NVIDIA reports $107.6 billion in data center sales as Rubin begins shipping

NVIDIA reported $107.6 billion in data center sales for fiscal Q3 2027 and began shipping Vera Rubin NVL72. See revenue breakdown and deployment context.

By Hypler editorial team · Source: NVIDIA.

Original source date: .

Hypler publication: . Hypler updated: .

Primary source last checked: .

Source check notes

NVIDIA earnings release and Form 8-K verify quarter end, release date, reported results and guidance. Shipment and partner deployment statements are attributed to NVIDIA.

Estimated reading time: 2 minutes.

Topics: NVIDIA earnings, Vera Rubin NVL72, data center revenue, AI compute capacity

Bar chart comparing NVIDIA fiscal Q3 2027 total revenue of $115.8 billion and its $107.6 billion data center component.
Bar chart comparing NVIDIA fiscal Q3 2027 total revenue of $115.8 billion and its $107.6 billion data center component. Credit: Hypler editorial team. Original Hypler editorial graphic

What NVIDIA reported

NVIDIA reported $115.8 billion in revenue for its quarter ended October 25, up 103% from a year earlier. Data center revenue reached $107.6 billion, up 110%, accounting for about 93% of the total. The November 17 release also said NVIDIA had started shipping Vera Rubin NVL72 and that named partners were running racks. The report did not quantify available Rubin compute across those partners.

NVIDIA projected $130 billion in revenue, plus or minus 3%, for the following quarter. The company said the quarter includes an extra week and assumed zero data center compute revenue from China. Both details matter when comparing successive quarters. The China assumption applies to the forecast and its stated product category.

Hypler analysis

Revenue tells us when vendors recognize sales. An inference team also needs to know when it can reserve a cluster, how much power is available, and what its workload can sustain. A rack can arrive before a site finishes commissioning its cooling or fabric. It can pass commissioning before operators can serve a model at their required latency. NVIDIA’s figures establish the scale of its data center business. They give buyers little basis for forecasting a specific deployment date.

For American AI leadership, the next numbers to watch are commissioned megawatts, usable accelerator hours and performance under sustained workloads. Those measurements tell us how much infrastructure is available for running models in research and production. Customers comparing deployment options should ask providers for a commissioning schedule, a usable capacity commitment and a workload-specific acceptance test.

Hypler analysis

Hypler’s view

A reported sale, a shipped rack, and compute available for a specific workload each answer different deployment questions.

Terms and definitions

GPU
Graphics processing unit: a processor designed for highly parallel computation, commonly used for AI training and inference.
Inference
Using a trained model to produce an output from new input.

More on this subject

Generated illustration of connected accelerator trays and cooling channels; not a vendor facility.
Generated illustration of connected accelerator trays and cooling channels; not a vendor facility. Credit: Hypler / AI-generated editorial illustration. Hypler editorial asset; not vendor or event photography

Original source

The article attributes technical and financial results to the primary source and identifies Hypler analysis separately. The supporting image is conceptual.