Ex‑OpenAI Stargate leaders to Meta
Reports say former OpenAI leaders from the Stargate data‑center project are joining Meta, suggesting personnel shifts in AI infrastructure teams. The report surfaced on social platforms as a recruitment and movement story within AI infrastructure circles. (x.com)
Three former OpenAI leaders from the Stargate data-center effort are set to join Meta, according to Bloomberg and The Information. (bloomberg.com, theinformation.com) Bloomberg reported the hires are Peter Hoeschele, Shamez Hemani, and Anuj Saharan, three executives tied to OpenAI’s push to line up vast artificial intelligence computing capacity. The Information said they are joining a new Meta compute unit. (bloomberg.com, theinformation.com) Stargate is OpenAI’s plan to build the warehouses of chips and power systems that train and run large artificial intelligence models. OpenAI said on January 21, 2025 that the project aimed to invest $500 billion over four years in United States infrastructure, with $100 billion to be deployed immediately. (openai.com, group.softbank) The staffing move lands as OpenAI has been reworking that infrastructure strategy. The Information reported this week that the three executives had left or were preparing to leave as OpenAI reshuffled leadership around Stargate. (theinformation.com) OpenAI also paused its Stargate project in the United Kingdom on April 9, citing energy costs and regulation. CNBC reported the company said it would keep exploring the plan when conditions support longer-term infrastructure investment. (cnbc.com, bloomberg.com) Meta is hiring into that backdrop while expanding its own spending on the physical side of artificial intelligence. In its latest guidance, Meta said 2026 capital expenditures will be $115 billion to $135 billion, driven in part by Meta Superintelligence Labs and related infrastructure. (cnbc.com, datacenterdynamics.com) Meta has already started shipping products from that reorganization. On April 8, Meta introduced Muse Spark and said it was the first model from Meta Superintelligence Labs after what the company called a rebuild of its artificial intelligence stack over the prior nine months. (about.fb.com, cnbc.com) The race here is not only about model researchers; it is also about the people who secure land, power, cloud contracts, and chip supply. Reuters reported on April 9 that CoreWeave expanded an agreement to provide Meta with $21 billion in cloud capacity as the company scales infrastructure for heavier artificial intelligence workloads. (msn.com) Neither report described the hires as public yet, and coverage republished by Business Standard said Meta declined to comment while the OpenAI employees did not respond to requests for comment. The immediate result is a clearer picture of where one part of the artificial intelligence talent fight is moving: from model labs into the teams that build the machine rooms underneath them. (hoodline.com, bloomberg.com)
