OpenAI to Deploy Its 1 Millionth GPU, Eyeing 100 Million Next?
OpenAI plans 1M GPUs by 2024, eyeing 100M for AGI dominance. Altman's moonshot vision reshapes AI infrastructure race.
"ChipPub" Publication: 20% Discount Offer Link.
OpenAI CEO Sam Altman is not known for being short-sighted, but his recent remarks have pushed the boundaries of even his typically bold technological rhetoric. In a new post on X, Altman revealed that OpenAI is expected to “deploy over 1 million GPUs” by the end of this year. That number alone is staggering.
Consider Elon Musk’s xAI, which earlier this year made waves with its Grok 4 model, running on approximately 200,000 Nvidia H100 GPUs. OpenAI’s computational power is five times that, but for Altman, it’s still not enough. “Very proud of the team...” he wrote, “but now they better start working on how to scale it up 100x, haha.”
The “haha” might sound like a joke, but Altman’s track record suggests otherwise. As early as February, he admitted that OpenAI had to slow down the rollout of GPT-4.5 because they had literally “run out of GPUs.” This is no small issue; given that Nvidia’s top-tier AI hardware is sold out through next year, it’s a wake-up call.