Nvidia's Demand Signal Holds Even as Microsoft Trims Its Budget

Nvidia shares fell 2.8% on Tuesday, caught in a broad technology selloff rather than any company-specific disclosure. The pressure stemmed from a market-wide reappraisal of whether the enormous sums being poured into artificial intelligence infrastructure will ever generate adequate returns. Amazon's heavy cash burn and Intel's weaker-than-expected outlook, which carried warnings about PC demand, deepened the gloom hanging over semiconductor stocks.

That distinction matters when sizing up Nvidia's move. The decline tracked an industry-wide retreat, so it cannot be read as a verdict on any single corporate announcement. What investors are questioning is the economics of the entire investment cycle, not the existence of Nvidia's expansion plans.

Expansion Plans Keep Rolling In

The company's own pipeline shows no sign of slowing. On Tuesday, Nvidia and Oracle unveiled plans to scale Oracle Cloud Infrastructure's superclusters to multiple gigawatts using NVIDIA Vera Rubin NVL72 systems, with NVIDIA Vera CPUs also set to enter OCI's infrastructure. A day earlier, on October 21, Nvidia announced agreements to expand substrate, advanced chip packaging, semiconductor testing and optical component capacity, with substrate capacity set to grow 2.5-fold within three years. The company also disclosed a $1.1 billion advance payment to reserve capacity at KYEC, funding new test facilities and equipment in the United States.

Those commitments create the conditions for additional output. They also explain why the profitability debate matters so much for Nvidia: the build-out has to justify itself economically. Tuesday's selling does not call the expansion plans into question — it challenges the earnings expectations investors attach to them.

Microsoft's Budget Mix Tells a Subtler Story

A closer look at Microsoft's spending plans offers a counterweight to the gloom. The software giant now expects roughly $170 billion in capital expenditures for calendar 2026, down from an earlier $175 billion projection. CFO Amy Hood attributed the revision to delays in some capacity deliveries for GPU compute services — a timing issue in the build-out, not a pullback driven by weak demand.

The composition of the spending is what stands out. About 75% of first-quarter investments in fiscal 2027 went to short-lived assets, chiefly CPUs and GPUs. Total first-quarter capex came in at $45.2 billion, and Microsoft expects more than $50 billion in the second quarter, with CPU and GPU spending set to accelerate in the second half of fiscal 2027.

Yet processor budgets are not the same thing as Nvidia orders. Microsoft's figures lump CPUs and GPUs together and disclose no amount attributable to Nvidia. They reinforce the demand picture without allowing any direct revenue calculation for the chipmaker.

Concentration Cuts Both Ways

Data centers accounted for 92% of Nvidia's total revenue, giving the capex plans of the big cloud providers outsized influence over its fortunes. That exposure runs deeper than a single business segment: according to Nvidia's annual report, two direct customers represented 22% and 14% of revenue in fiscal 2026. Such concentration magnifies the weight of a handful of budget decisions, even when overall appetite for AI compute remains strong.

Amazon, for its part, is holding firm to infrastructure spending of about $220 billion for 2026, and CEO Andy Jassy has promised "significantly more" in 2027. Supportive as that sounds, Amazon is also developing its own Trainium chips, so rising infrastructure budgets do not automatically translate into matching orders for Nvidia.

The takeaway for investors is that headline capex totals tell only part of the story. Microsoft's spending mix suggests continued willingness to invest, but the timing of capacity build-outs and the split between CPUs, GPUs and in-house silicon will determine how much of that money actually reaches Nvidia's top line.