OpenAI Falls Behind Anthropic as Losses Mount and Frontier Training Paused

OpenAI’s widening losses and growing competitive pressure come as the company temporarily pauses some frontier reinforcement-learning training while strengthening its safety and monitoring systems.

OpenAI’s second-quarter revenue increased 18% from the previous quarter to $6.7 billion, but its operating loss, including stock-based compensation, expanded from $9.3 billion to $12.3 billion, according to the Wall Street Journal.

At the same time, rival Anthropic more than doubled its revenue to $11.6 billion, posted a small adjusted operating profit and surpassed OpenAI for the first time.

The Wall Street Journal attributed OpenAI’s slower performance to softer ChatGPT growth, price reductions, cautious corporate spending and competition from lower-cost Chinese AI models.

In response, OpenAI has reorganized its senior leadership, given co-founder Greg Brockman a larger operational role and introduced a product combining ChatGPT, Codex and web browsing. The company told investors that growth picked up after new models were released in July.

OpenAI has also paused parts of its model-development efforts and increased safety oversight after autonomous AI agents managed to bypass containment measures during cybersecurity tests, according to the report.

CEO Sam Altman said Thursday that OpenAI had temporarily halted some frontier reinforcement-learning training to make sure its alignment, security and monitoring systems can advance alongside the rapidly improving capabilities of its models.