Zuckerberg: Meta building city-sized AI data center, going on $65 billion spending spree
The future of AI will be measured in gigawatts, GPUs, and the square mileage of your data centers.
Today Meta CEO Mark Zuckerberg announced that the company will be going big on capital expenditures this year, planning to spend up to $65 billion in 2025.
According the map Zuckerberg shared, the data center would occupy approximately 4.3 square miles (about 120 million square feet), an area that would cover a “significant part of Manhattan.”
In addition to the Richland Parish, Louisiana, supersized data center, Zuckerberg said the company will be able to fill it with powerful hardware:
“We’ll bring online ~1GW of compute in ’25 and we’ll end the year with more than 1.3 million GPUs.”
All of the Big Tech companies and AI startups have been in a bit of a measuring contest to see who has the largest number of powerful Nvidia GPUs, which are used for training and running AI services.
Zuckerberg teased the company’s upcoming Llama 4 AI model, saying he expected it would start contributing “increasing amounts of code to our R&D efforts.” Meta recently announced it would be laying off about 5% of its workforce, focused on the “lowest performers” in preparation for what Zuckerberg warned employees would be “an intense year.”
Meta’s massive capex outlay comes as the Trump administration is signaling to the AI industry that it wants the US to dominate the field and is throwing its support behind large AI infrastructure projects like the recently announced “Project Stargate” joint venture between Oracle, OpenAI, and SoftBank.
According the map Zuckerberg shared, the data center would occupy approximately 4.3 square miles (about 120 million square feet), an area that would cover a “significant part of Manhattan.”
In addition to the Richland Parish, Louisiana, supersized data center, Zuckerberg said the company will be able to fill it with powerful hardware:
“We’ll bring online ~1GW of compute in ’25 and we’ll end the year with more than 1.3 million GPUs.”
All of the Big Tech companies and AI startups have been in a bit of a measuring contest to see who has the largest number of powerful Nvidia GPUs, which are used for training and running AI services.
Zuckerberg teased the company’s upcoming Llama 4 AI model, saying he expected it would start contributing “increasing amounts of code to our R&D efforts.” Meta recently announced it would be laying off about 5% of its workforce, focused on the “lowest performers” in preparation for what Zuckerberg warned employees would be “an intense year.”
Meta’s massive capex outlay comes as the Trump administration is signaling to the AI industry that it wants the US to dominate the field and is throwing its support behind large AI infrastructure projects like the recently announced “Project Stargate” joint venture between Oracle, OpenAI, and SoftBank.
