as fast."

He gave Oracle as the sharpest case, saying it is building 7.1 gigawatts of data centers, "over $400 billion worth just for OpenAI," for a company whose revenue has been flat for fifteen years adjusted for inflation. Those figures come from overlapping announcements that describe different things: OpenAI's September 2025 Stargate update described nearly 7 GW of planned capacity and more than $400 billion of investment over three years across the whole Stargate program, and separately valued the Oracle partnership at more than $300 billion over five years. Zitron's conclusion rests on the dependency rather than the sum: "Without OpenAI, Oracle dies."

Then the indexes. NVIDIA, he said, is the largest company on the NASDAQ and accounts for 7% to 8% of the S&P 500, while doing "the most circular of financing, feeding companies money so that they can raise debt to buy more GPUs." He put a possible swing of 50% to 70% on NVIDIA's revenue, recalling that in 2022 the company was making single-digit billions. Asked about ordinary viewers with ordinary jobs, he described retirements contracting severely and not recovering: "20, 30, 40% off the top of these companies' stock value." So much of the S&P 500 and Russell 1000, he said, comes from those companies and the rest of the Magnificent Seven. He expects venture capital to take heavy losses too, arguing that since 2018 investors have received roughly 80 cents to $1.20 back per dollar committed, that more than half of last year's venture money went into AI, and that "most venture capital investments in AI are going to zero."

Paper gains are where his complaint gets specific. Venture firms, he said, are celebrating and raising off valuations rather than returns: "the valuation of Anthropic went up." He said Google boosted reported profits by $99 billion in a quarter through increases in the value of its SpaceX and Anthropic holdings. Amazon's quarterly filing shows the same mechanism on its own books: $53.4 billion of second-quarter other income, primarily from upward adjustments to Anthropic preferred-stock valuations, recorded outside operating income.

The executives' answer

The host had come prepared with the other side, reading out what the chief executives say. Sundar Pichai, chief executive of Google: "the risk of underinvesting is dramatically greater than the risk of overinvesting." Mark Zuckerberg of Meta: "I'd rather risk building capacity before it's needed than being late." And Amazon's Andy Jassy: "we're not investing approximately 200 billion in CapEx in 2026 on a hunch... we're investing to be the meaningful leader, and our future business operating income and free cash flow will be much larger because of this investment."

The reply was a line from Shrek: "Some of you may die, but that's a risk I'm willing to accept."

Jassy's 2025 shareholder letter sets out the reasoning behind the quote. He argues the roughly $200 billion of planned 2026 capital expenditure is backed by customer commitments, including an OpenAI commitment exceeding $100 billion, and that infrastructure spending typically precedes billing by six to twenty-four months, so much of 2026's AWS investment would be monetized in 2027 and 2028. The case assumes long-lived assets — over thirty years for data centers, five to six years for chips, servers and networking equipment — and presents near-term free-cash-flow pressure as the expected cost of building ahead of revenue.

Amazon's filing shows what that looks like in the accounts. For the twelve months to June 30, 2026, the company reported $161.403 billion of operating cash flow against $169.007 billion of property and equipment purchases net of sales and incentives, producing negative $7.604 billion of free cash flow under its own definition. A business generating substantial cash from operations can still show free cash flow below zero while it is building. Whether that is investment ahead of demand or spending ahead of a story is the question the two men could not settle.

Zitron's own version of the ending is a re-rating rather than only a crash: markets valuing these companies "like airlines" — big, profitable on existing products, and without a new thing. Asked what would make him wrong, he did not reach for a hedge about timing. He named conditions: "A bunch of hardware breakthroughs to make this profitable." And then, "New mathematics." He added a complaint about who gets asked such questions — the man with a blog, he said, rather than "the people spending a trillion dollars."

Share this article

Go to the original

Sources & further reading

  1. 01
  2. 02
  3. 03
  4. 04
  5. 05

Connected ideas and articles

From the conversation

Podcast episodes

Article history

Updates to this article

Tags

NVIDIA's open-model push is about GPU demand, the Moonshots panel says

On Moonshots EP #283, Peter Diamandis introduced a reported $6 billion NVIDIA arrangement with the coding startup Poolside as America's answer to Chinese open models. Emad Mostaque argued the real driver is selling more GPUs, while Alex and Dave disagreed about whether licensing-and-hiring deals exist to dodge antitrust review or simply to hire fast — and what happens to the half of Poolside that stays behind.

7 min read

Would she pay the real price? Zitron's test for AI adoption

On The Diary of a CEO, critic Ed Zitron praises a chatbot for reading a troubleshooting log and for helping fix his son's Minecraft mod, then argues that neither is worth a trillion dollars. The host counters with his fiancée's one-woman business and his chief of staff's inbox. The argument turns on tokens, subscription rate limits and who is paying the real bill.

7 min read

NVIDIA’s $500 billion financing plan faces the problem of aging GPUs

NVIDIA has signed memorandums with six financial institutions aiming to mobilize more than $500 billion in outside capital for customers’ AI infrastructure. On Moonshots, the panel debated whether rapidly changing chips can support long-term investments: Salim Ismail warned of stranded assets, Alex argued for financial hedges, and Emad Mostaque explained why older, paid-off GPUs can keep earning.

5 min read

Altman says economic inertia slowed AI's impact. His podcast panel disputes the cause

On Moonshots with Peter Diamandis, the panel watched Sam Altman explain that he expected GPT-4 to put software businesses up for grabs far sooner than it did, and that the economy's inertia has made the transition "smoother and slower." Salim Ismail blamed institutions that move at a different speed from the technology, Alex pointed instead at the abstraction layers of the economy and prescribed vertical integration, and Emad Mostaque objected that the models simply were not good enough until recently.

6 min read