Nvidia posts $57 billion in revenue and forecasts $65 billion
Nvidia closed its fiscal third quarter with $57.006 billion in revenue, up 62% year over year. Data centers accounted for nearly 90% of the business, and the company expects to reach $65 billion in the current quarter.
Nvidia has once again beaten its own financial records. The company generated $57.006 billion in revenue in the third quarter of fiscal 2026, which ended October 26, up 22% from the previous quarter and 62% from a year earlier.
The figure matters because it comes as investors and companies question whether the enormous spending on AI data centers will generate sufficient returns. For now, demand for Nvidia chips continues to grow: the company expects revenue of $65 billion, plus or minus 2%, in the current quarter.
Data centers now underpin nearly the entire business
Nvidia’s data center division generated $51.2 billion in revenue, up 66% year over year. That represents nearly 90% of the company’s total revenue. The business sells GPUs — specialized processors that train and run AI models — along with networking equipment and complete systems for large-scale computing facilities.
The rest of the business is far smaller. Gaming contributed $4.3 billion, up 30% from a year earlier; professional visualization generated $760 million; and automotive and robotics brought in $592 million. Nvidia remains a consumer brand best known for GeForce, but its results now depend primarily on the infrastructure purchased by cloud providers, AI labs and large corporations.
Jensen Huang, the company’s founder and CEO, said Blackwell sales were off the charts and that cloud GPUs were sold out. Blackwell is Nvidia’s latest chip architecture for large-scale AI and succeeds Hopper, the generation that made the company the central supplier in the current data center expansion.
Higher revenue, but exceptional profitability too
This is not simply a matter of selling more. Nvidia reported net income of $31.910 billion under GAAP accounting rules, up 65% from a year earlier. Its gross margin was 73.4%: for every $100 in revenue after selling its products, the company kept roughly $73 before operating expenses, taxes and other costs.
That profitability reflects the company’s negotiating position, although it does not guarantee that it will last indefinitely. Designing advanced chips requires enormous investment, but Nvidia sells a platform that is difficult to replace overnight: processors, networking, CUDA software and systems ready for AI deployment. For a customer already operating thousands of GPUs, switching suppliers entails technical costs and takes time.
The company also returned $37 billion to shareholders during the first nine months of the fiscal year through share repurchases and dividends. At the end of the quarter, it still had authorization to buy back another $62.2 billion worth of stock.
Demand is expanding from model training to inference
Nvidia’s argument is that the market does not depend solely on training ever-larger models. Inference is also growing: the day-to-day use of those models to generate responses, analyze documents, write code or run software agents. Every AI product that reaches users can become a recurring source of demand for computing capacity.
The company has announced large-scale infrastructure agreements with companies including OpenAI, Anthropic, Microsoft, Oracle, Google Cloud and xAI. They include a partnership with OpenAI to deploy at least 10 gigawatts of Nvidia systems for next-generation infrastructure. A gigawatt is a scale associated with major electricity facilities; the figure illustrates how far leading labs plan to turn computing into an industrial commodity.
A strong result does not settle the debate over spending
The results confirm that Nvidia is selling hardware at an extraordinary pace and that its customers are sticking to their expansion plans. That undercuts the idea of an immediate drop in demand for AI infrastructure.
But Nvidia’s accounts alone do not answer the sector’s most important question: how much money will the buyers of those chips ultimately generate from their assistants, cloud services and enterprise applications? Nvidia’s revenue measures investment in capacity; the final return on that investment will show up in the accounts of those renting or using that capacity.
For now, the forecast of $65 billion in revenue for the fourth fiscal quarter indicates that the spending cycle has not slowed. The next important signal will be whether that demand continues to spread across companies and countries or remains concentrated among a small group of major technology platforms.
