Amazon (AMZN) Stock Upgraded to Overweight on AWS Growth Outlook

“Upgrading AMZN to OW (from EW) on greater conviction in AWS acceleration in ’26. Raising AWS growth to +22% from prior/consensus of +19%/+18%. We see Project Rainier, a significant compute capacity build w/ partner Anthropic, as the primary driver of acceleration, contributing 5%/4% to AWS growth in ’26/’27. See build in our companion cloud industry note. We view AWS revenue acceleration as the key to the reversal of share underperformance YTD (flat vs. NDQ +17%). Increasing our 2026 / 2027 AWS revenue estimates by 3% / 7% as we see Project Rainier capacity additions starting in early 2026 supporting incremental Anthropic compute. Expect Project Rainier Phase 1 (1.3GW) to come online in Jan ’26 w/ 6-month utilization ramp and Phase 2 to start coming online in 4Q26. We estimate Indiana campus at full capacity (2.2GW) contributes ~$14B annual revenues to AWS. We forecast Anthropic contributes 7 points to AWS 2026 revenue growth (vs 3 points in 2025). See AWS market share losses peaking in 2025 (-470bps y/y), improving in 2026 and beyond as AWS accelerates, alleviating competitive concerns. While share losses remain material, we take solace in stronger industry growth and rising AWS estimates. Project AWS share losses modestly improve to -420bps y/y in 2026, -260bps y/y in 2027 and -180bps y/y in 2028. By 2029, see AWS at 32% share, down from 47% in 2024. However, over that period, see cloud industry 31% CAGR to $870B from $230B in 2024. Risks to our call include execution of scaling Project Rainier, performance of Trainium chips, further deceleration of core (non-AI) workloads, and greater AI-related margin headwinds than anticipated. See operational execution of Project Rainier (and data center capacity more broadly) as key risk given ongoing supply constraints and uncertainty around large Trainium deployments. We assume AWS OI margin compression of 270bps / 180bps y/y in 2026 / 2027 tied to scaling AI deployments.”