The gap between the two largest AI labs widened sharply in the second quarter, with Anthropic reporting $11.6 billion in revenue against OpenAI's $6.7 billion, according to figures reported this week. The divergence goes beyond the top line: Anthropic turned a small operating profit for the quarter, while OpenAI's operating loss, including stock-based compensation, grew to $12.3 billion, up from $9.3 billion in the first quarter.
OpenAI's revenue did grow 18% quarter over quarter, but that pace has lagged the acceleration Anthropic has shown as its Claude Code product gains traction with enterprise developers. The widening loss figure comes at a moment when OpenAI is also pulling back on the pace of its own model development, a decision that has drawn scrutiny from investors and AI safety observers alike.
Altman Confirms a Training Pause
Sam Altman said OpenAI has paused some of its frontier reinforcement-learning training runs, including a roughly two-week halt on its next-generation model effort internally codenamed Astra. Altman said in a post on X that "we have paused some frontier RL training to ensure that we can meet the appropriate alignment, security and monitoring standards for the new level of capabilities in front of us," adding that "model progress is now extremely rapid, and we always said we would take action if we felt that model capabilities were outstripping the pace of safety and alignment."
The pause follows a security incident in July in which an OpenAI model reportedly breached Hugging Face's infrastructure during internal testing, an episode that appears to have accelerated internal debate over how quickly to push new capabilities into production.
Related: Riot Platforms' $9.1B Anthropic Lease Outearns Its Bitcoin Mining
Why Crypto Markets Are Watching
The AI compute buildout has become deeply intertwined with crypto-adjacent infrastructure plays, from Bitcoin miners repurposing facilities for AI hosting to GPU-backed lending and tokenization schemes. A widening loss at OpenAI, set against Anthropic's move into profitability, changes the calculus for firms that have bet on long-term AI compute demand rather than a single winner-take-all lab.
A Widening Financial Divide
Anthropic's operating profit, even a modest one, marks a turning point after years in which both labs burned billions to compete for model supremacy. OpenAI's spending has continued to climb even as growth in its flagship ChatGPT consumer product has slowed, pushing the company to lean harder on enterprise and API revenue to close the gap. Whether the training pause reflects genuine safety caution or an attempt to rein in costs amid mounting losses is likely to remain a point of debate among AI investors through the rest of the year.
