On May 27, cloud data storage giant Snowflake officially announced a five-year strategic partnership agreement worth 6 billion USD with AWS (Amazon Web Services). For these two tech giants, this is not only a long-term infrastructure supply contract but also a deep strategic integration around AI computing architecture.

One, the Shocking Numbers Behind: A Contract Worth 14 Years of Past Revenue

The 6 billion USD signing amount is highly impactful. Since its establishment in 2012, Snowflake has cumulatively sold cloud services worth approximately 7 billion USD through AWS Marketplace. This means that the new contract amount is nearly equivalent to 85% of Snowflake's total revenue from AWS over the past 14 years.

This growth is driven by enterprises' massive spending on AI: In 2025 alone, Snowflake customers' cloud consumption on AWS doubled, reaching 2 billion USD.

Two, Core Drivers: AI Shifts from "Training" to "Automation"

The growth engine behind the contract is Snowflake's core AI tool - Cortex AI. As a hub for enterprise data, Snowflake enables companies to query databases or generate analytical reports directly using natural language through Cortex.

However, as AI scenarios shift from simple "model training" to "daily applications" and "AI Agent automation," the model of computing power demand has undergone fundamental changes:

  • GPU handles training and inference

  • CPU handles large-scale agent logic and auxiliary tasks

As enterprise AI applications scale up, CPU load grows exponentially, directly increasing the demand for high-performance and cost-effective processors.

Three, Strategic Shift: How Graviton Chips Are Reshaping the Cloud Computing Competition

A key highlight of this agreement is that Snowflake will gain more access to AWS' self-developed ARM architecture CPU Graviton chips.

  • Excellent Cost-Performance Ratio: Amazon CEO Andy Jassy pointed out that their self-developed chips offer better price-performance ratios than general solutions in the market. By deploying large numbers of Graviton chips, AWS not only reduces its own operating costs but also attracts major clients like Snowflake through lower pricing.

  • Securing AI Computing Ground: Previously, AWS had supplied millions of Graviton chips to Meta. Meta’s and Snowflake’s choices indicate that cloud computing giants are breaking into the NVIDIA-dominated computing market through self-developed CPUs.

Four, Industry Signals: The "CPU War" Between NVIDIA and Cloud Giants

This trend undoubtedly puts unprecedented pressure on GPU leader NVIDIA. Although NVIDIA CEO Jensen Huang launched a dedicated AI CPU called "Vera" last week and claimed to have secured 20 billion USD in orders, trying to hold its ground, the offensive from cloud providers is already underway:

  • Google has long been developing its own AI chips (TPUs).

  • Microsoft launched its self-developed Maia AI chip in January this year.

  • AWS has rapidly captured large-scale cloud customers through Graviton via a "price war."

Industry Deep Analysis

The 6 billion USD alliance between Snowflake and AWS signals the focus of competition in the second half of the AI era: In the AI era, data processing is no longer just about "who has stronger GPUs," but also about "who can process large-scale reasoning and automation tasks at a lower cost."

For Snowflake, by binding with AWS Graviton, enterprises can significantly optimize the daily operational costs of AI; for AWS, through such a large client as Snowflake, its self-developed chip ecosystem can accelerate its maturity. The outcome of this battle will determine who becomes the foundation of computing power in this AI wave.