Snowflake rockets 36% on earnings beat, will spend $6B on Amazon cloud

Amazon It said Wednesday that its cloud division had received a $6 billion spending commitment. SnowflakeThis includes the use of the company’s proprietary silicon and chips for artificial intelligence.

According to the press release regarding the agreement, Snowflake’s acquisition of services and technology from Amazon Web Services will occur over five years. Snowflake plans to expand the use of Amazon’s Graviton general-purpose chips as well as cloud-based graphics processing units for AI.

It’s the latest sign that AWS is gaining momentum in the AI ​​space as customers turn to the market-leading cloud for more advanced technologies. In April, Claude creator Anthropic said it aims to spend over $100 billion on AWS within a decade. Amazon also has an agreement with OpenAI.

Both of its deals with AI modeling companies include equity investments, while the Snowflake deal does not. Snowflake, which went public in 2020, has a market capitalization of just over $60 billion and has long relied on AWS.

Snowflake shares rose as much as 36% in extended trading following the announcement of strength results For the first quarter of its fiscal year ending April 30, the company reported revenue of $1.39 billion, up 39 cents in adjusted earnings per share and 33% year over year. Analysts surveyed by LSEG expected revenue of 32 cents per share and $1.32 billion.

Guidance has also been strong, with Snowflake claiming product revenue of $1.415 billion to $1.420 billion in the second quarter with an adjusted operating margin of 12.5%. Analysts surveyed by StreetAccount were looking for margins of 11.9% on product revenue of $1.37 billion.

Snowflake also says it’s acquiring AI startup Natoma for an undisclosed amount.

During Snowflake’s IPO, it was announced that a modified agreement had been made with an unnamed cloud provider for $1.2 billion in spending over five years, of which $350 million would come in the final year. A Snowflake spokesperson told CNBC that the provider is Amazon. In 2023, the deal increased to $2.5 billion.

The new regulation worth $6 billion means an average annual expenditure of $1.2 billion.

AWS announced its first Arm-based Graviton chip in 2018, and it remains the company’s most successful custom chip to date. Snowflake first talked about adopting Graviton in 2022.

Snowflake also has close ties Nvidiaafter a partnership announced In 2023. Snowflake was launched in November updates To simplify the process of running AI workloads on Nvidia GPUs.

Deal marks another big tech company opting for proprietary products Arm-based processors compared to processors based on traditional x86 architecture.

For decades, server chips were built on the x86 instruction sets he pioneered. Intel in the 1970s and Advanced Micro Devices a few decades later. Arm’s alternative power-saving architecture, Apple It adopted it for the first iPhone in 2007. But it was Amazon that brought Arm chips to data centers with Graviton. Cloud competitors Google And Microsoft followed Amazon in rolling out custom Arm chips.

Fast forward to 2026, and central processing units like Graviton are facing a renewed wave of demand as mass adoption of artificial intelligence moves from call-and-response chatbots to task-focused agency applications.

While GPUs like Nvidia are excellent at training AI models because they have thousands of tiny cores narrowly focused on performing many operations at once, CPUs have fewer powerful cores that run sequential general-purpose tasks. Agency AI requires a lot of general computing power to move large amounts of data for AI workflows by orchestrating across multiple agents.

in April, Meta He said that hundreds of thousands of Graviton chips will be used.

“Graviton is our industry-leading CPU chip that allows Meta to run the CPU-intensive workloads behind Agentic AI with the performance and efficiency they need at scale,” Amazon CEO Andy Jassy said on the company’s April earnings call.

WRISTWATCH: How are Amazon chips powering Anthropic without Nvidia?