NVIDIA demand off the charts

NVIDIA is not the news by itself this week. The real news is that its biggest customers just showed their cards again. In the span of a few days, Microsoft, Alphabet, and Amazon all talked up AI infrastructure demand, bigger buildouts, and tighter capacity. That matters because NVIDIA’s next earnings report lands on May 20, and the market is trying to figure out whether Blackwell demand is still as crazy as bulls claim. Turns out the customer commentary says yes. (investor.nvidia.com) ### What actually changed? The change was not a surprise product launch from NVIDIA. It was a fresh round of hyperscaler disclosures. Microsoft said on April 29 that its AI business had passed a $37 billion annual revenue run rate. Alphabet said the same day that Google Cloud revenue jumped 63% and backlog nearly doubled quarter on quarter to more than $460 billion. Amazon h(investor.nvidia.com). Put together, that is a giant demand signal aimed straight at NVIDIA’s data-center business. (microsoft.com) ### Why does that point to NVIDIA? Because these companies are not talking about abstract “AI investment.” They are talking about the exact kind of compute buildout NVIDIA sells into — GPUs, networking, and full AI systems. AWS explicitly tied its expansion to NVIDIA GPUs, including Blackwell and Rubin. Google said NVIDIA GPUs remain a core part of its AI accelerator portfolio and that it(microsoft.com)es. That is not a one-quarter buying blip. That is road-map level commitment. (aws.amazon.com) ### Where does Blackwell fit in? Blackwell is the current bridge between today’s AI training boom and the next wave of inference-heavy spending. Jensen Huang said in NVIDIA’s February results that Grace Blackwell with NVLink is “the king of inference today” and framed agentic AI as the next demand inflection. Basically, NVIDIA is arguin(aws.amazon.com)at framing is even mostly right, Blackwell demand should stay strong well beyond an initial launch spike. (nvidianews.nvidia.com) ### Why are investors so focused on May 20? Because NVIDIA’s last reported quarter was already enormous — $68.1 billion in revenue and $62.3 billion from data center alone for fiscal Q4 2026. The question now is not whether demand exists. The question is whether NVIDIA can convert that demand into enough shipped systems, enough gross margin, and enough guidan(nvidianews.nvidia.com)investors want proof NVIDIA can keep delivering “here it is.” (nvidianews.nvidia.com) ### Is this just cloud spending hype? Not really. The scale is too specific. Microsoft’s cloud revenue hit $54.5 billion for the quarter, with commercial remaining performance obligation at $627 billion. Alphabet said first-quarter capex was $35.7 billion, with the overwhelming majority going into technical infrastructure for AI. AWS is talking about a millio(nvidianews.nvidia.com)ructure spending at the biggest tech companies on earth. (microsoft.com) ### So what’s the catch? Crowded expectations. NVIDIA can have demand “off the charts” and still disappoint the stock if supply ramps unevenly, margins wobble, or guidance fails to clear the whisper numbers investors have built for themselves. The other catch is that hyperscalers are also pushing custom silicon — especially Google with TPUs. But right now, even the companies building their own chips are still expanding NVIDIA deployments at the same time, not replacing them outright. (blog.google) ### Bottom line? This story is really about customer budgets. And those budgets still look enormous. Heading into May 20, the cleanest read is that NVIDIA remains the main arms dealer in an AI infrastructure race that its biggest customers are still accelerating. (investor.nvidia.com)