NVIDIA’s Q3 FY2027 Results Show How Much Revenue Comes From Data Centres

NVIDIA’s Q3 FY2027 results show $115.8 billion in revenue for the quarter ended 25 October 2026. Data Centre revenue was $107.6 billion, about 93% of the total. Published on 17 November, the report puts AI infrastructure at the centre of the business. A separate Oracle announcement helps explain how the new Vera Rubin platform is moving into cloud services.
What NVIDIA’s Q3 FY2027 revenue tells you
NVIDIA’s results release says revenue rose 20% from the previous quarter and 103% from a year earlier. Those comparisons use different starting points. Quarter-on-quarter growth compares two adjacent periods, while the annual comparison uses the equivalent quarter one year back. Retaining the labels helps explain why the percentages differ.
The same applies to the fiscal-year title. FY2027 is the company’s reporting label for a period that ended during calendar 2026. When comparing two announcements, match the dates covered by the results before comparing the publication dates. The table should tell you both the period and the units used for its amounts.
Revenue measures recorded sales. A shipment update answers a question about product deployment, and an outlook describes management’s expectations. These can appear in the same announcement while referring to different points in time. For a useful comparison, first identify whether a number describes a completed quarter, an available service or a future estimate.
How Vera Rubin enters the cloud picture
NVIDIA and Oracle’s infrastructure announcement says Vera Rubin NVL72 systems and Vera CPUs are generally available on Oracle Cloud Infrastructure. It also describes plans to expand the deployment. General availability identifies a service milestone, while the expansion describes work the companies intend to carry out.
The announcement distinguishes GPU workloads from the supporting work performed on CPUs. Training and model inference can use GPU capacity, while an AI agent may also execute code and call tools. That supporting work can require CPU and memory resources. Choosing infrastructure for an application therefore involves more than selecting the processor with the largest advertised performance figure.
For a developer in India, the next practical checks are the service region, the instance family and the terms attached to access. A product announcement may cover a provider’s global offering. The provider’s account interface and region documentation will establish what a particular customer can request. The name of the hardware alone does not establish the available quota for an account.
How to compare the results with your own technology question
Start with the use you have in mind. If you are following AI infrastructure, the data centre share shows how heavily the reported business is weighted towards that market. If you are planning a cloud workload, look for the provider’s deployment information. If you are choosing a consumer device, use its own specification and price page to establish what it contains.
Keep the time period attached to any number you save. A revenue figure is easier to interpret when its quarter end and unit appear beside it. Likewise, write down a cloud service’s region and instance name when comparing availability. These details prevent a later update from being compared with an earlier figure that measured something different.
When reading a percentage, also check whether it refers to a change or a share of the total. A 93% contribution and a 103% annual increase describe separate relationships. Keeping those descriptions beside the figures makes a saved note easier to revisit.
The company’s financial reports page provides the accompanying documents for readers who want the full tables. Use those tables to follow the completed period and the product announcements to follow deployment. Together they show why AI infrastructure dominates this release and how a new platform is reaching cloud services. The circuit-board photograph is illustrative.



