Meta to Begin Production of In-House AI Chip in September, Memo Shows

Meta Platforms is set to begin manufacturing its proprietary artificial intelligence chip in September, a move aimed at significantly boosting its computing power and reducing its dependence on third-party suppliers, according to an internal company memo reviewed by Reuters. The initiative is a critical part of the company's strategy to control the spiraling costs of AI infrastructure and enhance the AI that powers its social media platforms.

The 'Iris' Chip Initiative

The custom silicon, codenamed "Iris," is part of a multi-generation project for Meta's Training and Inference Accelerators (MTIA). The memo indicated that the chip completed a six-week testing phase with no major issues, signaling positive momentum for the long-running internal project. Meta is reportedly working with Broadcom on design and Taiwan Semiconductor Manufacturing Co. (TSMC) on manufacturing.

By developing its own chips, Meta aims to augment the vast quantities of GPUs it purchases from market leaders like Nvidia and AMD. This strategic shift is designed to lower computing costs and provide the company with greater control over its hardware, a crucial step for tech giants competing in the AI space. "You can’t become an AI titan if you are dependent on another company for chips," Mike Gualtieri, a vice president at research firm Forrester, told Reuters.

Aggressive Infrastructure Expansion

The move is part of a broader, aggressive expansion of Meta's AI infrastructure. According to the memo, the company's goals include:

Ad

• Deploying 7 gigawatts of computing infrastructure in the current year.
• Doubling its total capacity to 14 gigawatts by 2027.

To achieve this, Meta added 1 gigawatt of power in the first half of the year and plans to add another 2.5 gigawatts by year-end. The company also intends to launch a new AI chip roughly every six months through 2027, a much faster cadence than the typical industry cycle.

Market Context and Supply Chain

Meta's push for hardware independence comes as it projects spending up to $145 billion on AI infrastructure this year alone. The intense demand for AI components has led to what some Morgan Stanley analysts have termed "chipflation," driving up prices for memory and processors.

To mitigate supply chain risks, the memo noted that Meta has secured long-term, multi-year agreements with key suppliers. These include deals with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment, ensuring a stable supply of critical components for its data center expansion.