Meta announced Meta Superintelligence Labs on June 30, 2025, bringing its foundation-model, AI product and research teams under chief AI officer Alexandr Wang. The reorganization combined work on existing AI systems with a new lab focused on developing the company’s next generation of models.
In his June 30 internal memo, CEO Mark Zuckerberg named former GitHub CEO Nat Friedman to lead AI products and applied research alongside Wang. The broader organization also included Meta’s Fundamental AI Research team, known as FAIR.
Wang’s appointment followed Scale AI’s June 12 announcement that its founder was joining Meta under an agreement that included a new investment giving Meta a minority stake in Scale. Scale said Wang would remain on its board.
Zuckerberg outlined parallel efforts to continue developing Meta’s Llama models and begin research on a new generation of models. He set a target for the research effort to reach the industry’s frontier within roughly a year of the June announcement.
Recruitment and staffing changes
Zuckerberg’s June memo named recruits from OpenAI, Google and Anthropic. The frontier-research team, known as the TBD Lab, was one part of the broader Meta Superintelligence Labs organization.
Meta spokesperson Dave Arnold said Ethan Knight was the only active staffer to have left the TBD Lab at the time of his comments. Arnold said Avi Verma and Rishabh Agarwal had never started roles in that team.
Agarwal publicly addressed his decision on X, writing: “It was a tough decision not to continue with the new Superintelligence TBD lab, especially given the talent and compute density.” He added: “But after 7.5 years across Google Brain, DeepMind, and Meta, I felt the pull to take on a different kind of risk.”
Infrastructure spending and revenue plans
Meta’s October 29 financial outlook described investments spanning its existing services and new AI development. The company expected advertising and engagement initiatives to support revenue growth in 2026, while progress on AI models and products would position it for new revenue opportunities in subsequent years.
Meta forecast 2025 capital expenditures, including principal payments on finance leases, of $70 billion to $72 billion. It said its computing needs had expanded beyond the previous quarter’s expectations and that it planned to add capacity through its own infrastructure and third-party cloud providers.
The company expected the dollar increase in capital expenditures to be larger in 2026 than in 2025. It also forecast that total expenses would grow at a significantly faster percentage rate in 2026, driven primarily by infrastructure costs, including cloud expenses and depreciation. Employee compensation was expected to be the second-largest contributor to expense growth.
Separately, Meta said in November that it was committing more than $600 billion in the United States by 2028. The commitment covered AI technology, infrastructure and workforce expansion, including data centers; it was broader than a data-center construction budget.
