OpenAI cites Anthropic's compute constraints behind Claude Mythos delay
What's the story
OpenAI has claimed a major advantage over rival AI start-up Anthropic, citing its early investment in computing resources. The revelation comes from an internal memo reported by Bloomberg, which notes that OpenAI believes it has outpaced Anthropic by "rapidly and consistently" expanding its computing capacity to meet the growing demand for AI.
Growth strategy
OpenAI defends infrastructure expansion plan
In a memo sent to investors after the launch of Mythos, OpenAI defended its ambitious infrastructure expansion plan. The company said this strategy, which many criticized as too expensive, has allowed it to keep up with the demand for AI products. OpenAI also revealed its computing capacity stood at 1.9 gigawatts (GW) in 2025, three times what it was the previous year, and projected it would reach low-double-digit range next year before hitting around 30 gigawatts by 2030.
Capacity comparison
Disparity in capacity could explain limited release of Mythos
In contrast to OpenAI's projections, Anthropic is estimated to have ended 2025 with 1.4GW of computing capacity and expects it to reach between seven and 8GW next year. This disparity in capacity could explain why Anthropic has limited the release of its new AI model, Mythos, to select partners only.
Market demand
OpenAI takes jab at Anthropic's CEO
OpenAI also took a jab at Anthropic's CEO Dario Amodei, who had previously spoken about the "genuine dilemma" of balancing risk in AI development. The company said Amodei underestimated the market's hunger for AI products. "In hindsight, that caution looks less like discipline and more like underestimating how fast demand would arrive," OpenAI wrote in its memo.