Nvidia’s Blowout Q4 Guidance Reignite the Entire AI Trade

Nvidia’s fiscal Q3 results and Q4 guidance both far exceeded expectations, effectively rescuing the U.S. equity market’s AI narrative.

The company reported revenue of $57.01 billion, up 62% year over year, surpassing the expected $54.92 billion. Adjusted EPS came in at $1.30, up 60%, also beating market expectations of $1.26.

The most closely watched segment—data-center revenue—reached $51.2 billion, up 66% year over year and ahead of the expected $49.3 billion. “Compute” (i.e., GPU) products contributed $43 billion, while networking products (the components enabling multiple GPUs to work together) contributed $8.2 billion. All remaining business lines posted growth of over 30%.

On guidance, Nvidia expects Q4 revenue between $63.7 billion and $66.3 billion, with a midpoint of $65 billion, up 65% year over year and easily above analyst expectations of $62 billion. Notably, Nvidia’s Q4 revenue guidance does not include China. If China shipments resume, revenue could deliver further upside surprises.

The company expects Q4 adjusted gross margin between 74.5% and 75.5%, with a midpoint of 75%, above analysts’ midpoint forecast of 74.6%. This indicates that despite rising raw-material costs, Nvidia’s pricing power remains extremely strong, maintaining roughly 75% gross margins.

Following the earnings release, Nvidia’s share price jumped 5% in after-hours trading.

Key Takeaways From the Earnings Call: Nvidia Leads the World in “Revenue per Watt”

During the earnings call, Nvidia’s management emphasized that its GPUs maintain high utilization over long periods. A100 GPUs shipped six years ago are still running at full capacity, whereas competing products from the same era have long been obsolete.

Nvidia’s CUDA ecosystem provides strong compatibility, extending GPU lifespan and creating a significant cost advantage over competitors.

Each new generation of Nvidia GPUs is gaining share in global data centers. Hopper accounts for 25%, Blackwell has reached 30%, and Rubin could be even higher. Power supply to data centers is limited, making “revenue per watt” crucial. Nvidia’s architecture delivers the highest revenue per watt globally, meaning that with the same electrical power, Nvidia’s systems generate the greatest economic output.

The company’s best-selling chip is Blackwell Ultra. From now through the end of 2026, Nvidia expects Blackwell and Rubin combined to generate $500 billion in revenue, with further upside from new partnerships such as those with Saudi Arabia and Anthropic. Over the next 12–18 months, Nvidia products will remain in persistent shortage, with cloud GPUs already fully sold out.

CEO Jensen Huang noted that the total capital spending by top cloud providers and hyperscale data centers will continue rising into 2026, now estimated at $600 billion, more than $200 billion higher than earlier forecasts. This provides strong long-term visibility for Nvidia’s growth.

Nvidia’s networking business more than doubled, driven by the rise of NVLink Scale-Up and strong growth in both Spectrum X Ethernet and Quantum X On The World. Giants such as Meta, Microsoft, Oracle, and xAI are using Spectrum X Ethernet switches to build gigawatt-scale AI factories.

Physical AI may become Nvidia’s next growth engine. Leading U.S. robotics companies are using Nvidia’s Jetson AGX Orin platform; Nvidia is also working with Uber, using the NVIDIA Hyperion L4 architecture to expand the world’s largest Level-4 autonomous-driving fleet.

As the holiday season approaches, gaming demand remains strong, with Steam active users hitting a record high. Nvidia GPUs remain an essential upgrade for gamers.

Huang also noted that Nvidia is expanding the CUDA ecosystem through strategic investments in leading AI companies such as OpenAI and Anthropic, gaining stakes in what he described as “the most influential companies of a generation” and ensuring long-term returns.

OpenAI’s weekly active users have grown to 800 million, with enterprise customers reaching 1 million, supported by healthy margins. Anthropic’s annualized revenue reached $7 billion as of last month, up from $1 billion at the start of the year. This underpins Nvidia’s confidence in making substantial investments in these two AI unicorns.