Amazon said Wednesday that its cloud unit has secured a $6 billion spending commitment from Snowflake over five years, deepening a long-running relationship between the two companies and underscoring rising demand for Amazon Web Services’ custom chips and AI infrastructure. The agreement covers Snowflake’s use of AWS services and technology, including Amazon’s Graviton processors and cloud-based GPUs for artificial intelligence workloads.
The deal is the latest sign that AWS is gaining momentum in the AI race as customers seek more advanced computing tools. It follows other major commitments on Amazon’s cloud platform, including Anthropic’s plan to spend more than $100 billion on AWS over a decade and a separate deal with OpenAI. Unlike those agreements, the Snowflake arrangement does not include an equity investment from Amazon.
Snowflake, whose market capitalization is a little above $60 billion, has long depended on AWS and is expanding its use of Amazon’s chips as AI demand grows. The company said it expects to broaden adoption of Graviton, Amazon’s Arm-based general-purpose chip, alongside cloud GPUs for AI workloads. The new five-year agreement implies average annual spending of about $1.2 billion, up from earlier commitments that were smaller but still substantial.
Snowflake shares jumped as much as 30% in extended trading after the company reported stronger-than-expected fiscal first-quarter results. For the quarter ended April 30, Snowflake posted adjusted earnings of 39 cents per share on revenue of $1.39 billion, up 33% from a year earlier. Analysts surveyed by LSEG had expected 32 cents per share on $1.32 billion in revenue.
The company also issued upbeat guidance. Snowflake said it expects a fiscal second-quarter adjusted operating margin of 12.5% on product revenue of $1.415 billion to $1.420 billion. StreetAccount-compiled estimates had called for an 11.9% margin and $1.37 billion in product revenue. Snowflake also disclosed it is acquiring AI startup Natoma for an undisclosed amount.
The new AWS contract is larger than the cloud spending agreement Snowflake disclosed at the time of its 2020 initial public offering, when it said it had an amended arrangement with an unnamed cloud provider for $1.2 billion over five years. A Snowflake spokesperson later confirmed that the provider was Amazon. By 2023, that agreement had increased to $2.5 billion.
Amazon’s Graviton chip, first announced in 2018, has become the company’s most successful custom processor and a key part of its push to reduce dependence on traditional x86 server chips from Intel and AMD. Snowflake first discussed adopting Graviton in 2022 and has also maintained close ties with Nvidia, announcing a partnership in 2023 and later adding updates to simplify AI workloads on Nvidia GPUs.
The deal reflects a broader shift in enterprise computing as AI adoption moves beyond simple chatbots toward more complex agentic applications. While GPUs remain crucial for training AI models, CPUs such as Graviton are increasingly used for the general-purpose computing needed to manage data movement and orchestrate AI workflows. Amazon CEO Andy Jassy recently said Graviton is helping customers like Meta run CPU-intensive workloads for agentic AI with better performance and efficiency at scale.
