Jensen Huang, Nvidia’s founder, president and CEO, speaks about the future of artificial intelligence and its effect on energy consumption and production at the Bipartisan Policy Center in Washington, D.C., on Sept. 27, 2024.
Chip Somodevilla | Getty Images
Nvidia revealed new chips for desktop and laptop PCs on Monday that use the same Blackwell architecture underpinning the company’s fastest AI processors for servers and data centers.
The chips, called GeForce RTX 50-series, will come pre-installed in computers ranging from about $550 to $2,000, the company said. Laptops with the chips will start shipping in March.
Nvidia unveiled the processors at CES in Las Vegas, where CEO Jensen Huang delivered a keynote address on Monday.
“Can you imagine, you have this incredible graphics card, Blackwell, I’m going to shrink it and put it in there,” Huang said, holding up a laptop.
Nvidia, which has soared past $3.5 trillion in market cap by selling AI chips to giant cloud vendors and other tech companies, was until the last few years known for selling graphics processing units (GPUs) to power video games. Nvidia’s first chip in late 1999 was designed to draw triangles and polygons quickly for 3D games.
“Of course, back then, we were a gaming company, and these GPUs were created to accelerate games,” Justin Walker, senior director of product at Nvidia, said on a press call.
Wall Street is less enthused about Nvidia’s gaming business these days given the explosion in AI and ever-increasing demand for more processing power. In the quarter that ended in October, Nvidia’s gaming sales accounted for under 10% of total revenue, compared to 88% from data center chips.
Nvidia has the vast majority of the AI GPU market for data centers, outpacing rivals Advanced Micro Devices and Intel.
But CES, formerly known as the Consumer Electronics Show, is all about consumer products, and the new chips announced Monday are primarily intended for gaming.
Nvidia says the RTX 50-series chips will support a feature called DLSS 4 that uses AI to boost gaming frame rates. They also can display character faces with more details, and will generally provide users with better graphics and higher resolution.
Nvidia’s gaming business is growing, with revenue increasing 15% from a year earlier in the latest quarter. But data center sales have at least doubled for six straight quarters, topping $30 billion in the most recent period.
Nvidia says that technical improvements made for its massive AI business will trickle down into its graphics cards for games.
“While we are now an AI company as well as a gaming company, our gaming side still benefits tremendously from the fact that we are an AI company,” Walker said.
The Blackwell GPU architecture and core design that the 50-series chips use debuted in the company’s AI accelerators, which were announced in March and started shipping later last year. Nvidia said they were designed and optimized to run neural networks used by OpenAI’s ChatGPT and Google’s Gemini.
The new chips for PCs and laptops will come in a number of different configurations. The company says the most expensive and powerful of the chips, the RTX 5090, will be sold individually for $1,999 and is twice as fast as its predecessor, the RTX 4090. Nvidia says it has 92 billion transistors, versus 208 billion transistors on the company’s B200 GPU for servers.
Nvidia says the chips will be optimized to run AI models and do computer graphics, not just run the latest games. The chips will be powerful enough for some game makers to integrate generative AI into their characters in games like “PUBG: Battlegrounds.”
The new processors will also will be powerful enough to to run large language models and image generation models from companies including Meta, Mistral and Stability AI, Nvidia said.
WATCH: Bernstein’s Rasgon says investors need Nvidia exposure
Read the original article here