Meta to Begin 'Iris' AI Chip Production in September, Memo Shows

Meta Platforms is set to begin manufacturing its own artificial intelligence chip in September, according to an internal memo reviewed by Reuters. The move is a key part of the company's strategy to significantly boost its computing power and gain more control over its AI hardware supply chain.

Strategic Shift to In-House Silicon

The custom chip, code-named "Iris," is part of a multi-generational project called the Meta Training and Inference Accelerators (MTIA). The initiative aims to develop in-house silicon specifically tailored to power the AI features on its Facebook and Instagram platforms. According to the memo, testing for the chip was completed in just six weeks without any major issues, signaling positive momentum for the long-running project.

Meta is collaborating with Broadcom on the chip's design and with Taiwan Semiconductor Manufacturing Co. (TSMC) for its production. By developing its own accelerators, Meta seeks to lower its massive computing costs and reduce its dependence on dominant chip suppliers like Nvidia and Advanced Micro Devices.

Doubling Down on AI Infrastructure

The push for custom chips supports an aggressive expansion of Meta's data center capabilities. The memo outlined plans to deploy 7 gigawatts of computing infrastructure this year and to double that capacity to a total of 14 gigawatts in 2027.

Ad

To facilitate this build-out, Meta has reportedly secured multi-year supply agreements for critical components. These include deals with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment. This proactive supply chain management comes as the tech industry grapples with soaring demand and rising prices for AI-related hardware, a phenomenon some analysts have termed "chipflation."

Market Implications

Developing proprietary chips is a strategic imperative for tech giants, or "hyperscalers," aiming to lead in the AI race. "You can’t become an AI titan if you are dependent on another company for chips," said Mike Gualtieri, a vice president and principal analyst at Forrester, in a comment to Reuters. This approach allows for greater efficiency, lower long-term costs, and hardware optimized for specific software needs.

While the custom chips are designed to augment Meta's hardware, the company will continue to purchase large quantities of GPUs from Nvidia and AMD. The memo noted that adopting the latest external GPUs at Meta's scale "has been a heavy lift, and it has cost us time." Meta declined to comment on the contents of the memo.