Meta’s push to build out capacity for its AI business has taken a step forward with the creation of a new initiative, Meta Compute.
CEO Mark Zuckerberg announced the “top-level” unit in a Facebook post Monday, and said he will assume responsibility for the planning and operation of the company’s fleet of data centers.
“Meta is planning to build tens of gigawatts this decade, and hundreds of gigawatts or more over time,” Zuckerberg said in the post. “How we engineer, invest, and partner to build this infrastructure will become a strategic advantage.”
Meta posted notice of its intention to make a major move in this area during its second quarter earnings call last year, when it unveiled plans for massive expenditure in the pursuit of what Zuckerberg calls “personal superintelligence,” or the idea that AI can surpass human performance in most cognitive tasks. The company committed a mammoth $72 billion to AI infrastructure in its 2025 fiscal year alone.
Zuckerberg tapped Santosh Janardhan, head of global infrastructure and co-head of engineering, to work in tandem with Daniel Gross, former CEO of Safe Superintelligence and hired by Meta last summer. Both bring a level of seniority that underscores the importance of the project for the company.
Janardhan will focus on Meta’s “technical architecture, software stack, silicon program, developer productivity, and building and operating our global data center fleet and network,” according to Zuckerberg’s post. Gross’ remit will be “long-term capacity strategy, supplier partnerships, industry analysis, planning, and business modeling,” Zuckerberg added.
They will be joined by a former advisor to President Donald Trump, Dina Powell McCormick, who this week joined Meta as president and vice chair and will be tasked with “partnering with governments and sovereigns.” This is likely to be a key role in terms of raising investment and securing approvals, particularly given the vast energy consumption the initiative will require.
Meta currently has a fleet of 30 data centers globally, most of them in the U.S., and last week confirmed a deal with three nuclear energy firms — Vistra, TerraPower and Oklo — to provide power for some of its facilities, highlighting the growing demand for energy.
Although the company’s Llama 4 AI models have met with a generally muted response so far, the belief is that Meta Compute can help the company carve out an advantage over its peers. As CFO Susan Li pledged last year, “We … expect that developing leading AI infrastructure will be a core advantage in developing the best AI models and product experiences, so we expect to ramp our investments significantly in 2026 to support that work.”
