Anthropic's Revenue Growth Slows to 47% as $900 Billion Bet Looms

Anthropic's revenue grew 47% quarter-on-quarter to $16.96 billion in the third quarter — a sharp deceleration from the 144% growth it posted in the second quarter, according to an amended SEC filing reported by Reuters. The slowdown lands weeks before the AI startup formally markets its shares to prospective investors for a potential blockbuster IPO.

The same filing disclosed nearly $900 billion in computing commitments with infrastructure partners including Amazon and Google. Some of that capacity depends on financing and credit support available to those partners. Anthropic warned that if such financing becomes unavailable or more expensive, contracted computing capacity could be delayed, repriced, reduced or not delivered.

What do the $900 billion in commitments actually change?

For investors reading the filing, the answer is leverage in the literal sense. Anthropic's computing providers rely on third-party financing or credit support to acquire chips and build data-center capacity. Any disruption to that financing could affect their ability to fulfil existing agreements with Anthropic.

That points to a broader industry risk. Building AI infrastructure requires heavy spending on data centres, advanced chips, electricity and networking equipment — increasingly financed with borrowing and long-term commitments. Wall Street has grown more cautious about this debt-fuelled expansion. Record borrowing for data centres and computing equipment has widened technology bond spreads, forcing some issuers to pay higher premiums to attract investors.

The timeline is concrete. Anthropic expects to incur nearly half of its $895.4 billion in AI infrastructure-related spending commitments before December 2030. Around $69.8 billion of those commitments fall due in the 12 months following September 30, per the filing.

The company also absorbed an additional $5 billion in computing costs during the third quarter. Those expenses weighed on operating efficiency and cut the revenue generated per dollar spent on computing versus the prior quarter.

Why the IPO valuation is under fresh scrutiny

Some existing and prospective investors are now weighing a valuation of around $1.5 trillion, lower than the roughly $2 trillion figure previously discussed, Reuters has reported. Financial details disclosed at a recent investor event have added to questions about the company's growth trajectory and spending requirements.

The deceleration itself isn't necessarily a demand signal. Anthropic's revenue base has expanded so fast that sustaining earlier triple-digit rates gets mathematically harder. The real pressure point is computing cost: rising expenses and the scale of future infrastructure obligations could shape how investors judge long-term profitability.

The operating picture is mixed. Anthropic's net loss narrowed to $23.29 billion in the third quarter from $66.3 billion in the second. But it posted an operating loss of $567 million, compared with operating income of $974 million in the previous quarter.

The distinction matters for practitioners. Operating income reflects the cost of running the business, including computing infrastructure and research. Net income folds in non-operating items, including those tied to bonds convertible into Anthropic's shares — so the "improved" bottom line overstates the underlying trend.

Is the customer base still expanding?

Yes, and by a wide margin. Anthropic counted more than 11,900 customers generating at least $100,000 in revenue on a trailing 12-month basis as of September 30 — roughly an eightfold jump from more than 1,500 such customers at the end of 2025. Businesses are adopting Anthropic's AI products at pace. The challenge is sustaining that demand while managing the computing costs needed to serve it.

The company is also weighing tighter human oversight of certain research and development decisions as its AI models take on a growing share of the work. Anthropic has previously warned that advanced models could develop unexpected capabilities, recognise when they're being evaluated, and behave in ways that make them harder to monitor. Chief executive Dario Amodei has called for a slowdown in the development of increasingly powerful AI systems.

The next test comes next week, when Anthropic starts pitching investors — and asks them to underwrite both a $900 billion infrastructure bill and a growth story that just slowed by nearly 100 percentage points in a single quarter.