Nvidia Sees Chip Sales Doubling Next Year Despite Shortage

Nvidia told investors that customer demand for its artificial intelligence chips could double next year, even though current supply cannot keep pace and shortages are likely to persist through 2028. CEO Jensen Huang made the projection during the companys second-quarter fiscal 2027 earnings conference, framing a clear disconnect between what customers want and what the company can deliver in the near term.

In the most recent quarter Nvidia reported revenue of $96.22 billion, a 106 percent increase from the same period a year earlier. The company said the datacenter business was the primary growth engine, producing $89 billion in revenue for the quarter, up 117 percent year over year and accounting for most of the companys overall advance. For the current quarter Nvidia guided to about $108 billion in revenue, with a 2 percent range of variation. On an annual basis the company estimates roughly 70 percent growth for fiscal 2028, which ends in January 2028.

Customers, orders and the role of HBM

Nvidia says demand now comes from a broader set of buyers than the cloud giants. Alongside Amazon, Microsoft and Google, the company reports orders from emerging AI startups, national sovereign AI projects pursued by governments, and enterprise customers across industries. One concrete example is a deal with Amazon Web Services to deploy an additional 2 million graphics processing units across 2027 and 2028, a commitment Nvidia cites as evidence that demand is both large and contracted over multiple years.

At the same time Nvidia and its finance team warned that supply constraints will limit shipments. Chief financial officer Colette Kress said at the companys financial conference that bottlenecks, particularly in HBM memory, are expected to continue through the end of fiscal 2028. HBM stands for high-bandwidth memory, a type of fast memory used in advanced accelerator modules. Nvidia relies on partners such as SK hynix and Samsung to supply HBM. When demand outstrips memory production capacity, memory prices rise, gross margins face pressure and the number of fully built systems that can be shipped falls.

Executives framed the situation as part of a much larger wave of investment in AI infrastructure. Nvidia said its chips are central to many compute clusters, datacenters and large language model projects, and that investment in AI infrastructure is expected to rise to multi trillion dollar levels by the end of the decade. Within that wider market, Nvidia positions itself at the center of a substantial portion of demand even as supply-side limits constrain near-term delivery.

Nvidia expects demand for AI chips to surge while supply constraints, especially in HBM memory, are likely to persist through fiscal 2028.