Deployment of the servers, dubbed Santa Barbara, expected to start in Q4 2025
Meta has placed an order for next-generation ASIC-powered AI servers with Taiwanese manufacturer Quanta Computer.
According to a report from Economic Daily News (EDN), Meta has ordered up to 6,000 racks and plans to start deploying the AI servers, code-named Santa Barbara, by the end of 2025.
The Santa Barbara servers will replace Meta’s existing Minerva servers, with the report adding that the next-generation servers will have a TDP (Thermal Design Power) of more than 180kW and require highly customized cabinets, sidecars, and water-cooling components, all of which will be supplied by SynMing Electronics
The report further claims that Meta has partnered with Broadcom for the supply of the ASIC chips, while Quanta Computer is handling the final assembly of the servers. Citing “supply chain sources,” EDN said the social media firm has finalized design decisions related to two to three new ASIC AI servers, and will decide on the final design this quarter, before starting trial production in Q4 2025.
Since the start of the generative AI boom, Meta has sought to significantly expand its AI server infrastructure and in-house ASIC development. In February 2024, the company was seemingly on the hunt for ASIC engineers, posting job adverts seeking workers with expertise in architecture, design, and testing in Bangalore, India, and Sunnyvale, California.
Although the job adverts didn’t reference any specific project, Meta has its own custom AI chips in development. Dubbed the Meta Training and Inference Accelerator (MTIA), the chips have been in development since 2023, with the company hoping to deploy them in its own data centers in an effort to reduce its reliance on Nvidia.
In March 2025, South Korean chip startup FuriosaAI turned down an $800 million acquisition offer from Meta.
Last week, Meta posted its Q2 2025 results, which saw the company report revenue of $47.5 billion for the quarter, a 22 percent year-on-year increase. While profits increased 36 percent to $18.3bn for the three-month period ending June 30, Meta’s costs also rose, up 12 percent to $27bn, as a result of increased spending on data centers, servers, and individual AI researchers.
On the company’s earnings call after the results were posted, CFO Susan Li said servers would "ultimately be the biggest bulk of capex spend," adding that the company was still planning when to deploy capacity.
