Amazon's commercialization of its in-house designed chips is accelerating. Cloud data company Snowflake has signed a five-year, $6 billion chip procurement agreement with Amazon Web Services (AWS), becoming a significant new client for AWS's Graviton processors. According to reports, Snowflake will pay $6 billion over the next five years to access AWS's Graviton CPUs within its data centers. The deal was officially announced, positioning Snowflake among AWS's largest CPU computing clients, alongside companies like Meta and Apple. This transaction not only propelled Snowflake's stock price to surge up to 35% in after-hours trading but also provides fresh market validation for Amazon's long-term strategy of betting on its own chip designs. The rapid adoption of AI agents is currently reshaping the demand structure for computing power. Autonomous AI agents require substantial CPU resources to coordinate and sequence computational tasks, driving a new wave of demand for central processing units. This trend is directly benefiting Amazon's Graviton chip business and validates Amazon's strategic positioning of Graviton as the "core computing power for the agentic AI era." Graviton Secures Major Client as CPU Demand Wave Builds The Graviton chip, launched by AWS in 2018, is a central processing unit designed for data center servers, catering to 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 increased significantly. Agents require the orchestration and sequencing of numerous processors to execute tasks, leading to sustained strong market demand for Graviton. Snowflake was founded on the AWS platform in 2015 and has since continued to deepen its partnership with Amazon. The company currently serves nearly 14,000 clients, including startups like Fetch and AI analytics platform Hex. This $6 billion long-term commitment marks an evolution of their relationship from platform dependency to a deep computing power alliance. This trend is also benefiting other CPU suppliers. Recent stock price gains and sales growth for companies like Intel, Advanced Micro Devices (AMD), and Arm Holdings are closely linked to the CPU demand expansion driven by AI agents. Dual Strategy: In-House Chips and NVIDIA Partnership Coexist The Snowflake deal is part of Amazon's broader chip strategy. Amazon's in-house chip business grew nearly 40% quarter-over-quarter in Q1, with an annualized revenue run-rate exceeding $20 billion. Its Trainium AI training chips have accumulated over $225 billion in revenue commitments, primarily from leading AI labs like Anthropic and OpenAI. Amazon stated that its "custom silicon business" is now among the world's top three data center chip businesses. Simultaneously, Amazon has not abandoned NVIDIA. CEO Andrew R. Jassy has explicitly stated that Amazon has no intention of moving away from NVIDIA hardware, anticipating that many enterprises will continue to choose NVIDIA chips, while another segment of customers may value Trainium's advantages in cost and performance. This dual-track strategy allows Amazon to reduce procurement costs and improve margins with its own chips while maintaining compatibility with the NVIDIA ecosystem and flexibility in customer choice. Notably, Amazon's Q1 capital expenditure reached $43.2 billion, primarily driven by AWS and generative AI infrastructure investments. Amazon believes data centers can operate for over 30 years, and servers and chips typically have a lifespan of five to six years, suggesting these investments will yield substantial long-term returns. AWS Growth Strong, AI Strategy Extends Beyond the Cloud The advancement of its chip strategy is injecting momentum into AWS's overall business. In Q1, AWS revenue grew 28% year-over-year to $37.6 billion, reaching an annualized run-rate of $150 billion. Customer spending on its Bedrock platform—which allows clients to access multiple foundation models to build AI applications—jumped 170% quarter-over-quarter, with the number of tokens processed in a single quarter surpassing the total from all previous years combined. OpenAI's models are now also available to customers through Bedrock. Amazon's AI initiatives extend beyond cloud computing. The company has deeply integrated AI into multiple business areas including warehouse automation, robotics, inventory forecasting, delivery optimization, advertising, customer recommendations, and supply chain management. Furthermore, its autonomous driving unit Zoox is set to partner with Uber, and its low-earth orbit satellite service is nearing commercialization, having secured commitments from entities like Delta Air Lines, AT&T, Vodafone, and NASA, and partnering with Apple to provide satellite connectivity for iPhone and Apple Watch. Financially, Amazon's Q1 total revenue grew 17% year-over-year to $181.5 billion, with adjusted earnings per share rising 75% to $2.80. Wall Street analysts project Amazon's earnings to grow approximately 21% in 2026 and another 14% in 2027. Among 57 analysts covering Amazon, 49 rate it a "Strong Buy," with an average price target of $315.67, implying roughly 16.6% upside from current levels.
Comments