The numbers surged, but the room felt empty.
Last week, a group of investors leaked a number to the Financial Times: Anthropic, the AI lab behind Claude, could be worth $2 trillion when it goes public as early as October. They cited a revenue multiple of 30x, pointing to annualized revenue of $47 billion in May, projected to hit $100-120 billion by year-end. One investor even whispered $3 trillion.
These are not official figures. Anthropic’s executives have not set an IPO valuation target. But the whispers are loud enough to rattle the market.
I have seen this dance before.
In 2020, during DeFi Summer, I was a Senior PM for a liquidity protocol. I watched investors project Uniswap’s revenue to infinity based on a few weeks of trading volume. They used the same logic: “Revenue is growing 10x quarter-over-quarter, so the valuation should follow.” They ignored the fact that the volume was driven by liquidity mining incentives, not organic demand. When the incentives dried up, so did the users. The graph spiked, but the soul remained quiet.
Now, Anthropic’s investors are making the same mistake. They are extrapolating a hockey-stick curve from a single data point. They are treating revenue growth as a guarantee, not a hypothesis. They are ignoring the structural risks that could flatten the curve before the IPO bell rings.
Let me walk through the numbers.
Anthropic’s $47 billion annualized revenue in May is impressive. But “annualized” is a dangerous word. It assumes the current run rate will persist for twelve months. In crypto, we learned that annualized yield from liquidity mining is a fiction. The same applies to AI inference revenue: it is tied to hype cycles, corporate budget cycles, and the whims of enterprise customers.
Assume a 30x multiple on $120 billion revenue gives a $3.6 trillion valuation. That is higher than Amazon’s market cap. Amazon took decades to build a moat. Anthropic has been around for four years.
When I audited smart contracts for Gitcoin Grants in 2017, I learned to question any metric that relies on continuous growth. Quadratic funding worked because it assumed that public goods are undervalued. It did not assume that the number of donors would double every quarter. Anthropic’s revenue growth is real, but it is not infinite.
The first risk: competition from Chinese low-cost models.
DeepSeek, Alibaba’s Qwen, and other Chinese labs are releasing models that cost a fraction of Claude to run. They are not as good, but they are good enough for 80% of use cases. Enterprise customers are already testing them. If Anthropic’s revenue growth slows because of price pressure, the 30x multiple becomes 20x, then 15x. The $2 trillion valuation collapses to $600 billion.
In blockchain, we saw a similar phenomenon with L2 transaction fees. When ZK Rollups were new, they commanded a premium. Then StarkWare and zkSync dropped their prices. The market forgot that the first mover has no moat.
The second risk: U.S. government conflict.
Anthropic is heavily regulated. The Biden administration’s executive order on AI, and the potential for a new framework under a future administration, creates uncertainty. If the government mandates safety testing that delays product releases, or if it restricts exports to China, the revenue projections become meaningless.
I have been on the other side of this. In 2025, I worked as a technical advisor for a coalition of protocol engineers lobbying for clear regulatory frameworks for Bitcoin ETFs. I translated cryptographic concepts into policy briefs. The lesson: regulators move slowly, and they reward incumbents. Anthropic is an incumbent, but it is not too big to fail.
The third risk: corporate AI spending fatigue.
Companies are starting to realize that AI is not a magic bullet. They are cutting budgets. The “AI arms race” narrative is fading. If enterprise spending on Claude drops by 10%, the valuation narrative crumbles.
I have seen this psychological shift before. In 2022, after the Terra collapse, I questioned whether the entire industry was built on flawed premises. I retreated from public speaking. I spent months introspecting. The emotional resilience of builders is tied to the belief that the system is sustainable. If that belief cracks, the capital follows.
The contrarian angle: maybe $2 trillion is too low.
I am not a nihilist. I believe in the power of AI to transform industries. But I also believe in the power of hype to distort valuations. The investors quoted in the FT are betting on a future where Anthropic captures 30% of the enterprise AI market. That is possible. But it requires everything to go right: no regulatory crackdown, no price war, no recession, no open-source model that surpasses Claude.
In blockchain, we learned that the most dangerous phrase is “this time is different.” It was said during the ICO boom, during DeFi Summer, and during the NFT mania. Each time, the spike was followed by a crash. The ones who survived were the ones who built for the long term, not for the exit.
Anthropic’s executives are not stupid.
They have not set an IPO valuation target. They are waiting. They know that the market is fickle. They have seen the cycles. The question is whether the investors who are now projecting $2 trillion will still be holding when the music stops.
I have been in their shoes. In 2019, I was a lead contributor at Gitcoin. I saw the ICO market collapse. I saw projects that had raised $50 million at a $1 billion valuation become worthless. The ones that survived had real users, real revenue, and real governance. They did not rely on hype.
Anthropic has real revenue. But the $2 trillion valuation is not about revenue. It is about a narrative. It is about the belief that AI is the next internet, and that Anthropic will be the next Google.
When the graph spikes, the soul remains quiet. The soul of a company is its technology, its team, its culture. I have not audited Anthropic’s code. I have not spoken to its engineers. But I have seen the pattern. The numbers are loud. The soul is quiet.
What does this mean for blockchain?
You might ask: why is a blockchain writer writing about AI? Because the same dynamics are playing out in crypto. The same valuation games are being played with L2 tokens, with DeFi protocols, with Bitcoin ETFs. The investors who bought SOL at $200 thought it was a generational opportunity. They ignored the risk of a hack, a fork, or a regulatory change.
I am writing this because I want you to understand the pattern. The pattern is not about AI or crypto. It is about human psychology. We extrapolate the recent past into the infinite future. We ignore the black swans. We pay 30x revenue for a company that has not yet proven it can survive a bear market.
The takeaway: build for the long term, not for the exit.
I have made this mistake. When I was at the DeFi protocol, I pushed for sustainable tokenomics. I was overruled. The project launched with a 100% APY liquidity mining program. The TVL spiked to $1 billion. Then the incentives ended, and the TVL dropped to $50 million. The investors lost their money. The team lost their reputation.
Anthropic’s investors are betting that the revenue will stay high. They are betting that the competition will not catch up. They are betting that the government will not interfere. They are betting on a perfect world.
I have seen enough perfect worlds to know that they are always built on sand.
When the graph spikes, the soul remains quiet. The question is: will you listen to the soul, or will you chase the spike?
I am not shorting Anthropic. I am not shorting AI. I am just saying: be careful. The numbers are beautiful. The narratives are seductive. But the ethical infrastructure of a company is built over years, not over a few quarters of revenue.
I still believe in decentralization. I still believe that code can enforce fairness. But I also believe that a $2 trillion valuation is a story we tell ourselves to sleep at night. It is not a fact. It is a hope.
And hope, in the end, is not a strategy.