Data Center Business Powers Record Revenue
Nvidia has reported another record-breaking quarter as demand for artificial intelligence (AI) compute continues to surge worldwide. The chip designer posted quarterly revenue of $68 billion, up 73 percent from the same period last year, making it one of the most profitable stretches in the company’s history. For the full year, Nvidia delivered $215 billion in revenue, underlining just how central its hardware has become to the global AI boom.
The main engine behind Nvidia’s performance remains its data center business. Out of the $68 billion in quarterly revenue, $62 billion came from data center sales alone. These products sit at the heart of modern AI infrastructure, powering large language models, recommendation systems and high-performance computing workloads in data centers around the world.
For the first time, Nvidia broke out its data center revenue into two separate lines: compute and networking. Compute revenue, which is largely driven by graphics processing units (GPUs), reached $51 billion. Networking products, including high-speed interconnect technologies such as NVLink, contributed a further $11 billion. Together, these numbers show how Nvidia has become a full stack provider for AI data centers, selling not only chips but also the connectivity needed to link thousands of GPUs into powerful clusters.
Token Demand and Fully Loaded GPUs
Chief Executive Officer Jensen Huang told analysts that demand for AI “tokens” — the basic units of text or data processed by AI models — has exploded. “The demand for tokens in the world has gone completely exponential,” he said, describing how enterprises, startups and cloud providers are racing to deploy AI services in many sectors of the economy.
Huang noted that even six-year-old GPUs in cloud data centers are running at full capacity and that pricing for available compute is rising. This suggests that the industry is still facing a supply-demand imbalance, where the appetite for AI compute outstrips the available hardware. As a result, customers are willing to pay premium rates to secure access to Nvidia-powered infrastructure.
No Revenue from China Despite Eased Controls
Despite a partial easing of U.S. export restrictions, Nvidia reported no revenue from chip sales to China during the quarter. Chief Financial Officer Colette Kress told investors that small quantities of H200 products destined for China-based customers had been approved by U.S. authorities, but these shipments had not yet translated into recognised revenue.
Kress added that the company remains uncertain about whether further exports to China will be allowed. At the same time, she pointed to rising competition inside China, where local chip-makers are gaining momentum and several firms have gone public. Over the longer term, she warned, Chinese competitors could become a serious factor in the global AI market, especially if export controls continue to limit Nvidia’s access to that segment.
AI Partnerships and Investment Talks with OpenAI
On the partnership side, Huang addressed reports that Nvidia has been discussing a potential $30 billion investment in OpenAI. He said the company continues to work towards a partnership agreement and believes it is getting close, but regulatory filings with the U.S. Securities and Exchange Commission have cautioned that there is no guarantee any investment will actually take place.
Huang also highlighted Nvidia’s existing relationships with other leading AI players, including Anthropic, Meta and Elon Musk’s xAI. These companies rely heavily on Nvidia hardware to train and deploy their cutting-edge models, further cementing the chipmaker’s position at the center of the AI ecosystem.
“In This New World of AI, Compute Is Revenue”
Responding to questions about whether massive capital expenditure on AI hardware is sustainable, Huang argued that data center investments are increasingly tied directly to revenue growth. “In this new world of AI, compute is revenue,” he said, explaining that without sufficient compute power, companies cannot generate tokens, and without tokens, they cannot grow AI-driven services or earnings.
He described the industry as being at an inflection point where tokens generated by AI models are not only useful for customers but also profitable for cloud service providers and software platforms. As more businesses integrate AI into their products, demand for GPUs and networking gear is likely to remain strong, keeping Nvidia at the forefront of the global race to build and monetise AI infrastructure.
