Meta AI chip production will begin in September, according to an internal company memo reviewed by Reuters. The Meta AI chip is part of the company's strategy to strengthen computing capacity and reduce dependence on external chip suppliers. The project supports artificial intelligence services across Facebook and Instagram.
The chip carries the codename Iris. It belongs to Meta's four-generation Meta Training and Inference Accelerators (MTIA) program. Engineers designed the processor internally for AI training and inference workloads. Broadcom assisted with the chip design, while Taiwan Semiconductor Manufacturing Co will manufacture it.
The memo stated that testing lasted only six weeks. Engineers reported no major issues during the evaluation process. The fast validation marks significant progress for the company's in-house semiconductor program, which started more than five years ago.
The Meta AI chip will complement graphics processing units purchased from Nvidia and Advanced Micro Devices instead of replacing them completely. Meta said adopting the latest GPUs across its large infrastructure has required substantial time and resources. Custom silicon aims to improve efficiency while lowering long-term computing costs.
Meta introduced Iris under its technical name in March alongside three additional AI processors. The company plans to release a new chip roughly every six months through 2027. Many semiconductor companies typically introduce new AI processors once each year.
The memo also outlined Meta's infrastructure roadmap. The company plans to deploy seven gigawatts of computing infrastructure during 2026. It expects to expand capacity to 14 gigawatts in 2027.
Meta expects AI infrastructure spending to reach up to $145 billion this year. Long-term agreements with Samsung Electronics, Sandisk, and Sumitomo Electric support this expansion. These partnerships help secure memory chips, flash storage, and fiber-optic equipment during ongoing supply shortages. Rising demand for AI hardware continues increasing component prices, making Meta AI chip development an important and promising step toward future computing growth.
