Meta Unveils "Iris" Custom AI Chip, Set for September Production

Meta plans to begin manufacturing its custom AI chip, codenamed Iris, from September as part of a strategy to boost overall computing power to 14 gigawatts next year, according to an internal memo reviewed by Reuters.

The chip cleared its bug-testing phase in roughly six weeks without turning up any significant problems, marking a notable milestone for Meta's in-house chip development efforts, which have faced challenges over the past several years.

Meta is designing the chip in collaboration with Broadcom, while Taiwan Semiconductor Manufacturing Co. will handle fabrication, aiming to reduce dependence on external suppliers such as Nvidia and AMD.

"You can't become an AI titan if you are dependent on another company for chips. The hyperscalers and even SpaceX all plan chips because it will be the only way to compete on price for model usage." – Mike Gualtieri, VP and Principal Analyst at Forrester

The Iris chip falls under Meta's MTIA program and is one of four planned generations, with the company planning to introduce a new AI chip roughly every six months through 2027 — a significantly faster cadence than the industry norm of annual releases.

Meta this year plans to deploy seven gigawatts of computing infrastructure, doubling to 14 GW in 2027, with projected AI infrastructure spending reaching as high as $145 billion.

As the custom AI silicon trend accelerates across hyperscalers, Meta's aggressive roadmap signals a broader industry shift toward vertical integration in AI infrastructure. While the company's massive capex commitments have drawn investor scrutiny, the Iris production timeline suggests that Meta is betting on long-term cost optimization and supply chain independence over short-term returns.