Meta is moving its custom AI chip "Iris" into production this September, an internal memo reviewed by Reuters reveals, part of an aggressive cadence that will see a new chip launch roughly every six months through 2027.
The cadence is unusually fast. Most chipmakers release new AI processors at intervals of a year or more. Meta's plan to nearly double that pace signals a company racing to reduce its dependence on Nvidia and AMD silicon. Testing for Iris took just six weeks and turned up no major issues, the memo showed. That marks a sharp turnaround for an in-house chip effort that has struggled since it launched more than five years ago.
The bug-testing completion and production timeline had not been previously reported. Meta declined to comment. Iris is part of a four-generation project called Meta Training and Inference Accelerators (MTIA), designed entirely in-house. Meta is working with Broadcom on the chip's design and Taiwan Semiconductor Manufacturing Co to produce it. The chip is meant to improve the massive quantities of GPUs Meta buys from Nvidia and AMD, not replace them entirely.
"Adopting the latest GPUs at a firm as large as Meta has been a heavy lift, and it has cost us time," the memo read.
Meta unveiled Iris under its technical name in March alongside three other AI processors. The company plans to deploy seven gigawatts of computing infrastructure this year and double that to 14 gigawatts in 2027.
The spending required is staggering. Meta expects to invest as much as $145 billion on AI infrastructure this year, a large slice of Big Tech's projected $700 billion-plus outlay on the technology.
To support that expansion, Meta has locked in long-term supply agreements with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment. Those deals have become critical as a memory chip shortage pushes companies like Apple to raise prices.
Memory and AI chip prices have risen so sharply that Morgan Stanley analysts have flagged "chipflation" as a macroeconomic concern.
