Cloud data storage giant Snowflake has solidified its foundational relationship with Amazon Web Services, announcing a new and substantial $6 billion five-year agreement. This massive commitment not only dwarfs previous spending but also signals a strategic pivot towards leveraging AWS’s unique offerings, particularly its homegrown AI-optimized chips, to power the next generation of enterprise AI applications.

The Landmark Agreement: A Foundation Reinforced

Snowflake, a company that has exclusively relied on AWS since its inception, albeit now also operating on Microsoft Azure and Google Cloud for multi-cloud strategies, is deepening its roots with Amazon. This new $6 billion contract represents a monumental increase in its infrastructure investment. To put this figure into perspective, AWS states that Snowflake has sold approximately $7 billion worth of its services via the AWS Marketplace since its founding in 2012. The new agreement alone is nearly equivalent to all the revenue Snowflake has generated through AWS over its entire history, underscoring the immense scale of this renewed commitment.

This substantial financial pledge is not merely a renewal but a dramatic acceleration of Snowflake’s cloud consumption. The company reports that its customers are significantly boosting their spending on AWS, projecting an astounding doubling of expenditure in 2025 to hit $2 billion for that calendar year alone. Such growth figures are atypical for mature cloud relationships and point to a powerful underlying force driving this expansion.

AI: The Catalyst for Unprecedented Growth

The primary driver behind this escalating investment is, unequivocally, artificial intelligence. Snowflake, a critical platform for enterprise data, is strategically positioning itself at the heart of the AI revolution. For several years, it has offered Cortex AI, a suite of AI building tools designed to unlock the value residing within its customers’ vast datasets. This makes intuitive sense: if an enterprise’s most crucial data lives on Snowflake, then the most effective AI tools should be integrated directly into that environment.

Cortex AI provides features that are becoming indispensable for modern businesses, such as natural language interfaces for database queries, enabling users to “ask” for data insights in plain English rather than complex SQL. It also facilitates automated summary reports and other intelligent automations, transforming raw data into actionable intelligence. As enterprises move beyond experimental AI projects to integrate AI into core operations, the demand for robust, scalable platforms like Snowflake, deeply integrated with powerful cloud infrastructure, intensifies exponentially.

Graviton’s Ascent: AWS’s Strategic Chip Play

A particularly noteworthy aspect of this multi-billion-dollar deal is Snowflake’s commitment to gain more extensive access to AWS’s homegrown ARM-based CPU chip, Graviton. This detail highlights a broader trend in the AI industry: while GPUs (Graphics Processing Units) are synonymous with the intensive training phases of large language models, CPUs play an increasingly vital role in the operationalization of AI.

As AI applications transition from initial training to daily usage, continuous inference, and sophisticated automation via AI agents, the demand for CPU compute skyrockets. CPUs handle the vast majority of tasks associated with managing, orchestrating, and serving AI models in production environments, making them critical for cost-effective scaling. Amazon CEO Andy Jassy proudly stated last month that Amazon’s proprietary AI chips offer “better price-performance” compared to many of Nvidia’s offerings, even as AWS continues to deploy Nvidia’s GPUs for specialized workloads. This cost efficiency is a significant advantage that AWS says it passes directly to its customers, making Graviton an attractive option for large-scale, sustained AI operations.

For cloud providers, developing their own silicon, like Graviton, offers multiple strategic benefits: greater control over the supply chain, optimized performance for their specific cloud architecture, and a substantial reduction in reliance on external chip manufacturers. This translates directly into cost savings for AWS, which are then leveraged to attract and retain major clients like Snowflake by offering compelling price-performance ratios.

The Cloud Giants’ Chip Challenge to Nvidia

The Snowflake deal is not an isolated incident but rather a testament to the growing success and strategic importance of AWS’s custom silicon initiatives. Last month, AWS secured another massive agreement to provide millions of Graviton chips to Meta for its burgeoning AI compute needs. This was a particularly significant win for AWS, especially given that Meta had only months prior signed a substantial $10 billion deal with Google Cloud, demonstrating the fierce competition and strategic shifting occurring in the cloud AI infrastructure market.

These multi-billion-dollar commitments serve as a clear signal to the market, and specifically to Nvidia, that competitive CPUs from the major cloud providers are aggressively entering the AI compute arena. Google has been developing its own Tensor Processing Units (TPUs) for years, optimizing them for AI workloads, and Microsoft recently launched its own custom AI chip, Maia, in January. This concerted push by the cloud giants represents a significant evolution in the AI hardware landscape, moving beyond generic hardware to highly optimized, proprietary silicon designed to meet the specific demands of hyperscale AI.

Nvidia’s Counter-Offensive: Defending the Turf

Not surprisingly, Nvidia, the undisputed leader in AI GPUs, is not ceding ground easily. CEO Jensen Huang, fresh off another quarter of record-breaking financial results, declared last week that he is more than ready to defend and even expand his company’s market share. He unveiled Nvidia’s new AI-specific CPU, Vera, proclaiming it represents a “brand new” $200 billion market opportunity for Nvidia. Huang confidently stated that Nvidia has already secured $20 billion in sales for Vera, underscoring the company’s intent to remain a dominant force across the entire spectrum of AI compute, including CPU-centric applications.

While the cloud providers’ custom silicon efforts are formidable and are successfully attracting massive deals, Nvidia continues to innovate and capture new market segments. The competition is intensifying, but it also reflects the sheer scale of investment and innovation pouring into the AI infrastructure space. Both strategies – specialized cloud chips and highly optimized general-purpose AI chips – are thriving simultaneously.

The Interconnected AI Ecosystem

The Snowflake-AWS deal, powered by Graviton, is a powerful illustration of the current dynamics in the tech industry. It shows how AI is not just a feature but the fundamental engine driving cloud consumption and infrastructure development. Regardless of which specific companies ultimately capture the largest share of the rapidly expanding AI market – be it specialized chipmakers like Nvidia or the vertically integrated cloud giants – the cloud providers are unequivocally benefiting from this unprecedented wave of innovation and adoption. Their role as the foundational compute layer ensures they will continue to capture a significant portion of the value created by the AI revolution.