In a bid to reduce its GPU costs amid an unprecedented component shortage, Meta is on track to begin manufacturing the latest versions of its AI-specific chip in September, Reuters reportedciting an internal memo.
At least one chip passed its testing phase in about six weeks, according to the memo. Meta is working with Broadcom on the chip design, however it will use Taiwan’s TSMC to manufacture them. It is also purchasing RAM from Samsung, storage from Sandisk, and fiber optic equipment from Sumitomo Electric, according to the report.
Goal detailed the four new chips, developed under its Meta Training and Inference Accelerator (MTIA) program, in March, some of which are currently in deployment or will be this year or next. The company is taking a modular approach to designing these chips, anticipating that its needs will change as AI evolves rapidly when the chips are in production.
“Each generation of MTIA builds on the previous one, uses modular chiplets, incorporates the latest insights into AI workloads and hardware technologies, and is deployed at a shorter cadence,” the company wrote at the time.
The chips are expected to help the company save on purchasing GPUs from chipmakers like Nvidia and AMD, although it still expects to spend big with those suppliers as well, Reuters reports. Meta intends to use MTIA chips to train models for its classification and recommendation algorithms, broader AI workloads, and targeted inferences for its applications. The social media company has been producing its own AI chips from 2023.
Meta has been spending heavily to secure enough computing power to power its various AI efforts. The company said this in April. wait capital expenditures between $125 billion and $145 billion this year, much of which will go toward its AI efforts.
The company has been closing data center and energy deals around the world, spending tens of billions to secure the computing capacity to train and deploy its new spark muse AI Model Series. It plans to deploy 7 gigawatts of computing this year and double that amount next year, according to Reuters, which cited the memo.
Also signed an agreement with ARM last year to secure computing for its recommendation systems, in addition to a multimillion-dollar deal with AMD for its Instinct GPUsand a Multimillion-dollar deal with Amazon to use the cloud giant’s local CPUs for AI-related needs.
Meta is not the only company trying to stem the tide of capital going to Nvidia. OpenAI last month sleepless an inference processor it is building with Broadcom, and Anthropic is it is said to be considering developing its own chips with Samsung. Amazon and Google Both develop their own chips for AI training and inference, and there is a multitude of startups building in space to meet growing demand.
Meta declined to comment.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
