Meta To Launch Custom AI Chip 'Iris' In September To Cut Nvidia, AMD Reliance

Meta is likely looking to cut down on its considerable computing costs and dependence on Nvidia and AMD.

Advertisement
Read Time: 2 mins
Meta is likely looking to cut down on its considerable computing costs and dependence on Nvidia and AMD.
Photo Source: Unsplash
Quick Read
Summary is AI-generated, newsroom-reviewed
  • Meta Platforms will start producing the AI chip Irisi in September to boost computing power to 14 GW
  • The company aims to reduce costs and reliance on Nvidia and AMD chip suppliers with in-house chip design
  • Irisi is part of Meta's four-generation MTIA project to enhance AI capabilities for Facebook and Instagram
Did our AI summary help?
Let us know.

Meta Platforms wil commence the manufacturing of a specialised AI chip known as 'Irisi" in September to double its computing capacity to 14 GW in the coming year, according to a memo cited by reports on Thursday.

This move is likely in order for the company to cut down on its considerable computing costs, as well as the reduction of its dependence on chip suppliers such as Nvidia and AMD.

Advertisement

The tech giant plans to create it's own custom-built AI chip to that it will design in-house to build up its AI capabilities and run its Facebook and Instagram social media platforms. 'Iris' is a part of the company's four-generation project for Meta Training and Inference Accelerators (MTIA).

The firm took six weeks to test the chip and did not notice any prominent issues, as per the memo cited by Reuters. Meta has partnered with Broadcom, the US-based semiconductor maker to designt the AI chip, along with enlisting Taiwan Semiconductor Manufacturing Co. to manufacture it.

Advertisement

Meta plans to use Iris to amplify the capabilities of its vast lineup of Graphics Processing Units (GPUs) that utilised for AI applications that the company buys from Nvidia and AMD.

ALSO READ: Meta Rolls Out Privacy Update That Disables Smart Glasses Camera If LED Is Blocked

The company described the process of implementing the latest GPUs as " a heavy lift that has cost us time" in its memo as cited by Reuters, due to its notable size. 

Advertisement

The company first announced Iris along with three more AI processors in March. The firm has set a goal to launch an AI chip every six months across 2027, in comparison to its peers, who release AI chips once a year.

Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.


Loading...