What Snowflake’s $6B AWS Commitment Really Means
Snowflake’s USD 6 billion (approx. RM27.6 billion) AWS commitment is a long-term infrastructure deal that binds its data warehouse and AI workloads to Amazon’s cloud, highlighting how enterprise analytics platforms are converging with large-scale AI hardware investments. Snowflake will spend that sum over five years on Amazon’s Graviton CPUs and GPU-accelerated EC2 instances, its largest cloud commitment so far. Under CEO Sridhar Ramaswamy, the company is repositioning itself from a cloud data warehouse to what it calls a “platform for the AI era,” with Cortex AI sitting on top of governed data. The scale and duration of the Snowflake AWS investment show this is not a side project but a directional bet: AI-driven analytics and agent-style applications are expected to become core to how customers query, summarize, and act on enterprise data rather than optional add-ons.
Graviton CPUs and AI Accelerators: A New Data Warehouse Stack
At the center of the deal is a specific hardware mix: AWS Graviton CPUs paired with AI accelerators in GPU-accelerated EC2 instances. Snowflake has already shifted substantial compute from Intel and AMD chips to Arm-based Graviton, and will now deepen that move as part of its enterprise AI infrastructure. Amazon’s latest Graviton generation offers 192 Arm Neoverse V3 cores with 12 memory channels, which suits the CPU-heavy parts of AI-enhanced analytics such as query planning, orchestration, and the non-model code paths that surround inference. Models themselves will run on GPU instances that Snowflake will use for AI training and inference, likely Nvidia-based according to AWS’s description of “GPU-accelerated” instances. This split stack — Graviton for core data warehouse compute and GPUs for AI — indicates how AI accelerators and CPUs are becoming a combined baseline for AI accelerators data warehouse architectures.
From Add-On AI to AI-Native Data Platforms
The Snowflake AWS investment is not only about raw compute; it signals a shift in product design. Instead of treating AI as an external service that periodically taps data warehouses, Snowflake is folding generative and agentic capabilities into the platform itself. Cortex AI already supports text-to-SQL, summarization, sentiment analysis, and entity extraction directly on governed data, while Cortex Code acts as an AI coding agent. According to The New Stack, Snowflake and AWS are expanding their joint go-to-market via AWS Marketplace, where Snowflake has surpassed USD 7 billion (approx. RM32.2 billion) in lifetime sales. That commercial traction suggests customers want AI services tightly coupled to their existing data controls and governance. The enterprise AI infrastructure pattern emerging here is clear: data, security, and AI logic live together, with the cloud provider’s silicon roadmap quietly shaping what is technically and economically possible.
ROI, Risk, and the Future of Data Warehouse Competition
Snowflake is effectively wagering USD 1.2 billion (approx. RM5.5 billion) per year that AI-accelerated analytics will pay back in usage and revenue. Amazon reports that Snowflake’s AWS Marketplace sales exceeded USD 2 billion (approx. RM9.2 billion) in 2025 alone, and Snowflake’s stock jumped more than 30 percent in after-hours trading when the deal was announced, signaling market confidence that AI workloads can justify the spend. Strategically, the commitment could help Snowflake lower unit costs for its core warehouse jobs via Graviton CPUs cloud computing while channeling savings toward pricey GPU capacity. For rivals, this raises the bar: competing data platforms now need credible access to both cost-effective CPU fleets and scalable accelerators, plus an AI-native experience built into the warehouse, not bolted on. The race is shifting from who stores data best to who can turn that data into AI-driven decisions fastest and at scale.