Jensen Huang pleased the crowd at the CES conference this week — just not the “right” crowd.
The Nvidia co-founder and chief executive was in rare form for the annual conference’s keynote address, even swapping his typical black leather racing jacket for a glittery one. He used the occasion to introduce products such as new videogaming processors and even an artificial-intelligence supercomputer the size of a large sandwich. He also announced new efforts in humanoid robots and self-driving cars, the latter of which he predicted will be the “first trillion-dollar robotics industry.”
The in-person audience was thrilled — particularly with a new family of gaming graphics cards based on the company’s Blackwell AI chip. But investors seemed disappointed, sending Nvidia’s stock sliding more than 6% Tuesday. That was double what the shares gained the previous day, when Nvidia seemed on track to surpass Apple and perhaps become the first public company to reach a $4 trillion market capitalization.
Nvidia was worth just one-third that amount a year ago, so even its current valuation at $3.4 trillion is nothing to sneeze at. But that rise, driven by explosive sales and earnings growth, naturally makes Nvidia’s stock highly-sensitive to any bumps in the road. It has also made investors highly-focused on the business segment of AI chips used in data centers that is driving the bulk of the company’s growth these days. Nvidia gave no new updates on that business at the conference, also known as the Consumer Electronics Show. In a note to clients, Cody Acree of Benchmark called Huang’s address “a solid, but somewhat muted CES keynote” and said many investors “were hoping for more concrete progress updates on the ramp of Blackwell,” as well as what comes after.
Nvidia’s data center segment alone is projected to generate about $113 billion in revenue for the fiscal year ending in January — more than the projected total revenue of any other chip company this year, according to FactSet. So that business naturally gets a lot of attention. Much of Nvidia’s current value stems from its ability to keep convincing giants such as Microsoft, Amazon, Google and Meta Platforms to spend tens of billions every year on AI systems. But Nvidia has long had extensive exposure to other industries, such as cars, games and high-end design. Many of those industries stand to be impacted by generative AI, or already have been. Those are important future opportunities for a company that could soon be valued at $4 trillion.
Being recognized for that will be a longer and slower process, though. Nvidia’s second largest business unit, graphics processors for gaming PCs, is on track to generate a little under $12 billion in revenue this year and is about a tenth the size of its data center segment. The company’s automotive segment is expected to grow to about $2 billion in revenue in the next fiscal year, though the company says total automotive revenue that year would be more like $5 billion if counting the data center computing necessary to power self-driving cars.
Such lopsided business realities will keep data centers in the driver’s seat of Nvidia’s stock price for a long while. The company has only just started shipping its Blackwell products. Huang described the demand as “enormous.” But Wall Street is already speculating on the next data center product family called Rubin, which isn’t expected to start shipping until 2026. That one needs to be big as well. Analysts expect Nvidia’s data center sales to eclipse the $200 billion mark in the fiscal year ending January 2027, according to Visible Alpha estimates.
Convincing investors to treat Nvidia’s other acts as more than a sideshow is no easy feat.