Meta to Begin AI Chip Production in September

Picture of Wird-e- Ali

Wird-e- Ali

Meta to Begin AI Chip Production in September

Meta Platforms plans to begin manufacturing its new in-house artificial intelligence chip, code-named Iris, in September as the company accelerates its AI ambitions and expands its computing infrastructure.

The chip is part of Meta’s Training and Inference Accelerator (MTIA) program, a four-generation initiative aimed at powering AI features across Facebook, Instagram, and the company’s other platforms. According to an internal memo reviewed by Reuters, Meta completed bug testing for Iris in just six weeks without identifying any major issues.

The successful testing marks a significant milestone for Meta’s custom chip program, which has faced several challenges since development began more than five years ago. The company declined to comment on the testing results or its production timeline.

Meta developed Iris in collaboration with Broadcom, while Taiwan Semiconductor Manufacturing Company (TSMC) will manufacture the processors. The custom chips are designed to complement—not replace—the Nvidia and AMD GPUs currently powering Meta’s AI infrastructure, helping reduce long-term computing costs and dependence on external suppliers.

The company plans to roll out four generations of AI processors over the next two years, introducing a new chip roughly every six months through 2027 to support recommendation systems, ranking algorithms, and generative AI applications.

Meta also announced plans to significantly expand its computing infrastructure, targeting 7 gigawatts of capacity by 2026 before doubling it to 14 gigawatts in 2027. The expansion is part of the company’s broader Meta Compute initiative, which oversees AI infrastructure, data centers, and supplier partnerships.

To support its AI roadmap, Meta expects to invest up to $145 billion in AI infrastructure in 2026. The company has also secured long-term agreements with Samsung Electronics, Sandisk, and Sumitomo Electric to ensure a stable supply of memory chips, storage, and fiber-optic equipment amid growing global demand for AI hardware.

Also read: PTCL Confirms WhatsApp Call Disruptions, Contacts Meta

Related News

Type to Search