Nvidia's Worst Nightmare: Amazon's Secret Weapon Is Stealing Customers with Better Prices

Core Viewpoint - Amazon's shares fell after announcing a significant increase in capital expenditures to $200 billion, primarily focused on AI infrastructure [1] Group 1: Capital Expenditures and AI Investment - Amazon plans to ramp up capital expenditures to $200 billion this year, indicating a strong commitment to AI infrastructure [1] - CEO Andrew Jassy emphasized the company's strategy to monetize AI capacity as quickly as possible [1] Group 2: AI Chip Development and Performance - Amazon has installed 1.4 million Tranium2 AI chips in its data centers, generating an annual revenue run rate of $10 billion, with growth exceeding 100% annually [2] - The upcoming Tranium3 chip is expected to provide a 40% improvement in performance-per-dollar compared to Tranium2, with full sales expected by mid-2026 [6] - Amazon's custom Graviton CPU also contributes significantly to its business, offering up to 40% more performance per dollar than leading x86 CPUs, with 90% of AWS's top 1,000 customers utilizing it [7] Group 3: Competitive Landscape and Cost Advantages - Amazon's Tranium chips reportedly offer 30% to 40% better performance-per-dollar than comparable Nvidia GPUs, which are known for their high costs [4] - AI startup Anthropic is utilizing Amazon's Tranium2 chips for training its AI models, showcasing the practical application and effectiveness of Amazon's technology [5]