Anthropic is about to become the first profitable AI lab
| Source: THE DECODER
Tags: Anthropic, profitability, Claude, Opus 4.7, agentic AI, enterprise AI, revenue
Anthropic is on track for $559 million in Q2 operating profit on $10.9 billion in revenue — a 130% year-over-year jump — making it the first major AI lab to turn profitable, two years ahead of its own projections, driven by enterprise coding tool adoption and agentic Claude workloads.
Details
Anthropic's Q2 numbers, reported by the Wall Street Journal, mark a historic inflection: the company expected to remain unprofitable until 2028 as recently as last summer. Now it projects $559 million in operating profit on $10.9 billion in revenue for the June quarter — growth that outpaces Zoom during the pandemic and pre-IPO Google and Facebook. The primary drivers are coding tools, which enterprises have adopted en masse since early 2026, and 'agentic' Claude usage — multi-step autonomous tasks that generate large token volumes. Demand has at times exceeded compute capacity, forcing Anthropic to throttle some users and sign new data center deals, including one with Elon Musk's SpaceX. CEO Dario Amodei joked that revenue growth had become hard to manage. There is a pricing shift embedded in the success story. Anthropic's Opus 4.7 uses a new tokenizer that breaks text into up to 47% more tokens than the previous model, raising effective per-request costs 12–27% for prompts over 2,000 tokens — confirmed by OpenRouter analysis. Anthropic cut its own compute costs from 71 cents to 56 cents per dollar of revenue quarter-over-quarter. Unlike OpenAI, Anthropic relies primarily on chips secured through investor deals with Google and Amazon.