Investors hoping for a surprise new revenue stream instead got an energetic keynote that was lacking in fresh developments
Jensen Huang showed off a robot during his Tuesday GTC keynote.
Jensen Huang brought the enthusiasm on Tuesday as he spoke for about two hours — supposedly without a teleprompter — about the big things ahead for Nvidia Corp.
But Wall Street wasn’t as excited about what he had to say. Nvidia’s stock, which was down nearly 1% ahead of Huang’s keynote, ended the day off 3.4%.
Nvidia clearly has a lot in its arsenal, as evidenced by a detailed roadmap for future product offerings and a vision to participate even more in industries like robotics and autonomous driving. But there were perhaps two things to nag at in CEO Huang’s speech. One is that many of the big opportunities ahead of Nvidia won’t translate into major revenue streams anytime soon. Another is that the keynote went largely as expected.
The things Huang said “had already been said,” according to Maribel Lopez, a tech-industry analyst and founder of Lopez Research.
Plus, there was hope that GTC would bring “a magic rabbit pulled out of a hat” in the form of a new revenue stream, she told MarketWatch. Instead, Huang mentioned areas like robotics and quantum computing that will take time to deliver payoffs.
Quantum computing “is still not going to be a short-term revenue driver for anybody,” Lopez explained. And while robotics is a more mainstream field, it’s also not a major short-term sales driver, in her view, and it won’t scale as fast as Nvidia’s core chip business has.
Jefferies analyst Blayne Curtis added that he was “hoping for more proof points” around the potential for Nvidia to expand its total addressable market and deliver advantages in total cost of ownership. Still, “what is clear is the scale and strength of [Nvidia’s] offerings across hardware/software and vertical domains,” he wrote in a note to clients.
In detailing the company’s product roadmap, which is now on a cadence of a new product about every 12 to 18 months, Huang said the next-generation platform called Vera Rubin is due in the second half of 2026. It will have 3.3 times the performance of Nvidia’s Grace Blackwell system, with 144 graphics processing units.
Following that will be Vera Rubin Ultra in the second half of 2027, an even more massive system which will have 14.4 times the performance of Grace Blackwell, and 576 GPUs.
Curtis wrote that “the updated roadmap does suggest Rubin will only be an incremental update in 2026 with Rubin Ultra the more meaningful leap ahead in 2027.”
The company also teased Feynman, which will follow the Blackwell and Rubin families with an expected 2028 launch. It’s named after theoretical physicist Richard Feynman, in keeping with Nvidia’s pattern of honoring prominent scientists through its product monikers.
Nvidia’s ability to innovate quickly has kept competitors at bay and could continue to do so. “Even when they screw up” — like they did initially with Blackwell — “they recover quickly,” Lopez said. The company is on a 12- to 18-month product curve that executives “don’t miss,” he said.
But Nvidia isn’t backing down from painting a vast picture of what’s ahead. “This isn’t like buying a laptop,” Huang explained. Customers need time to do things like create budgets and plan for power usage.
