Meta launches "Meta Compute" initiative to build gigawatt-scale AI infrastructure
This expansion occurs amidst an industry-wide race for generative AI-ready cloud environments. Meta's peers, including Microsoft and Alphabet, have also increased infrastructure investments, with Alphabet recently acquiring the data center firm Intersect.
ADVERTISEMENT
Meta has launched Meta Compute, a new top-level initiative focused on massively expanding the company’s artificial intelligence infrastructure. CEO Mark Zuckerberg announced the project, stating that the company intends to build tens of gigawatts of power capacity this decade, with plans to scale to hundreds of gigawatts over time. This move fulfills previous projections from CFO Susan Li regarding significant capital expenditure to secure a strategic advantage in AI model development.
The initiative will be led by three senior executives with distinct purviews. Santosh Janardhan, Meta’s head of global infrastructure, will oversee technical architecture, the silicon program, and the operation of the global data center fleet. Daniel Gross, co-founder of Safe Superintelligence and a recent hire, will lead a new group responsible for long-term capacity strategy and supplier partnerships. Additionally, newly appointed President and Vice Chairman Dina Powell McCormick will manage relationships with governments and sovereigns to finance and deploy the infrastructure globally.
This expansion occurs amidst an industry-wide race for generative AI-ready cloud environments. Meta's peers, including Microsoft and Alphabet, have also increased infrastructure investments, with Alphabet recently acquiring the data center firm Intersect. Meta’s plan anticipates a substantial increase in energy consumption, aligning with estimates that US electrical demand for AI could grow from 5 GW to 50 GW over the next ten years.
