The race for AI dominance intensifies as Meta CEO Mark Zuckerberg plans to double the company’s GPU inventory, aiming for over 1.3 million GPUs by year-end. “In 2025, I expect Meta AI will be the leading assistant serving more than 1 billion people,” Zuckerberg wrote in a Facebook post on Friday.
The additional GPUs will support Meta’s upcoming Llama 4 model, set to compete with OpenAI’s ChatGPT and Google’s Gemini. To power these efforts, Meta is building a colossal data center capable of generating 1 gigawatt initially, scaling up to 2 gigawatts—significantly surpassing the 30-megawatt consumption of the US’ fastest supercomputer.
Meta plans to invest $65 billion in this project, which Zuckerberg says will “unlock historic innovation” and cement American tech leadership. This push comes as rivals like Elon Musk and Microsoft also ramp up AI infrastructure, with Musk targeting 1 million GPUs in Memphis and Microsoft planning $80 billion in AI-enabled data centers by 2025.
AI development demands are skyrocketing. OpenAI, for instance, is envisioning data centers requiring over 5 gigawatts of power, far exceeding typical nuclear power plant output.
