Snowflake signs $6B AWS deal for AI CPU chips as cloud giants challenge Nvidia

Cloud data giant Snowflake has deepened its long-standing relationship with Amazon Web Services, signing a new five-year agreement worth $6 billion, the companies announced Wednesday. The deal grants Snowflake expanded access to AWS’s homegrown Graviton processors, a sign of how the shift from AI training to real-world usage is reshaping the cloud computing landscape.

A deal that nearly matches Snowflake’s entire AWS history

To put the scale of this agreement in perspective: since its founding in 2012, Snowflake has sold roughly $7 billion worth of its services through the AWS Marketplace. This single contract is close to that total, reflecting a surge in enterprise spending. Snowflake reports that its customers’ spending on AWS doubled in 2025 alone, reaching $2 billion for the calendar year.

Snowflake has historically run primarily on AWS, though it now also operates on Microsoft Azure and Google Cloud. The new deal underscores AWS’s continued dominance as Snowflake’s primary infrastructure partner, even as the broader market diversifies.

Why CPU chips matter more for AI now

The driving force behind this deal is the maturation of artificial intelligence. Snowflake’s Cortex AI tool, launched two years ago, allows enterprises to query their data using natural language, generate summary reports, and automate workflows. As AI moves from the training phase — dominated by expensive Nvidia GPUs — to everyday inference and agent-driven tasks, the demand for CPUs has skyrocketed.

AWS’s Graviton chips, based on ARM architecture, are designed specifically for these workloads. Amazon CEO Andy Jassy recently claimed that Amazon’s own AI chips offer “better price-performance” than Nvidia’s, though AWS continues to offer Nvidia hardware as well. The Graviton chips allow AWS to offer more cost-effective compute power, which the company says it passes on to customers.

Amazon’s chip strategy is attracting big names

This is not an isolated deal. Last month, AWS signed an agreement to supply millions of Graviton chips to Meta for its growing AI compute needs. That deal was particularly significant because Meta had previously committed $10 billion to Google Cloud. These wins signal that cloud providers’ custom silicon is becoming a serious competitive factor in the enterprise AI market.

Google has been developing its own AI chips for years, and Microsoft launched its Maia AI chip in January. The race to build affordable, high-performance AI infrastructure is intensifying, and cloud providers are leveraging their custom hardware to lock in long-term, high-value contracts.

Nvidia isn’t standing still

Nvidia CEO Jensen Huang, fresh off another record-breaking quarter, has made it clear that his company intends to defend its turf. Last week, he unveiled Nvidia’s new Vera CPU, calling it a “brand new” $200 billion market opportunity. He also disclosed that Nvidia has already sold $20 billion worth of the new chip. While Nvidia remains the dominant force in AI hardware, the rise of custom chips from cloud providers is creating a more competitive landscape.

What this means for enterprises

For businesses using Snowflake and AWS, this deal likely means continued access to powerful, cost-effective AI tools. The agreement signals that Snowflake expects its customers’ AI workloads to grow significantly, and that AWS is positioning its Graviton chips as a core part of that future. The broader takeaway is clear: the cloud providers are betting heavily on their own silicon, and they are winning multi-billion-dollar commitments that reflect the central role of AI in enterprise computing.

Conclusion

The Snowflake-AWS deal is more than a routine renewal. It is a strategic bet on the future of AI infrastructure, where CPUs — not just GPUs — will handle the bulk of everyday AI tasks. As cloud giants compete with Nvidia for this growing market, customers stand to benefit from more choice and potentially lower costs. The deal also reinforces AWS’s position as a leading provider of AI compute, even as rivals like Google and Microsoft invest heavily in their own custom chips.

FAQs

Q1: Why is Snowflake spending $6 billion on AWS chips?
Snowflake’s customers are rapidly increasing their AI workloads, particularly using the Cortex AI tool. The deal secures access to AWS’s Graviton CPUs, which are optimized for the inference and agent-based tasks that are becoming central to enterprise AI.

Q2: How does this deal affect Nvidia?
While Nvidia remains the dominant player in AI hardware, the Snowflake deal — along with similar agreements with Meta — shows that cloud providers’ custom chips are gaining traction. Nvidia has responded by launching its own Vera CPU, aiming to capture a new market segment.

Q3: What is the significance of AWS’s Graviton chip?
Graviton is an ARM-based CPU designed by Amazon. It offers a more cost-effective alternative to traditional x86 processors for many cloud workloads, especially AI inference. AWS claims it passes these savings on to customers, making it an attractive option for large-scale deployments.