Amazon (NASDAQ: AMZN | AMZN Price Prediction) and NVIDIA (NASDAQ: NVDA) both delivered blockbuster earnings, yet the CoreWeave narrative is quietly reframing which one actually owns the AI economy. Amazon reported AWS growth of 37% YoY, its fastest in 18 quarters. NVIDIA posted +85.2% revenue growth. Same AI wave, very different exposure to the neocloud threat.
AWS Is Accelerating, But the Cost of Defending It Is Enormous
Andy Jassy leaned hard into the AI narrative, telling investors “AWS is booming… our AI and Chips businesses each eclipsed run rates of more than $25 billion.” The $496 billion AWS backlog is growing triple digits. AWS operating margin expanded to 39% (up 650 basis points YoY).
The problem is the price of admission. Amazon spent $54.208 billion in Q2 capex alone and is guiding to roughly $200 billion for FY2026. Free cash flow flipped to negative $7.6 billion TTM. That is the tax Amazon pays for retrofitting a general-purpose cloud into an AI factory, exactly the friction specialized neoclouds like CoreWeave exploit with bare-metal, pure-NVIDIA clusters delivering 40 to 60% cheaper compute for frontier training.
NVIDIA Wins Whether AWS or CoreWeave Serves the Workload
Jensen Huang framed it bluntly: “The buildout of AI factories, the largest infrastructure expansion in human history, is accelerating at extraordinary speed.” Data Center revenue hit $75.25 billion, up 92% YoY, with networking (InfiniBand, Spectrum-X) nearly tripling. Gross margin sits at 75%, operating margin at 60.4%. Those supplier economics no hyperscaler can match while funding concrete, power, and cooling (we profiled seven of the power, cooling, and networking suppliers behind that buildout in a free report you can grab here).
| Lens | Amazon (AWS) | NVIDIA |
| Core Bet | Trainium + Graviton to displace GPU spend | Blackwell, Vera Rubin, NVLink ecosystem |
| Q2 Capex vs. Free Cash Flow | $54.2B capex, FCF negative | $48.55B FCF, +85% |
| Key Vulnerability | CoreWeave and neoclouds routing AI labs around AWS | China data center revenue excluded from guide |
The Trainium Question Is Real, But So Is the Loophole
Jassy noted “multi-year, multi-gigawatt commitments” from Anthropic and OpenAI to Trainium. That is a genuine share grab in custom silicon. Yet he also conceded AWS will “continue making AWS the best place to run NVIDIA chips” because customers demand choice. Every hyperscaler still funnels dollars to Santa Clara. CoreWeave takes the workloads AWS’s legacy architecture handles least efficiently, exactly the ultra-dense training clusters where NVIDIA’s networking stack shines.
Why I Lean NVIDIA Over Amazon on This Quarter
The CoreWeave story makes NVIDIA the cleaner AI expression. Amazon is a fine business, and AWS at a $169 billion run rate is not going anywhere. But its highest-margin AI dollars are the ones most contested by neoclouds, and it is spending unprecedented capex to defend them. NVIDIA collects a toll from AWS, CoreWeave, Anthropic, and OpenAI simultaneously. For investors weighing defensive scale and retail cash flow optionality, Amazon retains a distinct profile. For those focused on the picks-and-shovels exposure that benefits regardless of which cloud wins the frontier lab, the supplier’s economics look more insulated than the battleground’s.
Contact [email protected] for any questions or corrections.