Amazon's AWS achieved its fastest growth in 18 quarters at 37% YoY and expanded operating margin to 39%, but the company is spending unprecedented capex—$54.208 billion in Q2 alone and roughly $200 billion for FY2026—to defend against cheaper alternatives like CoreWeave, which offers 40 to 60% cheaper compute using bare-metal, pure-NVIDIA clusters.
NVIDIA, meanwhile, collects revenue from all cloud providers regardless of which one wins frontier AI workloads, making it the more insulated play on AI infrastructure buildout.
What happened
Amazon reported AWS revenue growth of 37% YoY (its fastest in 18 quarters) and an operating margin of 39% (up 650 basis points YoY), while NVIDIA posted +85.2% revenue growth with Data Center revenue hitting $75.25 billion. However, Amazon spent $54.208 billion in capex in Q2 alone and is guiding to roughly $200 billion for FY2026, while specialized cloud providers like CoreWeave offer bare-metal, pure-NVIDIA clusters delivering 40 to 60% cheaper compute for frontier training.
Why it matters
AWS's highest-margin AI dollars are increasingly contested by CoreWeave and other neoclouds that route AI labs around AWS's general-purpose architecture. Amazon is spending unprecedented capex to defend market share in the segment where its legacy infrastructure is least efficient—exactly the ultra-dense training clusters where NVIDIA's networking stack performs best. By contrast, NVIDIA collects revenue from AWS, CoreWeave, Anthropic, and OpenAI simultaneously, insulating it from any single customer's choice of cloud provider.
What to watch
Amazon noted "multi-year, multi-gigawatt commitments" from Anthropic and OpenAI to its custom Trainium chips, signaling a genuine attempt to displace GPU spending. AWS free cash flow flipped to negative $7.6 billion TTM, making the capital intensity of defending its AI position the critical metric to monitor going forward.
Ask the AI about this article →
Amazon and NVIDIA both delivered blockbuster earnings, yet the results reveal starkly different economic exposure to the emerging AI infrastructure market. AWS grew 37% YoY and expanded operating margin to 39%—the latter up 650 basis points—driven partly by AI and chips businesses that each eclipsed run rates of more than $25 billion. The $496 billion AWS backlog is growing triple digits. However, the cost of defending this position is enormous: Amazon spent $54.208 billion in capex in Q2 alone and is guiding to roughly $200 billion for FY2026, causing free cash flow to flip negative $7.6 billion TTM.
The fundamental problem is architectural. CoreWeave and other specialized neoclouds exploit the friction inherent in AWS's general-purpose cloud design by offering bare-metal, pure-NVIDIA clusters delivering 40 to 60% cheaper compute for frontier training—exactly the workload where AWS's legacy infrastructure is least efficient. This is precisely where NVIDIA's networking stack (InfiniBand, Spectrum-X) shines and where ultra-dense training clusters cluster together. While Amazon is attempting to carve out custom silicon share through multi-year commitments from Anthropic and OpenAI to Trainium, Jassy conceded that AWS must also "continue making AWS the best place to run NVIDIA chips" because customers demand choice.
NVIDIA's position is fundamentally different. With Data Center revenue hitting $75.25 billion (up 92% YoY) and gross margin at 75%, NVIDIA collects a toll from AWS, CoreWeave, Anthropic, and OpenAI simultaneously—regardless of which cloud provider wins any given workload. This supplier economics insulation stands in sharp contrast to AWS's battleground position, where the highest-margin AI dollars are the ones most contested by neoclouds.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Astromech, an AI startup co-founded by Ben Lamm and geneticist George Church, raised $20 million in funding le…
Alibaba is guiding its AI cloud revenue toward a US$10 billion run-rate in the next quarter, signaling that it…
Ode, a venture arm of Anthropic, has acquired Casper Studios to expand its enterprise artificial intelligence…
SK hynix, the world's largest supplier of high-bandwidth memory (HBM), has published a technical roadmap for c…
Alibaba Group's June quarter results show cloud AI revenue growing 45%, while capital expenditure (investment…
Anthropic launched Claude Academy on August 20, a free learning site that explains AI fundamentals and how to…
