Snowflake Signs $6 Billion AWS Deal to Secure AI CPU Capacity
Snowflake has finalized a massive five-year, $6 billion agreement with Amazon Web Services to secure access to custom Graviton CPUs, highlighting the surging enterprise demand for AI inference and autonomous agent workloads that rely heavily on CPU processing.
A Landmark Infrastructure Agreement
Cloud data platform Snowflake has finalized a massive five-year, $6 billion agreement with Amazon Web Services (AWS) to secure dedicated compute capacity. The deal, announced on Wednesday, represents a significant commitment to AWS infrastructure, nearly matching the total $7 billion Snowflake has generated through the AWS Marketplace since its founding in 2012. The scale of the contract underscores a rapid acceleration in enterprise cloud spending, with Snowflake reporting that its AWS-related customer spend doubled to $2 billion in 2025 alone.
AI Agents Drive the CPU Renaissance
While the artificial intelligence boom has traditionally been synonymous with graphics processing units (GPUs), this agreement highlights a critical pivot toward central processing units (CPUs) for next-generation AI workloads. Snowflake is specifically securing access to AWS's custom ARM-based Graviton processors. As AI transitions from model training to daily operational usage and autonomous agent deployment, CPU demand is skyrocketing. While GPUs remain essential for heavy training and complex reasoning tasks, CPUs are increasingly handling the bulk of AI inference, data orchestration, and agent-driven automation.
Strategic Implications for the Chip Market
The move signals growing confidence in cloud providers' homegrown silicon. Amazon CEO Andy Jassy recently emphasized that AWS's proprietary AI chips deliver superior price-performance compared to third-party alternatives, though the company continues to integrate Nvidia hardware across its cloud. For Snowflake, leveraging Graviton aligns with its strategy to scale Cortex AI, its enterprise AI toolkit that enables natural language database querying and automated reporting. By locking in long-term CPU capacity, Snowflake is positioning itself to handle the computational demands of widespread AI agent adoption while optimizing infrastructure costs. The deal serves as a clear indicator that the future of enterprise AI will rely on a balanced, hybrid approach to compute, with custom CPUs playing an increasingly dominant role alongside traditional accelerators.