First look at Nvidia’s AI system Vera Rubin and how it beats Blackwell

Nvidia’s earnings on Wednesday are expected to show booming sales of the company’s current rack-scale system. But all eyes are on its next AI system, Vera Rubin, which is scheduled to roll out later this year.

Vera Rubin, which is made up of 1.3 million components, will deliver 10 times more performance per watt than its predecessor, Grace Blackwell, the company claims. That’s a significant development when energy consumption is one of the most critical issues facing the artificial intelligence build-out.

CNBC got an exclusive first look at Vera Rubin at Nvidia’s headquarters in Santa Clara, California.

Nvidia says the new AI system is a complex web of parts sourced from around the world. Its core chips include 72 Rubin graphics processing units, or GPUs, and 36 Vera central processing units, or CPUs, primarily made by Taiwan Semiconductor Manufacturing Co. The other parts, from liquid cooling elements to power systems and compute trays, come from more than 80 suppliers in at least 20 countries, including China, Vietnam, Thailand, Mexico, Israel and the U.S.

One big challenge the company faces is the soaring costs of memory due to a global shortage from all the AI-driven demand. Dion Harris, Nvidia’s AI infrastructure head, said in an interview that the company has been giving suppliers “very detailed forecasts.”

“We’re aligning to make sure that everything we’re shipping will be met by our supply chain,” he said. “We’re in good shape.” 

It’s a critical moment for Nvidia, which dominates the market for AI processors but faces intensifying competition from Advanced Micro Devices as well as custom silicon from Broadcom and Google’s homegrown tensor processing units. Nvidia has plans to manufacture up to $500 billion of AI infrastructure in the U.S. through 2029, including making Blackwell GPUs at TSMC’s new Arizona fabs.

Grace Blackwell went into production in 2024, and changed the game on how much compute was possible with a single system. Vera Rubin, which is expected to ship in the second half of 2026, takes the company to another level. Nvidia CEO Jensen Huang announced in January that the system was in full production.

“These things are massive. They’ve got all the compute, all the networking, all the cabling, all the cooling,” said Daniel Newman of research firm Futurum Group. “They’ve got all these systems pulled together into a single rack built for absolute greatest efficiency and greatest performance. And that’s just not how servers were historically built.”

Last week, Meta announced plans to use Vera Rubin in its data centers by 2027. Nvidia’s list of other expected Vera Rubin customers includes OpenAI, Anthropic, Amazon, Google and Microsoft. The racks, manufactured in the U.S. and elsewhere, including in Taiwan and at a new Foxconn plant in Mexico, weigh nearly 2 tons and have about 1,300 total microchips, compared with Grace Blackwell’s 864. 

Vera Rubin is a simpler, modular system intended to ease installs and repairs. Each superchip slides out of one of the rack’s 18 compute trays in seconds. In the Blackwell system, those components are soldered to the board.

Nvidia said the new system will consume about twice as much power as its predecessor, but will be far more efficient because of that 10 times return on performance per watt.

Jordan Klein, an analyst at Mizuho Securities, said what “matters the most” is “how many tokens per power consumed can you get.”

“The more you can tweak that or move up the curve, the higher the return would be on the dollar you spend,” Klein said.

Vera Rubin is also Nvidia’s first system that’s 100% liquid cooled, which Harris said helps data centers consume “much less water” than traditional evaporative cooling.

Nvidia doesn’t share rack pricing, but Futurum Group estimates the price will increase about 25% from Grace Blackwell, bringing the system price to somewhere around $3.5 million to $4 million.

As major customers seek to diversify their reliance on the chipmaker, many are also filling AI servers with their own in-house silicon. CNBC visited an Amazon Web Services data center in October packed with “ultra-servers” made up of the company’s Trainium 2 chips. Meanwhile, Google’s data centers are loaded with racks of its TPUs.

Later this year, Nvidia will see some big competition when rival AMD ships its first rack-scale system called Helios. The chipmaker just secured a major commitment from Meta for up to 6 gigawatts of capacity.

“You’re going to see a lot of uptake because customers want more capacity, but they also want a viable second source to keep Nvidia honest,” Klein said.

With respect to the competition, Harris said: “Hats off to anyone who’s going to try. But this is certainly not a simple endeavor.”

Watch the video for a detailed look at Vera Rubin and how the system works.