Anthropic’s S-1 has brought the economics of a frontier AI company into view. The filing points to rapid growth and strong margins, while also raising a harder question: where would the much larger pool of spending needed to support a trillion-dollar business come from?
Strong financials meet a much bigger ambition
In an interview with The Verge’s Decoder, SemiAnalysis analyst Joey Brookhart described the filing as evidence that Anthropic had scaled efficiently. He pointed to its gross margins, compute margins, and free cash flow as encouraging signs. The filing also gives investors a rare, detailed view of how a leading AI lab operates financially.
But favorable performance today does not settle the question of how far the business can grow. The article says Anthropic could begin trading on the Nasdaq as soon as next month, at a valuation between $1.5 trillion and $2 trillion. The interview also considers the much larger possibility of a $10 trillion company, which rests on assumptions about AI’s future reach and the size of the market it could serve.
The scale of current spending is still concentrated. Brookhart said the two leading labs together had $120 billion of ARR, while the heaviest enterprise customers were spending $50,000 to $100,000 per employee. The broader Fortune 500 and Global 2000, despite their substantial IT spending, were not yet spending much on these services.
Coding is a foothold, not the whole market
Anthropic’s business is especially focused on coding. That gives the company a clear use case, but it also means that broader adoption is central to the larger growth case. The interview discusses a possible expansion into white-collar work, cybersecurity, and bio, where AI could support more kinds of business activity.
Brookhart’s argument is that getting to $1 trillion in revenue would require a shift from AI augmentation toward labor savings or replacement in some areas. In plain terms, helping an employee do a task is a different economic proposition from reducing the amount of work, or staffing, needed to complete it. The second could create a larger pool of spending, but the source presents it as a future possibility, not an established outcome.
Enterprise purchasing patterns could shape how quickly that expansion happens. The interview notes that the labs had not yet gone through a meaningful enterprise Q4 budgeting season since their ARR began to accelerate. That leaves open how large organizations will plan, approve, and scale AI spending as part of regular budgets.
Competition and control affect adoption
Anthropic is not alone in pursuing these customers. The interview describes several leading labs, open weight models, and companies with the ability to run AI on their own systems. It also notes that some CEOs are reluctant to hand their data to Anthropic or OpenAI, and that law firms are running AI on-prem.
Those choices matter for the market’s shape. Some coding tools have low switching costs, making it easy for customers to move between providers. Meanwhile, enterprises and AI-native companies are doing custom training and post-training on open weight models, then using them for inference or internal and external products.
Brookhart also pointed to well-funded competitors, including DeepMind at Google and Meta Superintelligence Labs, with similar training compute capacity. The article mentions SpaceX as well, in connection with plans for additional gigawatts and internal use. For Anthropic, the growth opportunity therefore depends not only on demand for AI, but on how much of that demand it can capture amid competing models and deployment choices.
The filing offers evidence, not a final answer
The S-1 provides more information about Anthropic’s finances than was previously available, and the interview treats its margins and cash flow as positive signals. Yet a filing can describe present economics more clearly than it can prove future adoption. The central uncertainty is whether businesses across enough fields will find AI valuable enough to support much greater spending.
Investors weighing Anthropic’s ambitions have to consider both sides. The company has shown rapid growth in a market led by coding demand, and it may have opportunities in areas such as cybersecurity and broader white-collar work. The path to a far larger valuation, however, depends on those opportunities becoming widespread enterprise use and meaningful labor savings, while customers choose Anthropic over rivals or open models.
That is why the S-1 is more than a snapshot of revenue and costs. It frames the assumptions behind the AI economy: that new uses will emerge, companies will commit larger budgets, and the value created will be large enough to sustain the growth investors are being asked to believe in.