Anthropic's updated S-1/A filing showed third-quarter revenue climbing to $16.956 billion, while the cost of compute and infrastructure grew even faster and pushed the company from an operating profit in the prior quarter to an operating loss. The figures place the economics of securing AI capacity at the center of scrutiny as Anthropic prepares for a potential public listing.
Revenue rose 46.7% from $11.555 billion in the second quarter, according to the filing, and was far above the $1.382 billion reported in the year-earlier period. Yet compute and infrastructure costs increased 77.5% quarter-on-quarter, reaching $9.863 billion from $5.558 billion.
That expense category consumed roughly 58% of quarterly revenue, up from about 48% in the second quarter. The widening share shows that Anthropic's rapid sales growth is being accompanied by a steep increase in the cost of serving customers and expanding its model-training capacity.
Operating result turns negative as compute costs rise
Anthropic reported an operating loss of $567 million for the third quarter, compared with operating profit of $974 million in the second quarter. The filing linked the change partly to higher short-term compute expenses and about $2 billion in non-cash expense related to a donation matching program.
The company's adjusted operating profit nevertheless rose to approximately $2.5 billion from about $2.2 billion in the preceding quarter. Its adjusted operating margin declined to 15% from 19%, indicating that profit growth did not keep pace with the expanding revenue base.
The distinction between the reported operating loss and the adjusted result will likely draw close attention in any public-market presentation. Adjusted earnings can exclude items that management considers non-recurring or non-cash, while reported operating income captures the full expense burden recorded in the period.
Anthropic also disclosed that it delivered about $5 billion of short-term committed compute during the quarter. The company said those commitments carried costs above comparable multi-year purchases, showing the premium attached to obtaining capacity quickly rather than through longer-term arrangements.
Short-term compute can help an AI company meet immediate customer demand and maintain access to scarce processing resources. It can also create a cost mismatch when capacity is acquired at elevated rates before the revenue associated with that capacity has fully matured.
Revenue growth slows from the prior quarter's pace
The third-quarter revenue increase remained substantial, but it marked a slowdown from the prior period. Revenue had expanded by roughly 144% quarter-on-quarter in the second quarter, according to the filing, before moderating to 46.7% in the latest period.
A slower growth rate does not imply falling demand; Anthropic's revenue nearly reached $17 billion in the quarter. Yet the comparison becomes more consequential when infrastructure expenses are rising faster than sales. Maintaining margins would require revenue from enterprise customers and other users to absorb a growing base of compute, data-center, and capacity-related costs.
The filing's figures underline the unusually capital-intensive structure developing around frontier AI models. Companies building and running large systems must secure enough computing power both to train new models and to handle user queries after launch. Those needs can produce large commitments well before the associated revenue is certain.
Anthropic listed procurement and lease commitments of about $895.4 billion as of Sept. 30, 2026 and thereafter. About $69.8 billion was due within 12 months after that date, according to the filing. The scale of those commitments places added weight on the company's ability to convert demand for its models into durable, high-margin revenue.
Net loss driven largely by fair-value accounting changes
Anthropic reported a net loss of $23.291 billion in the third quarter. That was wider than the $12.138 billion loss reported a year earlier, though sharply narrower than the $66.309 billion loss recorded in the second quarter.
The filing attributed $21.936 billion of the third-quarter net loss to fair-value changes associated with instruments including convertible notes, derivative liabilities, and simple agreements for future equity, or SAFEs. Such fair-value movements can be large for rapidly growing private companies because changes in implied valuation affect the accounting value of financing instruments.
That accounting effect means the headline net loss does not directly represent cash spent on operations during the quarter. It nevertheless illustrates how financing structures and valuation changes can materially affect reported earnings before a company reaches public markets.
The operating figures provide a more direct view of the tension in Anthropic's current model: revenue is expanding rapidly, adjusted operating profit is positive, and the company is simultaneously carrying a heavier infrastructure burden that reduced its adjusted margin and pushed reported operating results below zero.
Filing sharpens focus on capacity commitments
The S-1/A gives prospective market participants a clearer picture of how Anthropic is funding and contracting for the computing resources needed to support its business. The company's $5 billion of delivered short-term committed compute during the quarter suggests that access to capacity has become a near-term commercial requirement rather than a distant expansion plan.
Anthropic's next challenge is to show that the higher costs tied to that capacity can be converted into revenue without permanently compressing margins. With nearly $10 billion spent on compute and infrastructure in a single quarter, future disclosures will be measured closely against the company's ability to improve the relationship between sales growth, capacity commitments, and operating profitability.
Explore how prediction markets price AI risk and volatility in 2026 in this in-depth guide today.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
