Meta to Begin Manufacturing AI Chip 'Iris' in September, Targets Doubling Computing Capacity by 2027
Meta Platforms plans to begin manufacturing a custom artificial intelligence chip, code-named “Iris,” in September, according to an internal memo reviewed by Reuters. The chip is part of a four-generation project called Meta Training and Inference Accelerators (MTIA), designed in-house to power AI features on Facebook and Instagram.
The chip's testing phase took only six weeks and revealed no major issues, indicating a significant step forward for Meta's in-house chip development, which had faced challenges since its launch over five years ago. Meta is working with Broadcom for design assistance and Taiwan Semiconductor Manufacturing Co for production. This approach aims to reduce the company's reliance on external chip suppliers such as Nvidia and Advanced Micro Devices, potentially lowering computing costs.
The chip is intended to complement the large quantities of graphics processing units (GPUs) that Meta purchases from Nvidia and AMD for AI applications. The memo noted that adopting the latest GPUs at Meta's scale has been “a heavy lift” and has cost time. Analyst Mike Gualtieri of Forrester Research commented, “You can’t become an AI titan if you are dependent on another company for chips.”
Meta unveiled Iris under its technical name in March, along with three other AI processors. The company plans to release a new chip approximately every six months through 2027, a faster pace than the typical annual release cycle in the industry.
The memo also outlined Meta's computing infrastructure plans. The company aims to deploy seven gigawatts of computing capacity this year, adding one gigawatt in the first half and another 5.5 gigawatts by year-end. By 2027, Meta targets doubling that capacity to 14 gigawatts. One gigawatt of energy can power about 800,000 homes.
Meta expects to spend up to $145 billion on AI infrastructure this year, a significant portion of the more than $700 billion projected Big Tech expenditure on AI. To support expansion, Meta has secured long-term supply agreements with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment. These agreements are crucial amid a memory chip shortage that has raised prices for companies such as Apple.
Demand for chips used in AI and data centers has surged, leading to what Morgan Stanley analysts call “chipflation,” a macroeconomic concern as prices rise rapidly.
