Roth/MKM reiterates Buy on Amazon stock, keeps $285 target

Amazon’s expanded Anthropic partnership adds $5 billion now and another $20 billion planned, with Anthropic committing to spend more than $100 billion over 10 years on AWS. Roth/MKM says AWS could add 2 gigawatts of capacity by year-end 2026, implying $12 billion to $14 billion in incremental annual recurring revenue and lifting Anthropic’s AWS spend to roughly $20 billion to $24 billion by end-2026. The stock has already risen 27% over the past 30 days, and multiple firms have raised price targets ahead of Amazon’s April 29 earnings report.

Analysis

The market is starting to price AWS less like a cyclical cloud vendor and more like a utility tied to frontier-model capex. That re-rating matters because it compresses the perceived downside of Amazon's multiple: if one anchor customer can underwrite multi-year capacity growth, the debate shifts from demand uncertainty to execution and power availability. The second-order winner is not just AMZN, but the semi chain, networking vendors, and datacenter infrastructure names that benefit from a larger and longer backlog tied to AI training/inference buildouts.

The key risk is that the positive narrative is too linear on timing. Revenue recognition from capacity ramps and partner commitments will lag the headline partnership by quarters, while the market is already discounting 2026 outcomes today. If AWS hits power, chip yield, or deployment bottlenecks, the stock can de-rate quickly because expectations have moved from incremental share gains to a near-monopoly on AI workload growth. That makes the next 1-2 earnings prints more important for operating margin and capex discipline than for top-line beat size.

Consensus seems to be underpricing the possibility that this is a capex-intensive win with lower near-term margin quality than the bulls imply. More AI infrastructure spend can lift AWS revenue but still pressure consolidated free cash flow if Amazon front-loads buildouts faster than utilization fills in. In that setup, the trade is not simply long AMZN beta; it is long the names that sell picks-and-shovels into the build cycle, and selectively short high-multiple software exposed to AI budget reallocation if customers keep shifting dollars from app-layer spend to infrastructure.

Key Decisions for Investors

• Stay long AMZN into the April 29 print, but fund it with a tighter risk budget: use a call spread or reduce delta if the stock gaps another 5-7% before earnings, since the setup is now crowded and forward expectations are extended.
• Add to a basket of AI infrastructure suppliers over 3-6 months: long ANET / MRVL / VRT on any post-earnings pullback, as AWS capacity expansion should translate into sustained orders and better backlog visibility.
• Pair trade: long AMZN vs short a high-multiple, AI-exposed software name with weaker monetization visibility over the next 6-12 months; the market is likely to reward direct monetization of AI compute before it rewards application-layer narrative risk.
• If AMZN trades above the implied post-earnings move into the print, consider a downside hedge via puts or put spreads, because the stock has already re-rated on the same AI catalyst and the bar for another leg higher is now execution, not story.