Anthropic, the artificial intelligence startup behind the Claude chatbot, is making its case to public investors with a familiar pitch: rapid revenue growth. At an investor event in San Francisco, executives including CEO Dario Amodei and CFO Krishna Rao said third-quarter revenue rose 45% from the prior quarter to $16.8 billion, according to attendees. The figures were first reported by the Financial Times and cited by Reuters.
The growth is striking. Revenue climbed from $4.7 billion in the first quarter to $11.6 billion in the second, then to $16.8 billion in the third. That kind of acceleration is rare even in the fast-moving AI sector. But the company also faces a heavy cost burden: Reuters has reported that Anthropic has committed at least $518 billion over a decade to cloud services, computing, and infrastructure. It previously lost more than $8 billion on an operating basis even as revenue grew quickly.
So the IPO story is less “look at the growth” and more “can growth outpace the compute bill.” The company is reportedly targeting a valuation above $2 trillion, and investors are treating its listing as a bellwether for other potential AI IPOs, including OpenAI.
Why the compute bill matters
Running leading AI models requires enormous computing power. That means buying or renting chips, building or leasing data centers, and signing long-term contracts with cloud providers. These costs are not like typical software expenses, which often scale down when usage falls. Many of Anthropic’s commitments are likely “take-or-pay” style deals, meaning the company must pay regardless of whether it uses the capacity.
That distinction is crucial for investors. Public markets tend to reward fast-growing companies when they can show a path to turning sales into cash. For an AI developer, the key question is whether compute costs behave like flexible, per-use expenses that fall as models become more efficient, or like quasi-fixed bills that stay high even if growth slows.
Anthropic executives also said the company was profitable on an “adjusted” basis, a framing meant to ease concerns about the high cost of running AI models. But adjusted profitability can be misleading. It often excludes stock-based compensation and other non-cash charges, and it doesn’t necessarily capture cash going out the door for computing and infrastructure.
What “adjusted” profitability really means
When a company says it is profitable on an adjusted basis, it typically means it has stripped out certain expenses to present a cleaner picture of ongoing operations. Common adjustments include stock-based compensation, one-time charges, and sometimes acquisition-related costs. For a capital-intensive business like Anthropic, the adjustments may also exclude some of the very costs that are central to its operations.
Investors should look beyond the headline number. Adjusted profitability can be a useful signal, but it doesn’t replace a full view of cash flow. For Anthropic, the real test will be whether it can show improving gross margins and free cash flow over time, not just a one-quarter adjusted profit.
What it means for investors
For everyday investors, Anthropic’s IPO is worth watching for two reasons. First, it could set the valuation yardstick for the next major AI listing, such as OpenAI. If Anthropic can convince public markets that its growth can outpace its compute bill, other AI companies may command similar multiples. If not, the entire sector could face a reality check.
Second, the company’s cost structure is a reminder that AI is not a purely software business. It is also a hardware and infrastructure business, with heavy upfront commitments. That means the economics can be very different from a typical software company, where margins expand as revenue grows. For AI developers, margins may stay thin for years as they invest in capacity.
Anthropic’s reported work on custom chips could help bend the cost curve. If the company can reduce its reliance on expensive third-party chips, it might improve margins. But custom chip development is itself costly and time-consuming, and there is no guarantee it will pay off.
The broader backdrop is also important. The AI boom has driven massive demand for chips, with companies like TSMC reporting record quarters. But that demand also means higher prices and longer lead times for the hardware Anthropic needs. As TSMC's AI chip boom shows, the supply chain is stretched, and that could keep costs elevated.
Investors will also be watching how Anthropic’s IPO is priced and how the stock trades in the early days. A strong debut could boost sentiment for other AI listings, while a weak one could cool the market. The company’s ability to show it is bending its cost curve will be central to that outcome.
For now, the message from Anthropic is clear: growth is strong, but the bills are big. The IPO will test whether public investors believe the growth can outrun the compute bill. That is a question every AI company will eventually face.
