In more good news for Amazon, Snowflake signs $6B deal with AWS for AI CPU chips
Snowflake signed a $6 billion, five-year AWS agreement as demand for AI computing keeps rising, with Graviton chips central to the deal.
Intelligence analysis by GPT-5.4 Mini
Snowflake said its customers are spending more on AWS, and it is locking in a five-year, $6 billion agreement tied in part to Amazon’s Graviton CPU chips. The deal underscores how AI workloads are pushing cloud providers to compete harder on price, capacity, and custom silicon.
Snowflake stores a lot of company data, and many businesses now want that data to help power AI tools. To do that, Snowflake is making a very big five-year deal with Amazon’s cloud service.
Think of it like a school buying a huge supply of pencils for the next few years. The school still uses other tools too, but these pencils are an important part of how the work gets done. Here, the “pencils” are computer chips that help run AI jobs.
The story matters because AI is making cloud companies race to offer faster and cheaper computers. Snowflake is spending more, Amazon is trying to win more business, and both are betting that more AI use will keep growing.
Analysis
Snowflake’s new five-year agreement with AWS is large even by cloud-industry standards: the companies say it is worth $6 billion. The deal matters because Snowflake has already run on AWS for years, and this contract deepens that relationship at a time when its customers are spending more on cloud compute.
The article says Snowflake’s AWS-based customer spending doubled in 2025 to $2 billion for that year. Snowflake links that growth to AI adoption, especially through Cortex AI, its tool for building AI features on top of enterprise data. That fits the broader pattern in which companies are trying to turn the data they already store into chat-style interfaces, summaries, and agent-driven workflows.
A notable detail is that Snowflake is buying more access to AWS’s Graviton chips, Amazon’s in-house ARM-based CPUs. The article argues that as AI shifts from training to everyday use and automation, CPU demand rises alongside GPU demand, especially for agents and other support tasks. That makes custom CPUs a strategic lever for cloud providers: they can lower costs, improve margins, and try to lock in big customers.
The piece places this in a wider competitive fight. AWS has also sold Graviton chips to Meta, while Google and Microsoft have their own chip efforts. Nvidia still dominates much of AI infrastructure, but the article suggests cloud providers are using cheaper homegrown chips to win large deals and capture more of the AI spending wave.
Key points
- Snowflake signed a $6 billion, five-year agreement with AWS.
- Snowflake says customer spending on AWS doubled in 2025 to $2 billion.
- The deal includes more access to Amazon’s Graviton ARM-based CPU chips.
- The article frames the agreement as part of the broader AI infrastructure race.
- Cloud providers are using custom chips to compete on cost and capacity.
