So it’s been a big week for me after I published an exclusive covering OpenAI’s audited financials from 2024 and 2025, with reactions ranging from “oh my god, OpenAI spent $34 billion to make $13.07 billion in revenue!” to “actually, it’s good the company lost $21 billion.”
And that, my friends, is the Silicon Valley Bubble writ large – an industry that grew rich and famous off the back of a mythical pragmatism and meritocracy that’s morphed into a pseudo-cult built to protect venture capital investments at the cost of reality.
OpenAI lost $21 billion in 2025. $867 million – or around 6.6% – of its revenue came from SoftBank, I posit (though cannot confirm) is something to do with its supposed “Crystal Intelligence” (formerly Cristal Intelligence) initiative with OpenAI, announced in February 2025 as part of the alleged formation of SB OAI Japan, which only actually formed in November 2025.
I severely doubt that SoftBank was a significant cost center for OpenAI given the timeframe, which means that this likely inflated its revenue significantly, and in a way that was disproportionate to its expenses.
There is no justification for this kind of burnrate. There is no sensible, logical or rational reason to look at this company and think that there’s some magical way it’ll become profitable or sustainable. The fact that OpenAI magicked away billions of dollars of costs using bizarre “net losses attributable to noncontrolling members capital” suggests that there are connected entities that the company has yet to disclose, and nothing about this accountancy voodoo changes the fact that OpenAI spent $34 billion to make $13.07 billion. While I cannot speak to its exact intentions, I can see no reason to do this kind of thing outside of trying to obfuscate the horrible state of the company.
Every attempt to rationalize these losses only serves to prove that Silicon Valley itself is a bubble. This is no longer a community concerned with building the future, but building a capitalized consensus, an idea of where money should flow and to whom it should flow to. Gone are the days when plucky software engineers built “bootstrapped” companies that raised rounds based on their theoretical growth, total addressable market, and potential for industry capture. They’ve been replaced by a pseudo-philosophical belief that spending billions on training large language models will somehow turn into a theoretical computer that does its own research, eliminating the need for Silicon Valley to ever have another idea again.
That’s because the Valley has been captured by people that haven’t done any real work in years, haven’t built very much of anything, and thus will fall, well, for just about anything. They don’t see the problem in describing relatively boring cloud software that can write code based on natural language prompts as the path to a sentient computer, or the fact that these companies have mostly sold their software based on fear-mongering.
Per Cal Newport in the New York Times:
When it comes to A.I., we’ve become so used to this tone of helpless, stress-inducing prognostication that we’ve lost sight of its strangeness. Imagine if the Ford Motor Company put out a report saying that it feared its popular F-150 trucks might soon start bursting into flames, but that there was nothing the company could do about it because automotive technology was too inevitable and important to slow down. You’re probably struggling to picture this scenario because no reasonable consumer product company would ever act like this.
The A.I. companies could start behaving the same way. To do so would require that they stop treating A.I. like some inevitable force that they’re struggling to steward. It’s not. It’s a collection of specific tools that these companies are choosing to design and sell according to specific business plans. Accordingly, they need to talk about their offerings like any other consumer product. This means explaining clearly whom these products are for, justifying their benefits and, critically, taking full responsibility for any harm they might cause. Just because A.I. currently enjoys a high-tech sheen doesn’t make it exceptional with respect to common-sense safety standards.
This kind of specious hype and doom trolling exists to make you ignore the current state of AI models in favor of a theoretical better state that you can extrapolate from what you’re being fed by the companies. If you’re scared of AI, you assume that being able to get Claude Code to barf out a copy of some open source software is merely a precursor to automating all software, or even all jobs. If you’re excited about AI, you’re excited because you believe you’re on the ground floor, which will give you incredible advantages when all the things that Dario Amodei and Sam Altman have vaguely promised have come true.
To engage with AI hype is to become its supplicant. You cannot talk in the present tense. You cannot accept any negativity. You must ignore any signs that things are bad and repeat the necessary shibboleths. You must applaud literally any chart or weird, meandering blog that suggests that at some point something good will happen. The Silicon Valley bubble demands you ignore your lying eyes, because if you start thinking about things rationally — as in talking about the stuff LLMs do today and the underlying economics — things become increasingly more-worrying.
In a conversation with Cal on my podcast Better Offline, he also noted that some have tied their pride to their belief in the “incredible” future of AI, interpreting any naysayers as directly attacking their identity rather than critiquing software and the people building it. Perhaps it’s that they swallowed the hype after a particularly vigorous Claude Code session, perhaps it’s that they want to believe that Silicon Valley has “still got it,” but many AI boosters act as if they’re living in the cold, harsh realm of reality as they desperately grasp at straws.
They don’t actually want to hear contrarian points, nor do they want to know about the financials. All they want are more ephemeral talking points to parrot so that they can fool themselves into believing they’ll be rewarded by an industry built on doomerism, fantasy, deception and outright lies. While this existed in different forms in the past — with cryptocurrency, for example — nothing has ever captured the minds and wallets and hearts and social media presence of Silicon Valley more than AI, a technology that can mean anything you need it to, even if it can’t really do anything you’re promising.
The problem is that the world looks to Silicon Valley to explain what the future might be, and when Silicon Valley is captured by people that are either deranged pseudo-philosophers or cynical growth-drunk egoists, very little actual, real value is created. The stock market depends on Silicon Valley to create the next generation of growth — both in the form of new companies and the next chum for the Magnificent Seven to force upon its monopolized customers — but has never let a hype cycle poison its veins this thoroughly or destructively.
Today’s piece — the second (and final) part of the Silicon Valley Bubble series — is focused on how Silicon Valley’s reality distortion field has escaped containment, exploiting intellectual weaknesses throughout organizations and economies by promising a near-infinite source of capital.
The AI bubble has grown by promising everybody something — a cure for a tech industry that’s run out of hypergrowth ideas, a way for public (and private) companies to promise infinite growth, a way to paper over the collapse of growth throughout the software industry, and a way to convince the general public that the tech industry is an infinite flywheel of ideas rather than a machine custom-tweaked to extract capital through monopolies.
The problem isn’t simply that it will eventually need to make good on those promises, but rather, what those promises do in-and-of-themselves. Like a caustic acid, they’ve deformed and reshaped so much of what we consider to be the tech industry, changing incentives and eliminating what was once considered the guardrails against the kinds of reckless exuberance we’re now seeing.
Coming Up On This Week’s Where’s Your Ed At Premium
- The AI Media Bubble — the greatest mindshare exploitation of all time.
- The CFO Bubble — how the tech industry turned the adults in the room into co-conspirators in a financial con
- The Greater Software and SaaSpocalypse bubble — how AI is an attempt to paper over the collapse of the overall software industry.
- The GPU and AI Infrastructure Bubble — how the AI industry has helped set up a horrifying collapse that will have horrible micro and macro-economic consequences