Anthropic just said the quiet part out loud: demand has gone from strong to absurd. At its Code with Claude developer conference, CEO Dario Amodei said the company planned for 10x growth but saw 80x instead, pushing annualized revenue to a $30 billion run rate and leaving Anthropic scrambling for enough compute to keep up. That is the kind of number that makes investors grin and infrastructure teams reach for painkillers.
The company’s rise is unusually concentrated. Claude Code, its agentic coding tool, has become the main engine of growth, and Anthropic says enterprise adoption is doing much of the heavy lifting. The catch is familiar across AI: the product is scaling faster than the hardware behind it. In other words, the business is hot enough to create its own supply-chain problem.
Claude Code is carrying the company
Claude Code has been Anthropic’s breakout hit, with the company saying it reached $1 billion in annualized revenue within six months of launch and more than $2.5 billion by February 2026. Anthropic also said weekly active users doubled since January 1, and business subscriptions quadrupled since the start of 2026. For a product launched publicly in mid-2025, that is a ridiculous pace even by AI standards.
The appeal is practical, not magical. Claude Code reads a codebase, plans work, executes actions with real developer tools, checks the result, and adjusts. Anthropic says the average developer using it spends 20 hours a week with the tool, and that most of the company’s own code is now written by Claude Code. That self-usage loop is valuable because it tightens the product faster than a normal customer feedback cycle ever could.
- $30 billion: Anthropic’s annualized revenue run rate in April
- $1 billion: Claude Code’s annualized revenue within six months of launch
- 1,000+: enterprise customers spending more than $1 million per year
Enterprise demand is doing the heavy lifting
Anthropic says it now has more than 1,000 enterprise customers spending over $1 million per year on Claude services, double the number from February. Uber and Netflix are among the customers the company cited as part of that surge. That matters because enterprise buyers are stickier than hobbyists and far less forgiving when reliability slips, which is exactly where Anthropic has recently been feeling strain.
The company’s own revenue milestones show how fast this has moved: $87 million in January 2024, $1 billion by December 2024, $9 billion by the end of 2025, then $14 billion in February 2026, $19 billion in March, and $30 billion in April. Salesforce took about 20 years to reach $30 billion in annual revenue. Anthropic did it in under three years from a standing start.
That comparison is not apples to apples, because Anthropic is talking about an annualized run rate rather than full-year GAAP revenue. Even so, it is rare for a software company to go from essentially pre-revenue to dwarfing much of the Fortune 500 so quickly. OpenAI has reportedly argued the figure is overstated, which is a reminder that private-company math can be doing gymnastics in the background.
Anthropic’s compute bottleneck is the real story
Growth at this speed does not just create bragging rights. It creates shortages. Anthropic has already warned about strain on its infrastructure, and Amodei said the company planned for 10x growth while reality delivered 80x. That is not a nice forecasting miss. That is a signal that demand outran the hardware budget, the delivery schedule, and probably everyone’s sleep.
The most surprising fix came through a deal with SpaceX to use all the compute capacity at Colossus 1 in Memphis, Tennessee. Anthropic will get access to more than 300 megawatts of capacity and over 220,000 Nvidia GPUs, including H100, H200, and GB200 accelerators. Anthropic has also lined up more compute from Amazon, Google, Nvidia, and Microsoft, though much of that capacity is not expected online until late 2026 or early 2027.
The partnership is awkward and useful at the same time. Elon Musk has spent recent months criticizing Anthropic, but his data center now helps solve Anthropic’s immediate bottleneck. Rivalry in AI has a funny way of becoming logistics.
A near-trillion-dollar company still needs more servers
Anthropic is also said to be weighing a new funding round that could value it at more than $900 billion, after previous valuations climbed from $61.5 billion in March 2025 to $183 billion in September, then $380 billion in February. Secondary markets have already floated a $1 trillion implied valuation. The company is also reportedly considering an IPO as early as October 2026, with Goldman Sachs, JPMorgan, and Morgan Stanley already in early talks.
That valuation arc makes sense only if the company can keep turning compute into revenue faster than it burns cash. It has big infrastructure commitments with Amazon and Google, plus Nvidia capacity through SpaceX and Microsoft Azure, but the immediate question is simpler: can Anthropic keep customers happy while the hardware catches up? If it can, the company’s growth story gets even more bizarre. If it cannot, competitors will happily sell those users something less impressive but more available.
Amodei’s seven-month clock is the sharpest part of all this. He has already said he expects a single-person company worth $1 billion to appear in 2026. The real test is whether Anthropic can stop being the company with the hottest demand in AI and become the one that can actually serve it at scale.