Meta reportedly plans to begin manufacturing an artificial intelligence chip in September as part of a broader effort to expand its computing infrastructure and reduce dependence on outside chip suppliers. According to Reuters, the company’s data center chip is code-named Iris and is part of Meta’s multi-generation Meta Training and Inference Accelerator initiative. The chip is being designed in-house to support AI workloads across Meta’s platforms, including Facebook and Instagram.
The report cited an internal memo showing that Meta is targeting 14 gigawatts of computing power next year. The company reportedly plans to deploy 7 gigawatts of computing infrastructure this year and double that capacity in 2027.
Meta’s Iris chip is intended to supplement the large quantities of graphics processing units the company buys from Nvidia and AMD for AI applications. The chip is not expected to replace GPUs entirely, but it could help Meta lower infrastructure costs and improve control over its AI hardware roadmap.
Reuters reported that Iris completed testing in six weeks without major issues. That progress could represent an important step for Meta’s internal chip program, which has faced challenges since launching more than five years ago.
Meta is reportedly working with Broadcom to help design the chip and with Taiwan Semiconductor Manufacturing Co. to manufacture it. The custom silicon strategy is part of a broader trend among large technology companies seeking greater independence from external AI chip suppliers.
The company has been investing heavily in AI infrastructure as it expands model training, inference, coding tools, and AI-powered product experiences. Reuters reported that Meta expects to spend as much as $145 billion on AI infrastructure this year.
The memo also reportedly noted that adopting the latest GPUs at Meta’s scale has been costly and time-consuming. Custom chips tailored to Meta’s own workloads could help the company optimize performance, manage costs, and scale AI services more efficiently.
Meta unveiled Iris under its technical name in March alongside three other AI processors. Reuters reported that Meta plans to launch a new chip about every six months through 2027, a faster cadence than the annual or longer release cycles typically seen in AI chip markets.
To support its infrastructure expansion, Meta has reportedly secured long-term supply agreements with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment.
These supply agreements reflect the broader demand surge across AI infrastructure components. Memory, storage, networking gear, and AI accelerators have become critical inputs as large technology companies race to build enough data center capacity for AI training and inference.
The reported production timeline for Iris also highlights the increasingly strategic role of custom silicon in the AI race. As compute needs rise, companies such as Meta are looking for ways to control more of the hardware stack behind their AI products.
