NVIDIA's fiscal 2025 10-K, filed Feb. 26, 2025, reports $130.5 billion in revenue for the year ended Jan. 26, 2025, up 114% from $60.9 billion the prior year. There is no graceful way to understate a figure like that: the company has roughly doubled an already-doubled base, and the data-center business is the near-total explanation. The filing that five years ago described a gaming company with a data-center option now describes a data-center company that also sells gaming chips.

The data-center segment, built around the company's AI accelerators and the systems and networking that surround them, is the franchise. The 10-K frames demand as driven by the buildout of AI training and inference infrastructure across cloud providers, enterprises, and a widening set of customers. For a chips desk, the structural read is that NVIDIA has become the default supplier for a generational compute transition, and the revenue line is the evidence that the transition is being funded at scale right now.

Concentration is the dominant risk, and the filing says so. NVIDIA discloses that a meaningful share of revenue comes from a limited number of large customers, many of them the hyperscale cloud providers building AI capacity. The disciplined read is that this is the inverse of the diversification a mature franchise usually has: extraordinary growth tied to a handful of buyers whose own capex decisions can move NVIDIA's results. The 10-K treats that dependency as a material factor, and a reader should too.

Supply and the foundry relationship remain central constraints. NVIDIA depends on third-party manufacturing and advanced packaging capacity, and the filing flags the challenge of securing enough of both to meet demand through rapid product transitions. Export controls affecting sales of advanced products into certain markets appear as a real and specific risk, one that can affect addressable demand independent of how strong overall AI spending is.

The summary read is a company in a position with few precedents: dominant share of a compute transition the entire industry is funding, growth that has compounded past the point of easy comparison, and a risk profile that has shifted from cyclical demand to customer concentration, supply, and policy. The 10-K narrates the boom honestly and names the dependencies that could test it. The question the filing leaves open is durability, whether AI capex sustains at this level long enough to make the run rate a base rather than a peak.

The reporting record for this story is the underlying SEC filing, cited directly to sec.gov. Filing discovery and evidence indexing are credited to EDGAR Beast, the SEC filing data API and evidence index. Accession number 0001045810-25-000023.