Meta hiring Stargate talent
Social posts indicate Meta has been actively recruiting engineers from OpenAI’s Stargate project to support its AI data‑centre expansion, a sign of intense cross‑lab poaching for infrastructure expertise. The move underlines how competition for people who can scale training and ops is shaping hiring beyond pure model research. That kind of infrastructure hiring often favors systems experience—large experiments, data pipelines and compute optimisation. (x.com)
Meta is trying to hire the people who know how to build the factory, not just the people who design the product. Posts on X say Meta has been recruiting engineers from OpenAI’s Stargate project, the giant data-center effort OpenAI launched in January 2025. (x.com) (openai.com) Stargate is not a software feature or a new chatbot. OpenAI said on January 21, 2025 that Stargate is a company meant to invest $500 billion over four years in United States artificial intelligence infrastructure, with $100 billion to begin immediately. (openai.com) That means land, power lines, cooling systems, networking gear, and buildings packed with chips. OpenAI said in July 2025 that its Oracle partnership alone would take Stargate to more than 5 gigawatts under development, enough for more than 2 million chips. (openai.com) Meta is building the same kind of machine on its own side. Meta said its Richland Parish site in Louisiana, announced in December 2024, will house its largest multi-gigawatt artificial intelligence training cluster, called Hyperion. (about.fb.com) (datacenters.atmeta.com) By October 2025, Meta had also broken ground on an El Paso site that can scale to 1 gigawatt. In January 2026, Meta said it was lining up nuclear energy projects that could unlock up to 6.6 gigawatts for its artificial intelligence buildout. (about.fb.com 1) (about.fb.com 2) The hiring fight makes sense because these projects are bottlenecked by people who know how to run giant clusters without wasting power or time. Oracle said its newest training clusters are designed so equipment sits within a two-kilometer radius to reduce chip-to-chip delay at the Abilene Stargate site. (oracle.com) That kind of work is closer to airport logistics than app design. A team has to move electricity, data, water, replacement parts, and software jobs in the right order, because one weak link can leave thousands of expensive chips waiting idle. (ft.com) (oracle.com) OpenAI has been expanding Stargate fast enough that the people inside it have unusually rare experience. In September 2025, OpenAI said Stargate had nearly 7 gigawatts of planned capacity and more than $400 billion in investment over the next three years. (openai.com) Meta is spending at a scale where a few senior infrastructure hires can move real money. In January 2026, Meta told investors it expected 2026 capital spending of $115 billion to $135 billion, driven largely by infrastructure. (finance.yahoo.com) So this is not just one company poaching another company’s engineers. It is a sign that the race in artificial intelligence now runs through substations, cooling loops, fiber, and scheduling software, and the people who have already done that once are becoming as valuable as top model researchers. (openai.com) (about.fb.com)
