Securing a $6 billion deal with Snowflake, Amazon’s “AI chip strategy” begins to pay off

Amazon's commercialization of its self-developed chips is accelerating. Cloud storage company Snowflake has signed a five-year chip procurement agreement worth $6 billion with Amazon Web Services (AWS), becoming the latest heavyweight customer of AWS’s Graviton processors. According to the Wall Street Journal on Wednesday, Snowflake will pay $6 billion over the next five years to gain access to Graviton CPUs inside AWS data centers. The agreement was officially announced on Wednesday, making Snowflake one of AWS’s largest CPU computing customers, alongside Meta and Apple. This deal not only pushed Snowflake’s stock price up by as much as 35% after hours, but also provided fresh market endorsement for Amazon’s long-term strategy of betting on self-developed chips. Currently, the rapid adoption of AI agents is reshaping the structure of computing power demand. AI agents capable of autonomous task execution require a large number of CPUs to coordinate and sort computational tasks, fueling a new wave of demand for central processors. This trend is directly benefiting Amazon’s Graviton chip business and validates Amazon’s strategic positioning of Graviton as the “core computing power of the AI agent era.” Graviton Wins Major Clients, CPU Demand Surge Graviton chips were launched by AWS in 2018 as central processors for data center servers, covering a wide range of applications from personal devices to advanced AI systems. With the expansion of AI agent applications, the strategic importance of CPUs has significantly increased—agents need to schedule numerous processors to orchestrate and sort computational tasks, leading to sustained and strong market demand for Graviton. Snowflake was founded on the AWS platform in 2015 and has continually deepened its partnership with Amazon since. The company currently boasts nearly 14,000 clients, including rewards and advertising startup Fetch, as well as AI analytics platform Hex. This $6 billion long-term commitment marks an upgrade from platform reliance to deep computing power integration between the two companies. This trend also benefits other CPU suppliers. Intel, AMD, and Arm Holdings have all seen recent stock and sales growth closely related to expanding CPU demand driven by AI agents. Dual Strategy: Self-Developed Chips and NVIDIA Hardware Both Progress The implementation of Snowflake’s large deal is part of Amazon’s broader chip strategy. Amazon’s self-developed chip business grew nearly 40% quarter-on-quarter in Q1 2024, with annual revenue already exceeding $20 billion. Its Trainium AI training chip has accumulated over $225 billion in revenue commitments, mainly from leading AI labs like Anthropic and OpenAI. Amazon noted that its “custom chip business” is now among the world’s top three data center chip businesses. Meanwhile, Amazon hasn’t abandoned NVIDIA. CEO Andrew R. Jassy has made it clear that Amazon doesn’t intend to give up NVIDIA hardware, and expects many companies will continue to use NVIDIA chips, while others may prefer Trainium for its cost and performance advantages. This dual-track strategy allows Amazon to reduce procurement costs and boost profit margins via self-developed chips, while maintaining compatibility with the NVIDIA ecosystem for flexible customer options. Notably, Amazon’s capital expenditures reached $43.2 billion in Q1 2024, driven mainly by investments in AWS and generative AI infrastructure. Amazon believes that data centers can operate for over 30 years, and servers and chips usually have a service life of five to six years, making these investments likely to yield significant long-term returns. AWS Grows Robustly, AI Strategy Extends Beyond the Cloud The advancement of chip strategy is energizing AWS’s overall business. In Q1 2024, AWS’s revenue grew 28% year-on-year to $37.6 billion, with annualized revenue reaching $150 billion. Its Bedrock platform—which allows clients to use various foundational models to build AI applications—experienced a 170% sequential jump in customer spending, and the number of tokens processed in one quarter has surpassed the total of all previous years. OpenAI models are now available to clients through Bedrock. Amazon’s AI efforts are not limited to cloud computing. The company has deeply integrated AI into warehousing automation, robotics, inventory forecasting, delivery optimization, advertising, customer recommendations, and supply chain management. Its autonomous driving business Zoox is about to partner with Uber, and its low-orbit satellite service is nearing commercialization, with commitments from Delta Air Lines, AT&T, Vodafone, NASA, and a partnership with Apple to provide satellite connectivity for iPhone and Apple Watch. Financially, Amazon’s total Q1 revenue rose 17% year-on-year to $181.5 billion, with adjusted earnings per share up 75% to $2.80. Wall Street analysts expect Amazon’s earnings to grow about 21% in 2026 and 14% in 2027. Among the 57 analysts covering Amazon, 49 give a “strong buy” rating, with an average target price of $315.67, implying about 16.6% upside from current levels. Risk Warning and Disclaimer The market is risky and investment should be cautious. This article does not constitute individual investment advice nor does it take into account any user’s specific investment goals, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their particular situation. Invest at your own risk.