Table of Contents
Meta Compute will focus on the infrastructure layer supporting Meta’s AI ambitions
In sum – what to know:
Meta formalizes AI infrastructure strategy – The creation of Meta Compute signals a long-term commitment to AI data center development at unprecedented scale, reaching tens of gigawatts this decade.
Clear split between AI and infrastructure roles – While AI research leadership focuses on models and systems, Meta Compute is dedicated to capacity planning, power, suppliers, and financing.
Power and capital are central constraints – Nuclear partnerships, joint ventures, and multi-billion-dollar cloud deals underline the scale of investment required to support Meta’s AI roadmap.
Meta has established a new internal division, called Meta Compute, to oversee its rapidly expanding AI data center and infrastructure strategy, as the company prepares for buildouts measured in tens and eventually hundreds of gigawatts.
“Meta is planning to build tens of gigawatts this decade, and hundreds of gigawatts or more over time,” CEO Mark Zuckerberg said.
The new organization will be co-led by Santosh Janardhan, Meta’s head of global infrastructure, and Daniel Gross, formerly CEO of Safe Superintelligence (SSI).
Meta Compute will focus on the infrastructure layer supporting Meta’s AI ambitions. Janardhan will continue to oversee technical architecture, silicon, and data center operations, while Gross will lead long-term capacity planning, supplier partnerships, and business modeling.
The move follows heavy investment in AI talent and infrastructure. Zuckerberg recently disclosed that Meta spent approximately $14.8 billion to acquire a 49% stake in data labeling firm Scale AI and hired its CEO, Alexandr Wang, to lead a new superintelligence lab. While Wang will concentrate on AI systems and models, Meta Compute will handle the physical infrastructure required to support them.
Meta has already signaled rising capital expenditures, warning investors in October that capex in 2026 will be “notably larger” than its record spending in 2025. In addition to building its own gigawatt-scale campuses, Meta has signed or discussed multiple large infrastructure deals, including agreements with Google, CoreWeave, Nebius, and Oracle.
The company has also recently announced new nuclear energy partnerships with TerraPower, Oklo, and Vistra, which together could provide access to up to 6.6GW of nuclear power for future data centers.
In September 2025, Zuckerberg had said the company expects to spend at least $600 billion on U.S. data centers and related infrastructure by 2028.
