A major shift is underway inside Meta as the company moves to directly control the backbone of its artificial intelligence future. CEO Mark Zuckerberg has revealed a new internal program called Meta Compute, designed to build and operate large scale AI infrastructure entirely under Meta’s ownership.
The initiative focuses on deploying tens of gigawatts of computing capacity this decade, with long term plans to scale into the hundreds of gigawatts. Meta sees this infrastructure as a long lasting competitive edge, supporting advanced AI systems and its broader vision of personal superintelligence.
Leadership for Meta Compute is split across technical and strategic roles. Santosh Janardhan will oversee engineering, architecture, software, and data center operations, while Daniel Gross will manage capacity planning, partnerships and financial modeling. Dina Powell McCormick will work closely on government and sovereign relationships tied to infrastructure financing.
Key Objectives of Meta Compute:
- Build proprietary AI infrastructure at massive scale
- Reduce reliance on external compute providers
- Support next generation AI models and agents
- Secure long term energy access for data centers
The move highlights how deeply infrastructure is becoming tied to AI leadership as power availability and scale increasingly define success. As Meta accelerates its push toward owning AI infrastructure end to end, energy access is emerging as a defining constraint. The scale envisioned under Meta Compute aligns with the company’s broader strategy of securing dedicated power sources for AI data centers
