Snowflake pledges $6 billion to AWS over the next five years, with a focus on Graviton chips.

      The five-year agreement is 2.4 times larger than Snowflake’s AWS contract from 2023 and coincides with a 38% surge in shares following a strong Q1 earnings report. The strategic importance lies in the Graviton component.

      Snowflake has established a five-year, $6 billion agreement with Amazon Web Services, which both organizations are presenting as the most significant expansion of their 11-year partnership so far. The deal, revealed on Tuesday, includes commitments to utilize AWS Graviton, the cloud provider’s custom Arm-based CPU line, and to enhance product integrations for what Snowflake now refers to as “agentic enterprise” workloads.

      The increase in Snowflake’s AWS expenditures provides context for the magnitude of this announcement. The company pledged $1.2 billion during its 2020 IPO, which increased to $2.5 billion with a renewal in 2023. The new $6 billion deal represents approximately five times the commitment from 2020 and 2.4 times that of 2023.

      This expansion aligns with Snowflake’s own growth trajectory: the company reported Q1 fiscal-2027 earnings on Wednesday that significantly exceeded predictions, leading to a roughly 38% rise in stock price due to the combined earnings and AWS deal news.

      The Graviton aspect is particularly noteworthy. AWS Graviton, now in its fourth generation, is Amazon’s proprietary Arm-server processor line, intended to replace x86 chips from Intel and AMD in AWS data centers, offering significantly enhanced price-performance.

      Snowflake's decision to scale its data-cloud workloads on Graviton is an important validation of the Arm-server concept that has been gradually altering cloud-infrastructure economics for five years. Additionally, it serves as a valuable data point for Amazon in light of the recent announcement that ByteDance is developing its own Arm and RISC-V CPUs to mitigate pricing pressures from Intel and AMD.

      The transition to custom Arm-server silicon, driven by hyperscalers, is now a prominent narrative in data-center CPUs. For AWS, the Snowflake deal arrives amid a series of substantial AI-infrastructure commitments. Anthropic has pledged significant AWS spending over multi-year terms; OpenAI signed a notable agreement competing with Microsoft Azure on AWS earlier this year; and Meta has been visibly expanding its AWS presence for inference workloads. Snowflake marks the latest and largest non-foundational model agreement in this series.

      AWS’s ability to accommodate an additional $6 billion of demand over five years, given its already stretched data-center pipeline, is a clear indicator of how swiftly Amazon is expanding its capacity.

      The strategic backdrop for Snowflake involves the agentic-AI concept that the company is betting on. Snowflake’s strategy, similar to that of other enterprise data platform vendors currently, is that AI agents will primarily function on the trusted and governed enterprise data already within customers’ cloud-warehouse environments, rather than relying on external training datasets.

      Achieving this future vision necessitates significantly more computational integration with the underlying cloud provider, including direct access to AWS’s native AI tools (such as Bedrock, SageMaker, and the Q assistant) and closer marketplace and go-to-market integration.

      One remaining question is the comparison to Databricks. Snowflake’s most prominent competitor is more closely associated with Azure (through its 2023 partnership with Microsoft) and has aggressively positioned itself as multi-cloud agnostic. Snowflake’s extensive commitment to AWS, particularly with the explicit focus on Graviton, reflects a different strategic approach: partnering with a larger hyperscaler, securing customer acquisition through AWS Marketplace, and accepting the inherent single-cloud bias that accompanies this choice.

      Whether this strategy will succeed compared to Databricks's diversified approach remains an open question for the coming years. AWS Marketplace sales for Snowflake increased to $2 billion in 2025, signaling that the integration is already demonstrating commercial success. Neither company has specified which Graviton generation Snowflake is committing to. Snowflake CEO Sridhar Ramaswamy noted that further details will be shared at the AWS re:Invent conference later this year.