Meta to begin manufacturing 'Iris' AI chip in September as AI push gathers pace: Report

Meta is planning to begin manufacturing a new artificial intelligence chip from September as part of its broader strategy to expand its AI computing capacity to 14 gigawatts next year, according to an internal memo reviewed by Reuters.

The data centre chip, codenamed "Iris," is part of Meta's four-generation roadmap for in-house AI training and inference accelerators. The company aims to use custom-designed silicon to improve the AI systems powering its Facebook and Instagram platforms.

The memo also revealed that the new chip has been tested for only six weeks and has shown no major issues so far. The relatively rapid progress signals renewed momentum for Meta's in-house chip initiative, which has struggled to gain traction since it was launched more than five years ago.

More from Tech

Meta has designed the new chips to meet its own AI computing requirements, with Broadcom helping to develop the processors and Taiwan Semiconductor Manufacturing Co. (TSMC) handling production. The in-house chips are expected to reduce Meta's computing costs while lowering its reliance on external suppliers such as Nvidia and Advanced Micro Devices (AMD).

The chips will complement the large number of GPUs that Meta continues to purchase from Nvidia and AMD for AI workloads. Meta first unveiled the Iris chip under its technical name in March alongside three other AI processors. The company plans to introduce a new AI chip roughly every six months through 2027, a much faster cadence than the industry norm of one new chip every year or more.

Meta plans 7 gigawatts of computing capacity in 2026

According to the internal memo reviewed by Reuters, Meta plans to deploy 7 gigawatts of AI computing infrastructure in 2026 and double that capacity to 14 gigawatts in 2027.

The company is expected to spend as much as $145 billion on AI infrastructure this year, accounting for a significant share of Big Tech's projected spending of more than $700 billion on AI.

To support its infrastructure expansion, Meta has secured long-term, multi-year supply agreements with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fibre-optic equipment.

Such agreements have become increasingly important as technology companies race to expand data centres amid a global memory chip shortage. The supply crunch has already prompted companies such as Apple to raise prices.

Sandisk declined to comment on the agreements, while Samsung Electronics and Sumitomo Electric did not respond to Reuters' requests for comment.

Demand for key components, including memory and AI chips, has surged as companies rapidly expand data centre capacity to support increasingly compute-intensive AI workloads.

The sharp rise in memory and other chip prices has been significant enough for "chipflation" to emerge as a broader macroeconomic concern, according to Morgan Stanley analysts.