Meta Bets Big on Compute Power, Recasting Itself as a Global AI Infrastructure Heavyweight

Meta is making a decisive—and expensive—pivot to cement its role as one of the world’s most powerful AI infrastructure players. In a recent announcement on Threads, CEO Mark Zuckerberg unveiled Meta Compute, a new top-level initiative overseen by the company’s most senior leaders, underscoring Meta’s ambition to dominate the rapidly intensifying race to build the physical backbone of artificial intelligence.

The move signals that Meta does not intend to trail rivals such as Microsoft, Google, OpenAI, or xAI in the global data center and compute build-out. Instead, the company is doubling down on an audacious vision: constructing AI infrastructure measured not in megawatts, but 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,” Zuckerberg wrote, adding that how the company engineers, finances, and partners to build this infrastructure will become a lasting strategic advantage.

Under the new structure, longtime Meta executive Santosh Janardhan will continue to oversee technical architecture, software systems, custom silicon, and the daily operations of Meta’s sprawling global data center network. Meanwhile, Daniel Gross, a high-profile AI hire and former cofounder of Safe Superintelligence alongside ex–OpenAI chief scientist Ilya Sutskever, will lead long-term compute strategy—forecasting future capacity needs, selecting locations, securing chips and energy, and modeling the business impact of these massive investments.

Zuckerberg also named Dina Powell McCormick as Meta’s new president and vice chair, tasking her with forging government partnerships to finance and deploy data centers worldwide. Powell McCormick previously served as deputy national security advisor for strategy, bringing geopolitical and financial expertise to a role increasingly critical for hyperscalers seeking power, permits, and political alignment.

To some industry observers, the announcement raised eyebrows. Meta is already among the largest AI infrastructure operators, having broken ground last year on its massive Hyperion campus in Louisiana—a four-million-square-foot site Zuckerberg once compared in scale to lower Manhattan. Why formalize a new organization for work already well underway?

Analysts say the answer lies in perception and positioning. With competitors investing comparable or greater sums, Meta has faced a growing narrative that it is chasing leaders like Google and OpenAI. “This was Meta signaling—to investors and employees—that it remains a serious contender,” said Patrick Moorhead, founder of Moor Insights and Strategy.

Rick Pederson of Bow River Capital echoed that view, noting that while Meta spent over $70 billion on AI infrastructure last year and plans up to $600 billion more over the next two years, rivals are spending at similar levels. “This gave Zuckerberg a chance to clarify not just the commitment, but the strategy behind it,” he said.

Others see a deeper shift underway. Former OpenAI infrastructure policy head Lane Dilg argues Meta is now treating infrastructure less as a cost center and more as an investment portfolio. Data centers, GPUs, power contracts, and custom chips are no longer just operational necessities—they are strategic assets akin to financial instruments.

That perspective helps explain Gross’s leadership role. Known for blending AI development with compute strategy, Gross previously helped build the Andromeda Cluster, a large GPU network made available to startups at below-market rates. His recent recruiting push—seeking talent across deep learning, supply chains, semiconductors, energy markets, and geopolitics—suggests Meta is preparing for volatility in hardware costs, power availability, and global supply chains.

Powell McCormick’s appointment also reflects this reality. As hyperscalers increasingly invest directly in power generation and complex financing structures, her banking and political background could help Meta accelerate approvals and secure energy at scale.

Still, critics warn that the strategy carries risks. Investor Michael Burry publicly criticized the move, arguing that Meta’s embrace of gigawatt-scale infrastructure could erode returns on invested capital and transform the company into something resembling a capital-heavy utility.

Whether Meta Compute proves to be a masterstroke or a drag on profitability remains to be seen. What is clear, however, is that Meta is staking its future on owning not just AI models and platforms—but the physical infrastructure that powers the next era of intelligence.