Meta plans to bring AI chip into production in September: report

A worker stands at the Meta Lab in Los Angeles, California, USA on May 20, 2026.

Daniel Cole | Reuters

Metaplatforms The company plans to start making an artificial intelligence chip in September to boost total computing power to 14 gigawatts next year, according to an internal memo reviewed by Reuters.

The tech company’s data center chip, codenamed “Iris,” is part of a four-generation Meta Training and Inference Accelerators (MTIA) project that the tech company will develop internally. The plan is to use custom silicon to improve the AI ​​that powers social media platforms Facebook and Instagram.

Testing the chip took just six weeks and revealed no major problems, the memo shows. This relatively rapid progress signals positive momentum for an internal endeavor that has stalled since its launch more than half a decade ago.

Meta has tailored the chip to its own needs, working with Broadcom AVGO.O to help develop it and Taiwan Semiconductor Manufacturing Co 2330.TW to manufacture it. The approach is expected to help the company reduce its huge computing costs and increase independence from chip suppliers such as. B. to obtain Nvidia And Advanced micro devices.

The completion of bug testing and production timing have not yet been reported. Meta declined to comment.

The chip is intended to expand the large amounts of graphics processing units (GPUs) used for AI applications that Meta is buying from Nvidia and AMD.

However, introducing the latest GPUs at a company as large as Meta “was a difficult task and cost us time,” the memo said.

Meta introduced Iris under its technical name in March along with three other AI processors. It plans to release a chip about every six months through 2027, while companies typically release AI chips at intervals of a year or more.

Seven gigawatts of computing power in 2026

According to the memo, Meta plans to deploy seven gigawatts of computing infrastructure this year. The plan is to double that number in 2027, the memo said.

The company expects to spend up to $145 billion on AI infrastructure this year, a significant portion of Big Tech’s forecast spending of more than $700 billion on the technology.

To expand its computing infrastructure, Meta has secured long-term, multi-year supply contracts, the memo said. These include agreements with Samsung Electronics for memory chips, Sandisk for flash memory and Sumitomo Electric for fiber optic devices.

Such long-term agreements are important given the memory chip shortage affecting companies such as B. has become crucial to data center expansion goals Apple to increase prices.

Sandisk declined to comment. Samsung Electronics and Sumitomo Electric did not respond to requests for comment.

Demand for components like memory and AI chips has surged as tech companies look to expand their data centers to keep up with AI’s hunger for computing power.

Prices of memory and other chips have risen quickly and sharply enough that “chipflation” has become a macroeconomic problem, analysts at Morgan Stanley said.

Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.

https://www.cnbc.com/2026/07/09/meta-to-put-ai-chip-into-production-in-september-report.html