NVIDIA Corporation

NVIDIA Corporation designs and sells the accelerated-computing platforms — data-center GPUs, CPUs, scale-up and scale-out networking, rack-scale systems, and the CUDA software stack — that a large share of the world's artificial-intelligence infrastructure runs on, and it now describes itself without qualification as a data-center-scale AI infrastructure company. In fiscal 2026 it recorded $215.9 billion of revenue and $120.1 billion of net income; in the single quarter ended July 26, 2026 it booked $96.2 billion of revenue, up 106% from a year earlier.

This is no longer a chip story. Since the Hopper generation the company has sold an entire "AI factory" — the GPU, the CPU, the interconnects, the systems, and the software that binds them — and the revenue it captures per gigawatt of data-center capacity has risen from roughly $18 billion with Hopper to $25 billion with Grace Blackwell to about $40 billion with the Vera Rubin generation that began shipping in the current quarter. That shift is the engine of the results above, and it is also why the company's fortunes are now tied to a supply chain and a customer base it only partly controls.

The file turns on one question: whether the demand signal management describes — customer forecasts that imply roughly 100% growth — is a durable computing shift, or the largest concentration of capital-expenditure risk the equity market has ever underwritten. NVIDIA guides to about 70% revenue growth for fiscal 2028 and is explicit that the number is a supply constraint rather than a demand one. Everything below — the debates, the segment economics, the balance sheet — is an attempt to judge which of those two readings the evidence supports.