Demand
NVIDIA Data Center Revenue
Reported quarterly NVIDIA Data Center revenue in USD billions. A proxy for AI infrastructure demand—GPUs, networking, systems, and related software—not AI software revenue. Quarters are NVIDIA fiscal quarters (year ends late January), labeled Q4 FY23 and Q1 FY24, not calendar quarters.
NVIDIA Data Center Revenue
- NVIDIA
API access · Pro coming soon
As of 2026-01-01, NVIDIA Data Center revenue was $75.2 in the latest reported quarter.
Methodology
Values are NVIDIA-reported Data Center segment revenue from quarterly filings and investor commentary, converted from millions of dollars to billions. They are company-reported actuals, not estimates.
NVIDIA's fiscal year ends in late January, so its quarters do not match the calendar. Fiscal Q1 is February–April, Q2 May–July, Q3 August–October, and Q4 November–January. Axis labels follow NVIDIA's reported fiscal quarters (Q4 FY23, Q1 FY24, … Q1 FY27). OpenAI and Anthropic revenue charts use calendar quarters instead (Q1'23 is January–March 2023).
Data Center includes compute systems, networking such as NVLink, InfiniBand and Ethernet, associated software and services, and offerings such as DGX Cloud. NVIDIA has said large cloud service providers have recently represented about half of Data Center revenue.
This is a proxy for AI infrastructure investment, not for AI industry revenue. It can diverge from broader AI spending if NVIDIA gains or loses share to AMD or custom silicon, if networking mix shifts, or when a new GPU generation ships. Hardware sales are front-loaded capital expenditure: a cluster bought this quarter may generate AI services for years.
Beginning in Q1 FY27 NVIDIA splits Data Center into Hyperscale and ACIE (AI Clouds, Industrial & Enterprise). This series keeps the combined Data Center total for continuity.
Frequently asked questions
Is this a measure of AI industry revenue?
No. It is NVIDIA’s Data Center segment revenue: spending on NVIDIA compute, networking, systems, and related software used to build and run large-scale AI infrastructure. It is an excellent proxy for AI GPU demand and a very good proxy for hyperscaler AI capex and data-center buildout. It is a poor proxy for AI software/SaaS revenue, application usage, or the total economic value created by AI.
Why can this grow faster or slower than the AI market?
The series measures spending on NVIDIA, not total AI spending. Share shifts versus AMD or custom chips such as Google TPUs, mix between GPUs and networking, supply constraints, and Hopper-to-Blackwell-to-Rubin product cycles can all move NVIDIA revenue by more or less than underlying AI demand. Hardware revenue also leads usage: customers spend to build capacity, then monetize it over subsequent years.
Why are quarters labeled Q4 FY23 and Q1 FY24 instead of Q4 2022?
NVIDIA's fiscal year ends in late January, unlike a calendar year. Fiscal Q1 is February–April, Q2 is May–July, Q3 is August–October, and Q4 is November–January. We keep NVIDIA's own labels (Q4 FY23 through Q1 FY27) so a bar matches the quarter NVIDIA reported. That is different from the OpenAI and Anthropic revenue charts, which use calendar quarters: Q1'23 is January–March 2023.