Meta will use hundreds of thousands of AWS Graviton chips
Approximately 3.6 billion people use it Meta’s Apps and social networking company will operate every day 32 data centers Handling the load with the completion of new construction in Oklahoma. But this is not enough.
Amazon’s Meta has agreed to use Amazon’s general-purpose Graviton chips in a deal that will last at least three years, the cloud unit said Friday.
The arrangement shows that Meta is willing to splurge alongside its tech peers under CEO Mark Zuckerberg to meet high computing demand. Alphabet And Microsoft. In recent weeks, Meta signed a deal with CoreWeave worth a total of $48 billion and nebiusBoth rent access to the Nvidia graphics processing units, or GPUs, that run the AI models.
Amazon did not disclose the value of the Meta deal.
Meta balances infrastructure expansions with headcount reductions. On Thursday, the company announced plans to lay off approximately 8,000 employees, or 10% of its workforce.
Unlike Nvidia GPUs, ArmThe best cloud-based Graviton processors from Amazon Web Services can handle a wide range of computing tasks such as: Intel’s or AMDs Central processing units or CPUs. But Graviton can still be useful for AI workloads, especially for optimizations or post-training once models are trained with large amounts of data using large-scale computing clusters.
“Graviton is one of the most used platforms for pretraining by many base model companies, and Meta is now one of the newest platforms,” said Nafea Bshara, AWS vice president and distinguished engineer.
Bshara co-founded Annapurna Labs, a chip company that Amazon acquired in 2015. Since then, Amazon has developed special-purpose chips for training and running AI models, among other components. Graviton became a huge hit and was adopted Adobe, Apple And Snowflake. Earlier this week, Amazon-backed AI model builder Anthropic announced that it also plans to use Graviton processors.
AWS says Graviton provides best performance Using 60% less energy for the price of all computing options available through the EC2 computing service.
Meta has been using Graviton chips on a small scale and will now use hundreds of thousands of chips, making it one of the top five Graviton customers, Bshara said. He said the company has been renting Nvidia GPUs from AWS since 2017.
On Thursday, Intel CEO Lip-Bu Tan told analysts that demand for Xeon server chips exceeds supply.
“For the last few years, the story of high-performance computing has been almost exclusively about GPUs and other accelerators,” Tan said. “In recent months we have seen clear signs that the CPU is re-establishing itself as the indispensable foundation of the AI era.”
However, Bshara said Meta did not choose Graviton because other types of CPUs were not available.
“Expanding to Graviton allows us to run CPU-intensive workloads behind agency AI with the performance and efficiency we need,” Meta’s head of infrastructure Santosh Janardhan said in a statement.
WRISTWATCH: Amazon and Anthropic announce new deal to give Claude AI models up to 5 gigawatts of computing
