Nvidia’s latest financial results show that the race to build artificial intelligence infrastructure is still drawing extraordinary sums of money. The semiconductor company reported revenue of US$115.8 billion for its fiscal third quarter, up 103 per cent from a year earlier, as data-centre revenue more than doubled. The numbers are impressive, but their business significance is less about one company beating expectations than about the scale of computing investment now concentrated in a small set of suppliers and customers.
The results, announced on 17 November, cover the quarter ended 25 October 2026. Nvidia reported data-centre revenue of US$107.6 billion, up 110 per cent year on year. It forecast fourth-quarter sales of around US$130 billion, plus or minus 3 per cent, while noting that the upcoming quarter contains an additional week and assumes no data-centre compute revenue from China.
Why data centres are dominating
AI models require specialised processors, high-speed networking, memory and the infrastructure needed to keep them operating reliably. Large technology firms and cloud providers have invested heavily in this capacity to train models and deliver AI services to customers. Nvidia has been one of the leading suppliers of the systems that power that investment.
A useful comparison is the gap between the company’s total sales and the data-centre segment. With US$107.6 billion of US$115.8 billion in quarterly revenue coming from data centres, the business is overwhelmingly exposed to enterprise and cloud computing demand. The figure also demonstrates why consumer gaming products, although important, are no longer the primary explanation for Nvidia’s financial performance.
Nvidia said its Vera Rubin NVL72 computing platform had begun shipping, an indication of its drive to replace older systems with more capable infrastructure. New generations can improve performance, but customers must still weigh hardware costs, power availability, software compatibility and the financial returns from deploying AI services.
Growth does not eliminate business risk
The company’s outlook incorporates uncertainty arising from export restrictions. By explicitly excluding China data-centre compute revenue from its fourth-quarter assumptions, Nvidia is showing how government policy shapes the addressable market for advanced processors. Companies operating data centres in Asia should consider such regulatory conditions alongside demand when planning capacity.
Another question is whether AI applications will generate sufficient revenue or productivity benefits to justify sustained capital expenditure. Nvidia’s revenue records sales of computing equipment and related systems; it does not independently establish that every customer has achieved a positive return on those investments.
Investors reacted cautiously at points despite the strong forecast, according to Reuters reporting on 18 November. Such reactions may reflect expectations already embedded in share prices and wider concerns about the pace of AI investment. A company’s strong earnings and a fluctuating share price are not contradictory: one describes operating performance, the other reflects changing expectations about the future.
What the numbers mean for Asia
Asia’s role in semiconductor manufacturing, electronics assembly and data-centre development gives the trend regional significance. Demand for advanced chips can benefit suppliers of memory, equipment, specialised materials and supporting infrastructure. At the same time, export licensing, energy costs and concentration in particular platforms introduce risks for businesses dependent on uninterrupted access.
The next questions are whether large buyers maintain their spending, whether deployment constraints emerge and how successfully businesses turn AI capacity into useful services. Nvidia has provided a clear measure of the hardware boom, but it remains only one part of a wider commercial equation.
For executives, the lesson is not simply to buy more computing power. It is to understand which workloads produce measurable value, how resilient the supply chain is and whether forecasts account for policy changes. Those considerations are likely to outlast any single extraordinary quarter.
Sources and further reading
Sources: Nvidia: fiscal 2027 third-quarter financial results, 17 November; Reuters: market reaction, 18 November.