Nvidia reported another sharp increase in revenue on February 26, 2025, as demand for artificial intelligence infrastructure continued to drive sales of its data center chips and systems. The company's official fourth-quarter fiscal 2025 results showed revenue of $39.33 billion for the quarter ended January 26, 2025, up 78% from a year earlier and 12% from the previous quarter.
The results exceeded the Wall Street consensus reported at the time. Analysts surveyed by LSEG had expected approximately $38.05 billion in revenue and adjusted earnings of $0.84 per share. Nvidia reported both GAAP and non-GAAP diluted earnings of $0.89 per share.
The 78% figure refers specifically to Nvidia's fourth quarter of fiscal 2025, not its current growth rate. It was already a slowdown from the extraordinary expansion seen a year earlier, when the company's quarterly sales had more than tripled. Even so, the scale of Nvidia's business had become dramatically larger.
Nvidia's Q4 fiscal 2025 results at a glance
| Metric | Q4 FY2025 | Year-over-year change |
|---|---|---|
| Revenue | $39.33 billion | Up 78% |
| GAAP diluted EPS | $0.89 | Up 82% |
| Net income | $22.09 billion | Up 80% |
| Data Center revenue | $35.58 billion | Up 93% |
| GAAP gross margin | 73.0% | Down 3 percentage points |
For the full fiscal year, Nvidia generated $130.5 billion in revenue, an increase of 114% from fiscal 2024. Full-year Data Center revenue reached $115.2 billion, up 142%.
AI data centers were responsible for most of the growth
Nvidia's Data Center business generated $35.58 billion in the quarter, accounting for roughly 90% of total company revenue. Data Center sales increased 93% year over year and 16% from the previous quarter.
According to Nvidia's CFO commentary, the increase was driven by demand for accelerated computing platforms used for large language models, recommendation systems and generative AI applications.
The quarter was also the first major revenue period for Nvidia's Blackwell architecture. Nvidia said it generated $11 billion in Blackwell revenue during the quarter, describing the launch as the fastest product ramp in the company's history. Large cloud service providers were major Blackwell customers and represented approximately half of Nvidia's Data Center revenue.
Blackwell was therefore more than a future product story by the time these results were published: it had already become a multibillion-dollar business alongside continued demand for Nvidia's earlier Hopper-generation products, including the H200.
Why Nvidia focused so heavily on AI inference
One of the major questions surrounding Nvidia at the time was whether increasingly efficient AI models could reduce the amount of computing hardware required to run artificial intelligence. Those concerns intensified after the emergence of DeepSeek's R1 reasoning model.
Nvidia's management argued that greater model efficiency would not necessarily translate into lower overall demand for computing infrastructure. Instead, the company expected reasoning models and more intensive inference workloads to use substantially more computation for each complex task.
That was a forward-looking argument from Nvidia rather than a guarantee about future demand. However, the company was already reporting that a growing portion of its infrastructure was being used for inference—the process of running a trained AI model to generate answers, predictions or other outputs—rather than only for training new models.
Blackwell growth came with higher costs
The quarter was not strong across every metric. Nvidia's GAAP gross margin declined to 73.0%, from 76.0% a year earlier. Non-GAAP gross margin fell to 73.5%, from 76.7%.
Nvidia attributed the decline primarily to its transition toward more complex and more expensive Data Center systems. Blackwell products were increasingly being delivered not simply as individual GPUs but as tightly integrated computing platforms containing processors, high-speed interconnects, networking equipment and other components.
This made the Blackwell transition important for revenue growth while also putting temporary pressure on profitability during the production ramp.
Gaming and networking did not share the same momentum
The headline 78% revenue increase also masked significant differences between Nvidia's businesses.
Gaming revenue was $2.54 billion, down 11% from a year earlier and 22% from the previous quarter. Nvidia said limited supply of both Blackwell- and Ada-based GPUs contributed to the decline. The company had begun introducing its GeForce RTX 50 Series consumer graphics cards based on the Blackwell architecture during this period.
Networking revenue was approximately $3.02 billion, down 9% year over year. Nvidia was transitioning toward larger-scale NVLink systems and its Spectrum-X Ethernet platform as AI clusters grew in size.
Automotive revenue remained much smaller than Data Center revenue but increased rapidly, reaching $570 million, up 103% from the previous year. Nvidia attributed the increase primarily to sales of its self-driving computing platforms.
Nvidia forecast another record quarter
Alongside the February 2025 results, Nvidia projected approximately $43 billion in revenue for the first quarter of fiscal 2026, plus or minus 2%. That guidance implied growth of roughly 65% from the same quarter a year earlier.
The company ultimately exceeded that forecast. In May 2025, Nvidia reported first-quarter fiscal 2026 revenue of $44.1 billion, up 69% year over year.
But a major new risk had emerged by then. Nvidia said the U.S. government informed it in April 2025 that exports of its H20 products to China would require a license. The company recorded a $4.5 billion charge related to H20 inventory and purchase commitments and said it had been unable to ship an additional $2.5 billion of H20 revenue during that quarter.
That development demonstrated that Nvidia's future growth depended not only on AI demand and product execution but also on export controls and access to major geographic markets.
Later results showed that the AI infrastructure expansion continued
The February 2025 earnings report is now a historical snapshot of an earlier stage in Nvidia's AI expansion. The company continued to grow substantially afterward while moving beyond the initial Blackwell rollout.
In its second-quarter fiscal 2027 results published on August 26, 2026, Nvidia reported quarterly revenue of $96.2 billion, up 106% from a year earlier. Data Center revenue reached $89.0 billion, up 117%.
By that point, Nvidia said its newer Vera Rubin platform was in full production, showing how quickly the company's AI hardware cycle had advanced beyond the Blackwell launch that dominated the February 2025 report.
Export restrictions remained an important limitation, however. Nvidia's outlook for the following quarter assumed no Data Center compute revenue from China.
The broader takeaway from the original results is therefore more nuanced than the 78% headline alone. Nvidia's fiscal 2025 fourth quarter confirmed extraordinary demand for AI computing, demonstrated that Blackwell had reached meaningful commercial scale and showed Data Center revenue becoming overwhelmingly dominant. Later results confirmed that AI infrastructure spending continued to expand, while manufacturing complexity, product transitions and restrictions on sales to China remained material factors shaping the business.
0 Comments