Sam Altmanの兆ドルバブル：なぜOpenAIはAI業界の「リーマン・ブラザーズ」なのか？

Original article by Ed Zitron

Translation｜Odaily Planet Daily Qin Xiaofeng (@QinXiaofeng 888 )

Editor's Note: OpenAI secretly filed for an IPO with the SEC in June 2026 and is gradually approaching an IPO. The current timeline may be pushed back to 2027; if successful, it could become one of the largest IPOs in the AI sector.

Recently, independent tech commentator Ed Zitron published a lengthy critique of OpenAI. He argues that the real AI bubble is essentially the “OpenAI bubble” (OpenAI is the bubble); if OpenAI fails, it will become the “Lehman Brothers” of the AI era, potentially triggering a re-pricing of data centers, AI infrastructure, and tech stocks. He emphasizes that OpenAI's business model is unsustainable, relying on subsidies and circular financing, lacking genuine demand. (Odaily Note: Ed Zitron is known for his direct, financially-focused style of dissecting tech bubbles. This long-form article is the latest and most aggressive in Zitron's series criticizing the AI bubble. He has previously published multiple articles, such as an exclusive report on OpenAI's losses and the Silicon Valley Bubble series, among others.)

The original text is quite long. Compiled and translated by Odaily Planet Daily, Enjoy~

——————————————

Today's article is one of the largest free newsletters I've ever written, consolidating the work of the past six months.

And it all starts with one question: How much do you trust Sam Altman? The stock market, and to some extent the global economy, depends on your answer.

OpenAI has become one of the biggest liabilities in modern economic history. You could argue that OpenAI is no longer the focal point of the AI bubble – you could talk about open-source models or Anthropic or any other element – but without OpenAI, the AI industry wouldn't exist, and the justification for trillions of dollars in capital expenditure would vanish.

The AI bubble does not stem from any actual return on investment – whether it's purely monetary, like revenue or profitability, productivity improvements, or anything tangible or measurable. Instead, it's a cult-like psychotic episode that has infected the minds of some of the most powerful and wealthy individuals and institutions. The powerful mythology of one company has inspired – and is used to inspire – the largest capital misallocation in history.

Although this will annoy some people, I am fully convinced that the only reason this has lasted so long is that OpenAI hasn't collapsed yet. Its failure will be a watershed moment – the Lehman Brothers of the AI bubble, an event that will define the end of one era and the beginning of another, shaking those infected out of their delirium. Without this wake-up call, NVIDIA continues selling GPUs, the semiconductor industry's coffers keep swelling, and more spending commitments are made.

OpenAI plans to burn through over $852 billion by the end of 2030. It accounts for $748 billion of the remaining performance obligations of Microsoft, Amazon, and Oracle, in addition to at least $70 billion in RPO (Recovery Point Objective) from Cerebras, CoreWeave, Nebius, IREN, Lambda, and Nscale (according to Kakashii data), and plans to spend an uncertain tens of billions of dollars on Broadcom's "Jalapeno" chip. It plans to spend $50 billion or more on compute this year, which I estimate exceeds 50% of global AI compute spending (Openai accounts for over 50% of all AI compute infrastructure).

OpenAI can afford these expenditures only because its latest (presumably fully completed) $122 billion funding round has received at least $50 billion, $20 billion of which came from SoftBank (totaling $30 billion, with the third tranche due on October 1, 2026). NVIDIA mentioned in its latest quarterly report that it "estimates that an AI research and deployment company contributed a significant share of [its] revenue for the first quarter of fiscal 2027 by purchasing cloud services from its customers," referring, of course, to OpenAI.

The AI Bubble is the OpenAI Bubble – Headed for an Inevitable End

OpenAI is the reason everyone is watching AI. In March 2019 (according to JustDario), NVIDIA acquired a company called Mellanox, which produces the high-speed networking technology needed to create AI GPU clusters. Four months later, Microsoft invested $1 billion in OpenAI and began purchasing AI GPUs and building AI infrastructure for it. By March 2020, NVIDIA shipped its A100 GPU. In May 2020, Microsoft announced it had built a supercomputer for OpenAI, equipped with "over 285,000 CPU cores [and] 10,000 GPUs."

ChatGPT launched in November 2022, perfect timing for the tech industry – which was then running out of ideas and teetering on the brink of a prolonged downturn. The IPO market had collapsed, rising interest rates ended the zero-interest-rate era, pandemic-era over-hiring was ending with some of the industry's worst layoffs, global venture capital shrank after historic overinvestment in 2021, and tech stocks were hit hard.

For the first time, the tech industry was forced to live within its means – something it has historically been reluctant to do. Big tech was unpopular with investors and the public. The excesses of the past decade – coupled with growing frustration over "tech exceptionalism" (the idea that rules governing the rest of the world don't apply to Silicon Valley) – tested the patience of regulators and legislators. And without "one more thing" – a sensational, game-changing product category – it no longer had an excuse for extravagant spending or regularly breaking the unwritten and written rules that govern society.

