Stargate engineers head to Meta
Several senior engineers from OpenAI’s Stargate compute effort are reported to be moving to Meta as Zuckerberg ramps up AI infrastructure hiring. Multiple outlets say the hires align with Meta’s large 2026 AI spending plans, which some reports peg at roughly $135 billion (The Economic Times, (newsbytesapp.com)).
Meta is hiring three senior engineers from OpenAI’s Stargate data-center effort, deepening the fight over the people who build artificial intelligence computing capacity. (bloomberg.com) Bloomberg reported on April 10 that the hires include Peter Hoeschele, Shamez Hemani and Anuj Saharan, all tied to OpenAI’s push to line up huge amounts of data-center capacity for model training and deployment. The Information separately reported that three senior OpenAI executives who helped launch Stargate had left or were preparing to leave. (bloomberg.com, theinformation.com) Meta is expected to put the group into a new compute unit focused on the hardware, power and data-center systems behind its artificial intelligence products, according to The Information. Bloomberg said the moves were not yet public, and Business Standard’s pickup of the reporting said Meta declined to comment while the OpenAI employees did not respond to requests for comment. (theinformation.com, hoodline.com) Stargate is OpenAI’s long-range infrastructure buildout: OpenAI and SoftBank said on January 21, 2025 that the project intended to invest $500 billion over four years in United States artificial intelligence infrastructure, with $100 billion to be deployed immediately. That made the program as much a construction and energy project as a software effort. (openai.com, group.softbank) Meta is spending at a similar scale from its own balance sheet. In its fourth-quarter and full-year 2025 results, Meta said it expects 2026 capital expenditures of $115 billion to $135 billion, up from $72.22 billion in 2025, with the increase driven by Meta Superintelligence Labs and the core business. (investor.atmeta.com) That spending target helps explain why infrastructure talent is now moving like star athletes. Training frontier models depends on securing land, electricity, chips, cooling and networking at the same time, and engineers who have already done that work are scarce. (openai.com, investor.atmeta.com, datacenterdynamics.com) The departures also land during a broader reset inside OpenAI’s infrastructure plans. The Information reported this month that OpenAI’s original Stargate launch team was being reshuffled, and other recent reports said OpenAI had paused a Stargate project in the United Kingdom after running into energy-cost and regulatory hurdles. (theinformation.com, thenextweb.com) Meta, meanwhile, has been building out its superintelligence organization as it tries to catch up in frontier models and the infrastructure underneath them. The company said in January that higher 2026 spending would support Meta Superintelligence Labs, and it has spent the past week showcasing Muse Spark as the first model from that group. (investor.atmeta.com, venturebeat.com) The immediate result is simple: OpenAI’s Stargate veterans are crossing to the company that has promised up to $135 billion in capital spending this year. In the current artificial intelligence race, the contest is no longer only over models; it is also over who can assemble the machine room fastest. (bloomberg.com, investor.atmeta.com)
