Snowflake rockets 35% on earnings beat and plan to spend $6 billion on Amazon cloud

Key Points:

• Snowflake is enhancing its investment in Amazon Web Services, which includes the acquisition of custom Arm-based Graviton chips and graphics processing units.
• Amazon disclosed that Snowflake has committed to a $6 billion investment in AWS over five years.
• Snowflake also released strong quarterly financial results and guidance.

Amazon announced on Wednesday that its cloud division has secured a $6 billion spending commitment from Snowflake, which will encompass the use of Amazon's custom silicon and chips targeted at artificial intelligence.

This partnership entails a five-year plan during which Snowflake will procure services and technology from Amazon Web Services (AWS). The company plans to amplify its utilization of Amazon's Graviton general-purpose chips and its cloud-oriented graphics processing units for AI applications.

This agreement underscores the growing momentum at AWS, as clients increasingly gravitate towards the leading cloud provider for advanced technology solutions. Notably, in April, Anthropic, the creator of Claude, expressed plans to invest over $100 billion in AWS over the next decade. Additionally, Amazon has established a partnership with OpenAI.

Both of these agreements with AI-focused companies include an equity investment, in contrast to the Snowflake deal, which does not include such provisions. Snowflake, having gone public in 2020, currently boasts a market capitalization exceeding $60 billion and has historically relied on the services of AWS.

Following the announcement of robust fiscal first-quarter results for the period ending April 30, Snowflake's shares surged by as much as 35% during after-hours trading. The company reported adjusted earnings per share of 39 cents and revenue of $1.39 billion, reflecting a 33% year-over-year increase. Analysts polled by LSEG had anticipated earnings of 32 cents per share and revenue of $1.32 billion.

Furthermore, Snowflake’s outlook is optimistic, projecting a 12.5% adjusted operating margin for the fiscal second quarter and estimating product revenue between $1.415 billion and $1.420 billion. Analysts consulted by StreetAccount had forecasted an operating margin of 11.9% along with product revenue of $1.37 billion.

In addition, Snowflake revealed its acquisition of AI startup Natoma, although financial terms for the transaction were not disclosed.

At the time of its IPO, Snowflake had indicated a modified agreement with an undisclosed cloud provider, committing to $1.2 billion in spending over five years, with $350 million allocated for the final year. This provider was later identified as Amazon, as confirmed by a Snowflake spokesperson. By 2023, this agreement had expanded to $2.5 billion, and this new $6 billion pact suggests an average annual expenditure of approximately $1.2 billion.

AWS first unveiled its Arm-based Graviton chip in 2018, marking it as the most successful custom chip developed by the company to date. Snowflake initially discussed adopting Graviton technology back in 2022.

Additionally, Snowflake maintains a collaborative relationship with Nvidia, stemming from a partnership established in 2023. In November of that year, the company highlighted improvements aimed at facilitating the execution of AI workloads on Nvidia’s graphics processing units.

This collaboration further illustrates the trend among large technology firms opting for custom Arm-based processors over traditional x86 architecture.

For many years, server chips have predominantly been based on x86 instruction sets, developed by Intel in the 1970s with contributions from Advanced Micro Devices in subsequent decades. The power-efficient Arm architecture gained prominence after being adopted by Apple for the first iPhone in 2007. However, it was Amazon that pioneered the integration of Arm chips into data centers with its Graviton series. Competitors like Google and Microsoft have since followed suit by introducing their own custom Arm chips.

As we move into 2026, central processing units like Graviton are experiencing renewed demand, driven by the widespread adoption of AI which is evolving from basic chatbot capabilities to more complex task-oriented applications.

While GPUs like those from Nvidia are particularly effective in training AI models due to their numerous small cores optimized for executing multiple operations simultaneously, CPUs, with their smaller number of powerful cores, are designed for sequential general-purpose tasks. The requirements of agentic AI demand substantial general compute power to manage extensive data flows across multiple agents.

In April, Meta indicated plans to leverage hundreds of thousands of Graviton chips for its operations.

"Graviton is our industry-leading CPU chip, which enables Meta to execute the CPU-intensive workloads associated with agentic AI while maintaining the necessary performance and efficiency at scale," remarked Amazon CEO Andy Jassy during the company's April earnings call.