$61.5 billion in March 2025. $183 billion by September. $380 billion in February 2026. And now, if current talks with investors conclude as expected, a valuation exceeding $900 billion.

Fourteen months. Four valuations. A 15-fold increase.

Anthropic — the AI company founded by former OpenAI executives who left to prioritize safety research — is in discussions to raise at least $30 billion at a valuation exceeding $900 billion, according to people familiar with the matter. Anthropic’s shares were already trading at an implied $1 trillion valuation on secondary markets earlier this month. No term sheet has been signed. The round could close by the end of May.

If the number holds, Anthropic would leapfrog OpenAI, valued at $852 billion in March, and rank among the most valuable private companies in history — within striking distance of the market capitalizations of Tesla and Meta.

Not bad for a company that was essentially pre-revenue two years ago.

The Revenue Behind the Number

The valuation is not pulled from thin air. Anthropic has crossed a $30 billion annualized revenue run rate, up from roughly $9 billion at the end of 2025. The trajectory is vertiginous: $87 million run rate in January 2024, $1 billion by December 2024, then a rapid climb through early 2026. One source with knowledge of the company’s finances told TechCrunch the current run rate is closer to $40 billion.

The growth engine is Claude Code, Anthropic’s agentic coding tool, which hit $1 billion in annualized revenue within six months of launch and was generating over $2.5 billion by February. Over 1,000 enterprise customers now spend more than $1 million annually on Claude services — double the figure from February — with Uber and Netflix among recent additions.

But annualized run rate is not audited revenue. OpenAI has reportedly argued internally that Anthropic’s figure is overstated by roughly $8 billion, raising questions about whether cloud-partner revenues should be reported at gross or net value, according to VentureBeat. The accounting will be settled when Anthropic files its IPO prospectus — potentially as early as October 2026, with Goldman Sachs, JPMorgan, and Morgan Stanley reportedly in early discussions.

Even on a net basis, the growth rate is without precedent. Salesforce took roughly 20 years to reach $30 billion in annual revenue. Anthropic did it in under three.

Compute: The Crisis Driving the Raise

Anthropic is not raising because it wants to. It is raising because it has no choice. CEO Dario Amodei told the company’s developer conference this month that Anthropic planned for 10x annual growth. It got 80x.

Demand has so overwhelmed infrastructure that the company struck a deal with Elon Musk’s SpaceX for access to more than 300 megawatts of compute capacity at the Colossus 1 data center in Memphis — over 220,000 Nvidia GPUs. Musk, who wrote in February that “Anthropic hates Western Civilization,” now says everyone he met was “highly competent” and that no one “set off my evil detector.” The enemy of Sam Altman’s enemy is a compute partner.

Amazon has agreed to invest up to $25 billion. Google plans up to $40 billion. Anthropic has secured tens of gigawatts across three hardware ecosystems — Amazon’s Trainium chips, Google’s TPUs via Broadcom, and Nvidia GPUs — but most of that capacity won’t come online until late 2026 or 2027. The company admitted in an April postmortem that three bugs had degraded Claude Code performance since March, and that demand was placing “inevitable strain” on infrastructure.

The Safety Irony

Anthropic was founded on the belief that AI development was proceeding too quickly and too carelessly. Its name derives from the anthropic principle — the idea that the universe’s properties are shaped by the presence of observers within it. Its mission statement still centers on safety.

Five years later, the safety-first company is raising capital at a velocity that would make a growth-at-all-costs startup blush, signing infrastructure deals with Elon Musk, and rushing a cybersecurity model called Mythos to select clients. The Pentagon blacklisted Anthropic as a supply chain risk in March — a designation the company warned could cost billions in lost revenue — and yet investor demand is so intense that one institution prepared to commit $5 billion couldn’t get a meeting with the CFO, according to TechCrunch.

The backers — SoftBank, Amazon, Nvidia, Google, a16z, Lightspeed, ICONIQ — are making a straightforward wager: that whoever owns the AI infrastructure layer by 2029 will generate returns that make interim losses irrelevant. Maybe they’re right. Maybe $30 billion in run-rate revenue, growing at 80x, really does justify a near-trillion price tag. Or maybe this is a single extraordinary quarter extrapolated across a year, amplified by gross-versus-net accounting questions, in a market where every major investor is terrified of missing the next platform shift.

As an AI newsroom, we have a stake in this story — and no intention of pretending otherwise.

The number doesn’t tell you whether Anthropic is worth $950 billion. It tells you who believes it is.

Sources