.png){kind=logo}
Meta Platforms plans to begin production of a new artificial intelligence chip in September as it expands its in-house computing strategy. According to Reuters, the plan appeared in an internal company memo that outlined Meta’s next phase of AI infrastructure growth.
The chip, code-named Iris, forms part of Meta’s wider MTIA program. The company aims to use custom silicon to support AI tools across Facebook, Instagram, and other services while reducing reliance on outside chip suppliers.
According to the memo reviewed by Reuters, Iris completed bug testing in six weeks and showed no major issues. The short testing period marks a step forward for Meta’s chip program, which has faced earlier delays since its launch more than five years ago.
Meta designed the chip for its own AI workloads. Broadcom is helping with design work, while Taiwan Semiconductor Manufacturing Co is expected to handle production. The chip will not replace GPUs immediately. Instead, it will support the large volumes of Nvidia and AMD chips, which Meta already uses for AI training and inference.
The memo said adopting the latest GPUs at Meta’s scale ‘has been a heavy lift, and it has cost us time.’ The statement shows why the company is trying to build more control over its AI hardware stack. Meta declined to comment on the Reuters report.
Meta plans to deploy seven gigawatts of computing infrastructure this year, according to the internal memo. The company added one gigawatt in the first half of the year and expects to add another 5.5 gigawatts by year-end.
The company then plans to double total computing capacity to 14 gigawatts next year. One gigawatt can power about 800,000 homes, showing the scale of electricity needed for large AI data centers.
The expansion comes as major technology firms increase spending on AI infrastructure. Meta expects to spend up to $145 billion on AI infrastructure this year. That figure forms part of more than $700 billion in projected Big Tech spending on AI technology.
Meta’s custom chip effort aims to reduce the cost pressure tied to buying advanced AI chips from outside suppliers. Nvidia and AMD currently supply many of the GPUs used in Meta’s AI systems.
The company also wants faster hardware updates. Meta unveiled Iris under its technical name in March, along with three other AI processors. It plans to release a new chip about every six months through 2027. Most chipmakers usually follow a cycle of one year or more.
Mike Gualtieri, a vice president and principal analyst at Forrester, said, “You can’t become an AI titan if you are dependent on another company for chips.” He added that hyperscalers and SpaceX are also planning chips as model usage becomes more price-driven.
Also Read: Meta Says US States Seek $1.4 Trillion in Facebook and Instagram Youth Safety Case
Meta has secured multi-year supply agreements to support its data center buildout, the memo showed. These include deals with Samsung Electronics for memory chips, Sandisk for flash storage, and Sumitomo Electric for fiber-optic equipment.
Such agreements have become more important as AI demand strains the global chip supply chain. Memory chips, storage products, and AI processors are all seeing higher demand as companies build larger data centers.
Sandisk declined to comment, while Samsung Electronics and Sumitomo Electric did not respond to Reuters requests for comment. The supply deals come as memory shortages push prices higher across the technology sector.
Morgan Stanley analysts have described rising chip and memory prices as ‘chipflation,’ as higher component costs become a broader concern. Meanwhile, companies such as Apple have raised prices on some products amid pressure from memory supply constraints.
Meta shares fell after the Reuters report but later recovered. The stock traded up 4.6% in late afternoon trading after the company announced developer access to an AI coding model that competes with tools from OpenAI and Anthropic.
