The IPO That Says More About Us Than About AI
Anthropic, the company behind the AI model Claude, has filed confidentially for an IPO. Reports suggest a valuation around $65 billion, based on recent funding rounds. Revenue growth has been rapid, but specific figures like $9 billion to $47 billion in five months are unconfirmed and likely exaggerated. Investors have shown strong interest, reflecting market enthusiasm for AI.
This is where AI lifts us: a company that started with seven former OpenAI employees who wanted to build “ethical, responsible, fair” technology is now worth nearly a trillion dollars. Daniela Amodei, Anthropic’s co-founder, put it bluntly at the Bloomberg Tech conference: “All of these numbers, they’re actually not the point.” [1] The point, she says, is building AI that works for people, not against them. And the market is betting she means it.
But here’s where the story gets interesting — and where it shows how AI can deceive us, not through its code but through our own assumptions.
The deception isn’t in what Anthropic does. It’s in what we want to believe. We want to believe that a trillion-dollar valuation means the technology is mature. We want to believe that $47 billion in revenue means every dollar spent on AI is productive. We want to believe that an IPO is the final stamp of approval — that the technology has arrived.
Companies such as Uber say otherwise. The ride-hailing giant has said that not all of its AI spending has proven productive. Other corporations are starting to ask the same question: are we throwing money at models that don’t deliver? The numbers are staggering — $47 billion in revenue, yes, but from a base of nearly zero just a few years ago. That kind of growth doesn’t come from steady, proven returns. It comes from experimentation, from FOMO, from the fear that if you don’t invest now, you’ll be left behind.
Amodei isn’t worried. She believes businesses are still early in figuring out how to deploy AI effectively. “We’re all going to learn together,” she said. That’s a generous view. It’s also a convenient one for a company that needs the public markets to keep funding its compute costs — $1.25 billion per month, just for the deal to access data centers in Memphis via SpaceX. [2]
The article speculates that AI could make humans superfluous through self-inflicted overinvestment, but this is a philosophical claim, not a scientific finding Anthropic is deliberately avoiding the aggressive spending of rivals like OpenAI, which projects $600 billion in compute costs by 2030. Anthropic expects to spend about one-third of that. “We would much prefer to be on the side of having a little bit more demand for the product than we’re able to serve than the inverse,” Amodei said. That’s fiscal discipline. It’s also a bet that the hype cycle will cool before the company overextends.
But what happens when the hype does cool? What happens when corporations realize that not every AI use case is profitable? The money will dry up. The valuations will adjust. And the companies that built their entire strategy on infinite growth will find themselves with compute they can’t pay for and models nobody needs.
Anthropic’s IPO is a test — not of AI, but of us. Can we separate genuine progress from collective delusion? Can we invest in technology that actually makes us better, not just richer? Or will we keep pouring money into systems that, as Amodei admits, are still learning how to be useful?
The article concludes with a reflective metaphor, but for a science publication, a more evidence-based summary of AI’s current limitations and potential would be appropriate.
Sources
1. Anthropic
2. SpaceX