Snowflake rockets 36% on earnings beat and plan to spend $6 billion on Amazon cloud

Snowflake rockets 36% on earnings beat and plan to spend $6 billion on Amazon cloud

On Wednesday, Amazon announced that its cloud division has secured a $6 billion spending commitment from Snowflake, which encompasses the utilization of the company’s custom silicon and chips specifically designed for artificial intelligence applications.

The allocation of services and technologies from Amazon Web Services (AWS) by Snowflake is scheduled to unfold over a five-year period, as detailed in a press release regarding the agreement. Snowflake aims to enhance its deployment of Amazon's Graviton general-purpose chips, alongside cloud-based graphics processing units (GPUs) tailored for AI tasks.

This development is a strong indication that AWS is gaining traction in the AI sector, as clients increasingly turn to the dominant cloud provider for sophisticated technological solutions. Notably, in April, Anthropic, the creator of Claude, announced plans to invest over $100 billion in AWS over the next decade, while Amazon also maintains a partnership with OpenAI.

Both agreements with these AI-focused companies involve an equity investment; however, the Snowflake agreement does not include this provision. Snowflake, which went public in 2020, currently boasts a market capitalization exceeding $60 billion and has consistently relied on AWS for its infrastructure needs.

Following the announcement of robust results for its fiscal first quarter ending April 30, Snowflake's shares experienced a remarkable surge of up to 36% in after-hours trading. The company reported adjusted earnings of 39 cents per share on revenues of $1.39 billion, reflecting a year-over-year growth of 33%. Analysts surveyed by LSEG had anticipated earnings of 32 cents per share and $1.32 billion in revenue.

Moreover, the guidance for the upcoming quarter was optimistic, with Snowflake projecting a fiscal second-quarter adjusted operating margin of 12.5%, and product revenue ranging from $1.415 billion to $1.420 billion. Analysts from StreetAccount had forecasted an adjusted margin of 11.9%, alongside expectations of $1.37 billion in product revenue.

Additionally, Snowflake confirmed its acquisition of AI startup Natoma for an undisclosed amount.

At the time of Snowflake's initial public offering, the company revealed a revised agreement with a yet-to-be-named cloud provider, outlining a spending plan of $1.2 billion over five years, with $350 million allocated for the final year. A spokesperson later confirmed to CNBC that the provider was indeed Amazon. By 2023, this agreement had increased to $2.5 billion.

The new $6 billion partnership proposes an average annual expenditure of $1.2 billion.

AWS introduced its first Arm-based Graviton chip in 2018, and it remains the company's most successful custom chip to date. Snowflake first expressed interest in integrating Graviton technology in 2022.

Furthermore, Snowflake has established a strong relationship with Nvidia, following a partnership announced in 2023. In November of that year, Snowflake highlighted new updates aimed at simplifying the execution of AI workloads on Nvidia GPUs.

This agreement exemplifies a growing trend among major technology firms to favor custom Arm-based processors over traditional x86 architectures.

Historically, server chips have predominantly been constructed on the x86 instruction sets originally developed by Intel in the 1970s and later by Advanced Micro Devices. Arm's power-efficient architecture gained mainstream adoption when Apple utilized it for the first iPhone in 2007. Amazon has since led the charge in integrating Arm chips into data centers through its Graviton family, with cloud rivals such as Google and Microsoft following suit.

As we look ahead to 2026, central processing units like Graviton are witnessing a resurgence in demand as the widespread adoption of AI evolves from basic query-and-response chatbots to more complex task-oriented agentic applications.

While GPUs, such as those developed by Nvidia, excel at training AI models due to their myriad of small cores designed for simultaneous operations, CPUs typically possess fewer, more powerful cores that handle sequential general-purpose tasks. Agentic AI necessitates substantial general compute power to manage extensive data flows for AI workflows across multiple agents.

In April, Meta announced plans to leverage hundreds of thousands of Graviton chips.

"Graviton is our industry-leading CPU chip, which enables Meta to execute CPU-intensive workloads essential for Agentic AI with the required performance and efficiency at scale," stated Amazon CEO Andy Jassy during the company's earnings call in April.