1 The exterior of an AWS data center in Oregon, United States. 2 Inside an AWS data center. 3 AWS develops its own high-performance chips designed to reduce the time and expense of training generative AI models and uses them in its data centers. 4 AWS's new chip, the Graviton 3 processor. /Courtesy of AWS

Amazon Web Services (AWS) has added a major customer for its in-house central processing unit (CPU) "Graviton."

The Wall Street Journal (WSJ) reported on the 27th (local time) that cloud data platform corporations Snowflake signed a deal to use AWS's Graviton chips.

The contract is worth $6 billion over the next five years (about 9 trillion won). Snowflake, through this deal, will gain access to Graviton chips installed in AWS data centers. Launched on the AWS platform in 2015, Snowflake has risen to become one of the largest customers of AWS's in-house CPU.

Other major customers of Graviton include Meta and Apple. Earlier, Meta was reported last month to have decided to adopt tens of millions of AWS Graviton chips. CNBC reported that Meta became one of AWS's top five Graviton customers through the deal, with a term of at least three years.

The Graviton5 that Meta is adopting is built on a 3-nanometer process, delivering up to 25% higher performance than the previous generation and reportedly cutting energy use by 60% compared with existing compute options. AWS is developing Graviton as a custom CPU to improve compute performance and expense efficiency for cloud servers.

The deal also shows that demand for Generative AI and data analytics is fueling competition in cloud infrastructure. Reuters said Snowflake, through the partnership, will expand its use of not only Graviton but also AWS's AI infrastructure. Snowflake shares surged in after-hours trading following an upward revision to its outlook and the AWS deal announcement.

※ This article has been translated by AI. Share your feedback here.