OpenAI's existence legitimized an era of frenzy and extravagance. Lacking new hyper-growth ideas, hyperscalers pointed to ChatGPT with its "fastest-growing user base ever" and the Microsoft "supercomputer" that built it, telling investors they'd be left behind if they didn't invest. Amazon, Meta, and Google announced their own vague "supercomputers" in 2023.

By the end of 2023, NVIDIA had sold 500,000 A100 GPUs, and the only reason it could do so was ChatGPT's rapid growth. Sam Altman's brief ouster only inflated the AI bubble further – adding a layer of dreary courtly intrigue to a tech industry lacking imagination and personality – and further cemented Microsoft's role as OpenAI's paternalistic funder, ensuring Altman's return to the helm.

To clarify, by "rapid growth," I mean OpenAI reached 100 million weekly active users by the end of 2023, with monthly revenue of about $108 million. Microsoft invested another $10 billion that year, most of it in the form of Microsoft Azure credits.

OpenAI is also the reason Anthropic exists – not only because several of its founders came from OpenAI, but also because Google and Amazon both agreed to provide it with a total of $6 billion in 2023 as a means to "compete" with Microsoft's new darling, giving both further reason to spend hundreds of billions to "not miss out on AI."

When you take the word "AI" out of the equation, it all seems a bit absurd. $16 billion in equity investments, plus over $150 billion in capital expenditures by the end of 2023, all essentially justified because one website became very popular.

And the only reason these two companies can grow is that hyperscalers fund their entire infrastructure.

In the fourth quarter of 2023, global venture capital fell to its lowest level since the third quarter of 2016, with US startups accounting for $183.6 billion of the year's total investment. Venture capital itself couldn't – and wouldn't – truly support OpenAI or Anthropic at the scale of infrastructure they require. Without hyperscalers inflating the value of these two companies, there would be no hunger from hyperscalers or debt financiers for data centers – which is almost entirely due to OpenAI's success.

If you removed OpenAI from the period between 2020 and 2024, the AI bubble wouldn't have inflated at all. No other major AI company showed any signs of life – not those peddled by hyperscalers, funded by VCs, or launched by other tech companies.

The only reason any hyperscaler's AI efforts generate revenue – and aside from OpenAI and Anthropic, it's quite meager! – is because they know they just have to sit there and keep saying "AI is the future" until customers eventually give in and try it… largely because everyone is talking about ChatGPT.

Anthropic wasn't considered a frontrunner until early 2025, and its ability to continue raising funds was solely because people wanted to invest in the next OpenAI. Anthropic's initial funding rounds and infrastructure building were justified solely on the basis of competing with OpenAI.

The $178.5 billion in US data center debt transactions in 2025? Almost entirely justified by OpenAI's growth and its voracious demand for compute, because besides OpenAI (and later Anthropic), no one else was using massive clusters of tens of thousands of GPUs, and such a market for compute at this scale didn't seem to emerge for months or years afterward.

The largest compute consumers remain Microsoft (for OpenAI), Google (for Anthropic), Amazon (for OpenAI and Anthropic), CoreWeave (for OpenAI and Anthropic), Meta (imitating other hyperscalers' moves), and Oracle (for OpenAI). Otherwise, there's little evidence – and I've looked carefully – that AI compute demand exceeds tens of billions of dollars, which is already generous.

All this investment – whether in AI startups or data centers – exists to fund the next OpenAI, or to become the landlord for the next OpenAI.

The assumption – because no one has seriously thought about it – is that since there is one OpenAI, more OpenAIs will blossom. Since there is one big compute customer, the template for future compute-intensive startups has been set… and, again, because no one seriously thinks about anything, no one realizes that the reason there is no second OpenAI is that OpenAI and Anthropic are financial psy-ops orchestrated by the world's largest software companies.

OpenAI and Anthropic Are Hyper-Scale Psy-Op Specialists for Silicon Valley's Monoculture

The harsh truth is, you can't fund an AI lab through venture capital. While OpenAI and Anthropic have raised nearly $300 billion in the past few years, their actual infrastructure costs – the GPUs and data centers powering their services – are entirely funded by hyperscalers, costing perhaps another $250 billion in the process, as Microsoft stated it had spent $100 billion on its relationship with OpenAI by early 2026.

However, the true cost isn't just financial; it's the experience and industrial knowledge required to execute large-scale infrastructure bailouts. Besides Google, Microsoft, and Amazon, no other company possesses the scale or construction experience for the kind of AI clusters needed by OpenAI (and later Anthropic).

We know this for several reasons. First, because before 2023, few companies were building the scale of AI computing clusters that OpenAI or Anthropic require. The closest might have been cryptocurrency mining companies, and tellingly, many of today's new cloud providers (most notably CoreWeave) originally operated warehouses full of ASIC chips to mine Bitcoin and Ethereum.

Second, because, according to conversations with data center insiders, the entire Overton window of what constitutes a "large" facility has shifted. Previously, a 50MW data center would be considered a significant (even notable) development. These were exceptions, not the rule; most data centers were much smaller. The only companies with experience building at this scale were mostly hyperscalers.

