Amazon's AI capacity crunch and performance issues pushed customers to rivals including Google

Core Insights - Amazon's cloud business, particularly AWS's Bedrock service, is facing significant capacity constraints that have led to lost revenue opportunities and customer migration to competitors like Google Cloud [2][3][5] Capacity Constraints - Bedrock has encountered "critical capacity constraints," resulting in customers like Epic Games and Vitol considering alternatives, leading to tens of millions in lost or delayed revenue [2][3][4] - The capacity issues have affected various industries, including finance, gaming, and tech, with companies like HelloFresh and Ryanair also impacted [12][13] Revenue Impact - The document revealed that delays in quota approvals have led to at least $52.6 million in projected sales being postponed, with specific projects like Epic Games' $10 million Fortnite project being shifted to Google Cloud [3][4] Infrastructure Expansion - Amazon CEO Andy Jassy emphasized the urgent need to ramp up cloud infrastructure, particularly for AI, stating that AWS has added over 3.8 gigawatts of power in the past year and plans to double its capacity again by 2027 [9][10] Customer Migration - Customers are migrating workloads away from Bedrock due to latency and feature parity issues, with companies like Thomson Reuters and Figma opting for Google Cloud or Anthropic's own platform [16][17] - Financial startup TainAI shifted 40% of its workloads from Bedrock to Google's Gemini Flash, saving $85,000 daily, highlighting the competitive pressure AWS faces [21] Competitive Landscape - Bedrock is losing ground to Google's Gemini models, which offer larger quota limits and better performance, raising concerns about AWS's cohesive product vision for AI inference [19][20][21] - The document warns that without a clear strategy, AWS risks missing out on lucrative opportunities in the AI market [22]