Amazon is 'last place' in the AI cloud race. Why that might not be a 'death sentence.' (www.businessinsider.com)

Bernstein analyst Mark Shmulik declared AWS “last place” in the AI cloud race, arguing Amazon trails Microsoft and Google on AI momentum due to slower revenue growth, limited GPU capacity, and the fact many AI startups are building on other clouds. He points to Microsoft’s early OpenAI tie-up and Google Cloud’s “full‑stack” approach—proprietary Gemini models plus TPU chips and fewer capacity bottlenecks—as reasons AWS has ceded positioning in the expensive, GPU‑hungry LLM era. That narrative helps explain why Amazon stock has lagged peers and why investors worry AWS could miss out on the large, infrastructure-driven wave around generative AI and inference workloads. Shmulik cautions, however, that being late isn’t fatal. He highlights several concrete signs AWS can recover: easing capacity constraints, higher developer engagement, its second‑best quarter ever for net new dollar growth, and the Anthropic partnership—an $8B investment tied to Project Rainier, an AI supercomputer built on Amazon’s custom chips that could shift inferencing away from Google. Bernstein models Project Rainier contributing up to ~2.6% of AWS revenue in 2026 and >4% in 2027, with AWS revenue growth forecast at 18% this year to $127B and 21% in 2026–27. The takeaway: AWS faces real technical and market disadvantages now, but strategic deals, custom silicon, and improving capacity give it plausible paths to regain competitiveness before re:Invent.