Meta lured three leaders of the Stargate project from OpenAI, which became a blow in the middle of the AI race.

Meta attracts key OpenAI specialists after a sweeping AI project review

Three senior OpenAI employees who worked on the Stargate project—a plan to build data centers (DCs) costing hundreds of billions of dollars—have moved to Meta. Their departure coincided with a sharp rise in Meta’s AI infrastructure spending and a reevaluation of several Stargate initiatives within OpenAI.

OpenAI publicly thanked these specialists for their contributions and continues to recruit experts to implement its infrastructure expansion plans. In November, the company also hired former Intel executive Sachin Katti to lead large‑scale compute capabilities.

Meta ramps up AI investment
Mark Zuckerberg announced an intention to significantly increase the budget for DCs and the specialists needed to compete in artificial intelligence. According to company forecasts, by 2026 capital expenditures will reach $135 billion, and several more hundred billion dollars are planned to be invested by the end of the decade. These funds will support Meta Superintelligence Labs and its new Muse Spark model.

What is Stargate?
The Stargate project was first presented at the White House last year as a joint venture between OpenAI, Oracle, and SoftBank with a budget of $500 billion. Later, the name became a generic term for all DC development plans within OpenAI. The company claimed it was outpacing competitors in the speed of expanding compute infrastructure for its AI models.

Key sites of the project were:

- Abilene (Texas) – here Peter Hoeschele and his team led DC work.
- A UK site – the project was frozen, and collaboration with Oracle to lease additional capacity was terminated.

In a recent note to investors, OpenAI emphasized that it has begun providing its own compute capabilities earlier than Anthropic and considers this a competitive advantage.