By treating OpenAI as a "venture capital-backed startup," hyperscalers created the illusion that it's the next big kind of company, which will in turn create the next huge demand center for cloud computing – except these companies exist solely because the hyperscalers themselves are enabling their existence, funding them with vast sums of money, allowing them to burn cash at will.

This is why the idea that OpenAI will continue to grow indefinitely is central to the AI bubble myth. The existence of one OpenAI allows others – no matter how illogically – to imagine the existence of more OpenAIs, which in turn means those OpenAIs will need as much compute as OpenAI.

Foolish investors who believe this nonsense can rationalize it through countless buy-side analysts or bought-and-paid-for media pundits – who talk about "insatiable demand for compute," point to capacity constraints (caused by slow data center construction and OpenAI/Anthropic gobbling up most of the world's compute) and rising GPU prices as evidence of massive actual demand, without ever really thinking deeply.

The biggest trick the hyperscalers play is never backing down. By pouring over a trillion dollars into AI capex without showing a single dollar of profit, they justify anyone investing in AI data centers with the logic that "the world's largest companies can't all be wrong" – even though the very reason they're doing it is to expand capacity for OpenAI and Anthropic, which are companies the hyperscalers themselves incubated.

The hyperscalers spending so much money on AI infrastructure is fundamentally illogical and insane, and the reason few are willing to say so is that until recently, suggesting it was a waste of money was considered radical – almost entirely because of OpenAI's existence and continued growth. (Note: I realize Anthropic has gained a lot of attention and grown rapidly in the past year, but it can do this A) because of the OpenAI myth, and B) because it, too, is being incubated and allowed to operate at massive losses.)

Whether you derive utility from LLMs or not is irrelevant because it isn't the actual basis for data center investment. While accelerated progress in code generation (which itself is only possible with massive subsidies) might have helped Anthropic grow, the vast majority of data center capital expenditure is chasing the phantom of what AI might become, not tied to any company's overall revenue or economic reality – except for their compute spending, of course.

This is the underlying greed driving this wasteful, reckless, and destructive era – the belief that there will be another OpenAI, and as I said, a chance to be the landlord for the next OpenAI. And because media and analysts rarely have original thoughts, everyone rationalizes (and continues to rationalize) this waste through the same clichés, saying it's "like Uber (no!)" or "like Amazon Web Services (between 2003 and 2015, Amazon spent $29.7 billion in capex, inflation-adjusted)."

Like any great investment bubble, the more money pours in, the stronger the FOMO, the more dollars can be justified, and the more complex and distorted the myth becomes – which is why you see prominent VCs claim AI labs have "90%+ inference profit margins," a totally unsubstantiated claim that AI adherents cling to and repeat so often that it's taken as gospel, likely because they don't want to face the fact that burning $14,000 worth of tokens on a $200/month ChatGPT subscription is possible.

This myth only grows in an environment deliberately deprived of good information. We've been in this terrible bubble for four years and still have no consensus on the actual cost of large language models, a testament to the industry-wide effort to suppress this information.

OpenAI, Anthropic, Microsoft, Google, and Amazon have gone to great lengths – according to discussions with sources familiar with their infrastructure – to obscure the true underlying costs of their operations, and Silicon Valley, an industry supposedly composed of free thinkers and individuals, is all too happy to accept any convenient myth that might sustain its dreams.

Ultimately, they've all become useful idiots for the hyperscalers. Their obsessive attachment to OpenAI – and by extension Anthropic – looks like a decision made in the name of "democratizing powerful AI," when in reality every dollar ends up with Microsoft, Google, Amazon, or Oracle, which then pass it to NVIDIA or Broadcom, which then pass it up to TSMC, SK Hynix, Samsung, or Micron.

Investing in AI startups? They pay some AI lab, which pays a hyperscaler. Investing in AI infrastructure companies? The money goes to NVIDIA, then upstream to semiconductor companies. Ultimately, whether they die or get acquired (because none will go public), all value ends up in the hands of the hyperscalers who created this phantom era and then inflated it into something very dangerous.

Why the AI Bubble Cannot Survive Without OpenAI

The problem, however, is that this industry cannot survive without OpenAI under any circumstances.

When people discuss the possible collapse of OpenAI, they act purely cowardly, either saying "it won't be that bad" or vaguely claiming it's "too big to fail."

If OpenAI – the company with the most funding, the most infrastructure, the most attention, and the most AI talent – collapses, it will likely happen after AI data center debt and venture capital funds are almost entirely exhausted.

See, Goldman Sachs' Jeffrey Papai recently noted that replicating the hundreds of billions of dollars raised by hyperscalers over the past four years will be "very difficult" – $244 billion in 2026 alone, if you include NVIDIA and SpaceX – which is a problem because, as of Q3 2026, they can no longer fund their data center capex with cash flow.

And to be clear, hyperscaler capex doesn't have to stop entirely for NVIDIA to fall. It just needs to slow down significantly enough that Jensen Huang can no longer deliver investors 60%+ year-over