Nvidia announced a surprise beat for the fourth quarter of fiscal year 2026 (Nov. 2025–Jan. 2026), topping market expectations, and said this year's results could also grow sharply. The company said the era of physical AI has already arrived and has begun to be reflected in revenue, citing examples from Hyundai Motor Group's Boston Dynamics and LG Electronics. Jensen Huang, Nvidia chief executive officer (CEO), said, "We are investing with a strategic focus on expanding the ecosystem."
Nvidia said on the 25th (local time) that fourth-quarter fiscal year 2026 revenue came in at $68.13 billion (about 97.4327 trillion won). That was up 73% from a year earlier, marking a record high. The result beat the $66.2 billion consensus compiled by market researcher LSEG.
Operating profit for the period was $44.3 billion (about 63.2382 trillion won), up 84% from a year earlier. Operating margin reached 65%, and adjusted earnings per share (EPS) was $1.62.
Of Nvidia's total fourth-quarter fiscal year 2026 revenue, 91.4% ($62.3 billion; about 88.9332 trillion won) came from the data center business. That segment's revenue rose 75% from a year earlier, driving overall growth. In addition, gaming brought in $3.7 billion, professional graphics $1.3 billion, and automotive and embedded $600 million.
Annual revenue was $215.9 billion (about 308.1972 trillion won), up 65% from a year earlier and setting a record high. Annual adjusted EPS was $4.77. The company plans to pay a cash dividend of $0.01 per share on Apr. 1 to all shareholders of record as of Mar. 11 this year.
Colette Kress, Nvidia chief financial officer (CFO), said on a conference call, "Since ChatGPT emerged, the scale of the data center business has grown 13-fold," and "We expect revenue to increase sequentially by quarter this year."
◇ "Sending Vera Rubin samples to customers… mass production in the second half"
Nvidia guided first-quarter fiscal 2027 (Feb.–Apr. this year) revenue at $78 billion (±2%; about 111.345 trillion won). That is also more than 7% above the market estimate of $72.6 billion. Nvidia said physical AI solutions and the next-generation AI chip "Vera Rubin" will drive growth.
Nvidia unveiled Cosmos, a physical AI deployment model, and Isaac GROOT, a custom model for humanoid robot inference. CFO Kress said, "The era of physical AI has arrived, and it has already contributed more than $6 billion to revenue," adding, "Through Cosmos and Isaac GROOT, we are also driving Robotics development." Citing Boston Dynamics, LG Electronics, Neura Robotics, and FRANKA Robotics as examples, Kress said, "We are accelerating the adoption of industrial physical AI, with major companies' robots running on Nvidia's solutions."
Huang cited "ecosystem expansion" as a key achievement last quarter. Of OpenAI, the developer of ChatGPT, Huang said, "A company that may come along once in a generation," adding, "We are pleased to have been with them since the early days of the business."
However, Huang avoided a direct answer to speculation that the previously promised $100 billion investment had been reduced to $30 billion. Huang said, "We are continuing to work on a partnership agreement with OpenAI," adding, "I think we are close to sealing it."
Huang also cited collaborations with Meta, Anthropic, and xAI, saying, "We have the capability to build full-stack AI infrastructure from the ground up or support it in the cloud, enabling us to maintain unmatched partnerships with leading model development corporations across all stages of scaling training and inference AI." He added, "We want to put as many ecosystems as possible on top of Nvidia."
Huang expressed strong confidence in the next-generation AI chip "Vera Rubin." Huang said, "Computing demand is growing exponentially, and the inflection point for agent AI has arrived," adding, "As corporations rapidly adopt agent-based AI, they are competitively investing in computing, and this is a key driver of the AI industrial revolution and future growth." Huang added, "Grace Blackwell is the current leader in inference, showcasing innovative technology that dramatically lowers cost per token," and "Vera Rubin will further strengthen this leadership."
Nvidia recently delivered samples of Vera Rubin to customers. CFO Kress said, "Our plan to begin mass-production shipments in the second half of the year is unchanged," adding, "There is no doubt that demand and interest are strong, and we expect virtually all customers to purchase Vera Rubin." Kress added, "We also plan to continue selling Blackwell, so there will be some overlap with the timing when Vera Rubin begins to hit the market." However, on initial volumes for Vera Rubin, Kress said, "It is still early to be specific," adding, "The key is when we can start full-scale supply of Vera Rubin and how quickly customers can build out their data centers."