Anthropic told investors it will be profitable for a second consecutive quarter, with gross margins exceeding 80% before partner revenue-sharing and training costs, according to the Financial Times1,2.
The disclosure, reported on September 13, marks a financial milestone for the AI lab, which has positioned itself as a safety-focused competitor at the frontier of large-language-model development.
The 80%-plus gross-margin figure excludes partner revenue sharing and training costs. ANALYSIS Because those two line items, particularly training compute, represent substantial expenditures for frontier labs, the headline margin figure does not equate to operating or net profitability on a fully loaded basis; the metric instead reflects the unit economics of serving inference to customers before those deductions.
The profitability update arrives the same day Anthropic CEO Dario Amodei published an essay calling on the AI industry to slow the pace of frontier development, a position that drew a public response from White House AI adviser David Sacks, who wrote that Anthropic and OpenAI "are already free to pace the frontier and should do so for business reasons, instead of first demanding a preferred regulatory framework". ◆ The juxtaposition of a strong financial quarter and a call for voluntary restraint sharpens the tension between Anthropic's commercial trajectory and its stated safety posture, a dynamic explored in prior coverage of the lab's internal-safety reckoning[1].
Anthropic's profitability claim lands amid a period of surging cloud-infrastructure demand driven by AI workloads. Oracle, one of the major cloud providers, reported fiscal first-quarter cloud-infrastructure revenue that more than doubled year over year[2]. ◆ Rising cloud revenue across hyperscalers and infrastructure vendors points to expanding enterprise AI deployment, the demand side that underpins Anthropic's margin structure.
The Financial Times report did not surface in full-text form in the available evidence; the accessible details are limited to the headline-level claims carried by aggregators.