Hook: The $65 Billion Question
We don’t often see a number that stops the conversation mid-sentence. But when I heard that Anthropic’s annualized revenue run rate hit $65 billion at the end of July — a staggering $25 billion ahead of OpenAI — I felt the familiar pull of a deeper truth. The bear market didn’t kill innovation; it just redirected capital. And now, the AI industry is sprinting so fast that its own shadow cannot keep up.
Yet as I sat in a Nairobi co-working space, refreshing Bloomberg terminals and reading the confidential prospectus filings, I couldn’t shake the feeling that these numbers, as impressive as they are, are built on a foundation of sand. Because who audits the models? Who verifies the data? The same question that haunted me during the 2017 DAO hack — code is law, but flawed by human hubris — now applies to the outputs of large language models. Anthropic’s run rate tells us nothing about the integrity of the outputs themselves.
Context: The Numbers Behind the Hype
Let me paint the landscape. According to figures shared by people familiar with the matter, Anthropic crossed roughly $9 billion in annualized revenue run rate at the end of 2025. By May 2026, that number had jumped to $47 billion. Then, in just two months — from May to July — it added another $18 billion, reaching $65 billion. That’s a 622% expansion across seven months. For context, OpenAI is on track for a run rate above $40 billion, roughly double its end-of-2025 level. The preliminary second-quarter revenue for Anthropic topped $11.5 billion, against $787 million in the same quarter a year earlier. Quarterly revenue more than doubled from $4.73 billion in the first quarter. The company also posted positive adjusted operating income for the period — a rare feat in the capital-intensive AI world.
These numbers come from a confidential prospectus filed with the SEC in June, and investor meetings are already underway. Bloomberg expects a Wall Street debut as soon as this fall, with the Financial Times reporting that investors anticipate a valuation of $2 trillion. That’s not a company — that’s a small country.
But here’s the rub: neither Anthropic nor OpenAI officially confirmed these run rates. The numbers trace to “people familiar with the matter,” and the two firms may not calculate the metric the same way. In crypto, we call this “selective transparency” — and it’s exactly the kind of opacity that decentralized protocols were built to solve.
Core: The Technical Gap No One Is Talking About
Based on my audit experience during the 2022 bear market, I spent hundreds of hours tracing ZK-rollup scalability solutions and building a visualization tool for proof generation times. I learned that the hardest part of any trustless system is not the math — it’s the narrative. You can have the most elegant STARK proof in the world, but if users don’t believe the data is authentic, the protocol fails.
Now, apply that lesson to Anthropic. The company’s $65 billion run rate is a function of enterprises buying API access to Claude, its flagship model. But those enterprises have no way to verify that the responses they receive are generated by the exact model they’re paying for, or that the data used for training hasn’t been poisoned. In a centralized AI world, trust is a binary choice: you either take the company at its word, or you don’t.
This is where blockchain becomes more than a financial ledger — it becomes a truth machine. Imagine a world where every inference request to Claude is accompanied by a zero-knowledge proof that the output was generated by a specific model version, with a specific set of weights, and that no data leakage occurred. That’s not science fiction. I prototyped a similar system in 2025 called “TruthLayer,” a decentralized registry for AI-generated media. We integrated watermarking algorithms with IPFS storage and had 500 beta testers within a month. The tech worked. What we lacked was the economic incentive for adoption.
But now, with Anthropic’s run rate soaring, the incentive is clear. Enterprises paying millions for AI services will soon demand cryptographic assurances. The Wall Street Journal recently reported that several Fortune 500 companies are already exploring on-chain attestation for AI outputs. The cost of a false AI response — whether a hallucinated financial analysis or a biased hiring decision — far exceeds the cost of a few gas fees.
The poetry of liquidity that I wrote about during DeFi Summer now applies to AI attention. Just as Curve’s stableswap invariant replaced traditional banking intermediaries with mathematical elegance, so can blockchain replace the black-box trust model of centralized AI. The run rate is not the story — the underlying trust architecture is.
Contrarian: The Delusion of Speed
I know what the pragmatists will say. “Chris, you’re overcomplicating this. Anthropic is growing 622% in seven months. They don’t need blockchain. They need to hire more engineers and scale faster.” And they’re right — in the short term.
But here’s the counter-intuitive truth: the faster a centralized AI company grows, the more vulnerable it becomes to a single point of failure. The bear market didn’t kill the need for resilience; it taught us that survival matters more than gains. In 2022, I watched projects with $1 billion TVL collapse overnight because a single oracle failed. The same logic applies to AI. If Anthropic’s training data is compromised, or if a rogue employee inserts a backdoor, the entire $65 billion run rate evaporates in a day. The market will not forgive opacity twice.
Moreover, the SEC filing for the IPO will require unprecedented transparency. The company will have to disclose its data sources, model governance, and security protocols. Smart investors will look for on-chain proof of these claims. I have already spoken to three venture capital firms that are conditioning their AI investments on the existence of a verifiable identity layer for model outputs. The institutional bridge I helped build in 2024 — translating blockchain jargon into business value — is now being crossed from the other side.
Some might argue that blockchain adds latency and cost. But the same argument was made against rollups in 2022. Today, Optimistic and ZK-rollups process millions of transactions per day at negligible cost. The technology is ready. What’s missing is the trigger — and Anthropic’s IPO might be it.
Takeaway: The Horizon Is Hybrid
About Me: I’m Chris Thompson, a 29-year-old decentralized protocol PM in Nairobi who started coding in 2017 because I believed that code could be a social contract. I’ve seen three market cycles, survived two bear markets, and built a prototype for AI verification that 500 strangers trusted. I’m not a financial analyst, and I don’t own Anthropic stock. But I do own a conviction: the next trillion-dollar company will not be purely centralized or purely decentralized — it will be a hybrid that uses blockchain to prove what it claims.
Anthropic’s $65 billion run rate is a signal, not a destination. It tells us that the market is ready for AI at scale. But scale without trust is a castle built on a marsh. The question is not whether Anthropic will succeed — it’s whether they will be the first to integrate the cryptographic infrastructure that makes their success verifiable.
We don’t need to wait for the IPO. We need to build the rails. The bear market didn’t break our curiosity — it sharpened it. Now, let’s apply that same resilience to the AI industry. Because the real revolution isn’t a run rate. It’s the ability to say, with mathematical certainty, that what you see is what was built.