Cloud data storage company Snowflake has just signed a $5 billion deal with Amazon Web Services (AWS, Amazon's subsidiary), marking one of Amazon's largest AI infrastructure contracts in recent times.
According to announcements from both sides, Snowflake will continue to expand operations on the AWS platform to meet the increasing demand for artificial intelligence (AI) services.
While it now supports both Microsoft Azure and Google Cloud, AWS has remained Snowflake's core platform since the company's founding in 2012.
The scale of this agreement is particularly noteworthy. AWS said that the total revenue that Snowflake has generated through AWS Marketplace since 2012 is about 7 billion USD. This means the new contract is almost equivalent to the entire business value that Snowflake has built on AWS over the past decade.
Snowflake said that the sharp increase in customer demand for AI services is the main reason for this investment. In 2025 alone, Snowflake customers' spending on AWS doubled, reaching about 2 billion USD.
The main driving force for growth comes from Snowflake's Cortex AI platform. This is an AI toolkit that allows businesses to exploit data in natural language, create automatic summary reports and support building AI applications for daily work.
A noteworthy point in the agreement is that Snowflake wants to expand access to the Graviton chip line developed by AWS itself based on ARM architecture. This is the type of CPU that Amazon is promoting to serve the new AI wave.
Technology experts believe that when AI moves from the training stage to practical deployment and operation of "AI agents", the need for CPU processing increases very rapidly.
While the GPU is responsible for model training and inference, the CPU handles many background tasks such as data management, process automation, and AI system coordination.
Last month, Amazon CEO Andy Jassy announced that AI chips self-produced by Amazon have a price-to-performance ratio (the efficiency between cost and product processing power) much better than current solutions on the market.
Although AWS still uses Nvidia chips, Amazon is striving to reduce dependence on the world's largest GPU manufacturer.
Graviton's biggest advantage is lower costs. Amazon said they can transfer this savings to customers, helping businesses deploy AI at more competitive prices.
Not only Snowflake, many large technology corporations are also increasing the use of AWS chips. Last month, AWS was said to have signed an agreement to supply millions of Graviton chips to Meta to meet the increasing demand for AI computing. Not long ago, Meta also signed a $10 billion contract with Google Cloud.
