Nvidia surges 2.9% on GTC 2026 announcements
Vera Rubin AI platform delivers 5x performance boost as Huang charts three-generation roadmap to 2028
Nvidia shares rallied 2.9% on Monday as Chief Executive Officer Jensen Huang unveiled a sweeping product roadmap at the company's GPU Technology Conference, headlined by the Vera Rubin AI platform that delivers a fivefold improvement in inference performance over its predecessor.
The stock climbed to $185.50, adding $5.26 in afternoon trading on the Nasdaq, giving the chipmaker a market capitalization of $4.38 trillion. The advance extended Nvidia's year-to-date gain and reinforced its position as the dominant player in artificial intelligence hardware despite growing competition from Advanced Micro Devices and custom silicon initiatives from cloud providers.
The centerpiece of Huang's keynote was the Vera Rubin platform, which the company said is in full production and represents a significant leap beyond the Blackwell architecture. The new system offers a 10x reduction in inference token costs and requires four times fewer GPUs to train equivalent Mixture-of-Experts models, addressing key concerns among enterprise customers about the escalating costs of AI deployment.
"The Rubin Revolution is underway," Huang told attendees at the San Jose conference, referring to the platform named after the pioneering astronomer. "We're not just building chips—we're building the infrastructure for agentic AI."
The Vera Rubin architecture features the R100 GPU built on a 3-nanometer process with 336 billion transistors, paired with HBM4 memory delivering more than 3.0 TB/s of bandwidth. The flagship VR200 NVL144 rack system combines 144 Vera Rubin GPUs with a new Vera CPU featuring 88 custom ARM cores. Microsoft Azure has been named the first cloud provider to validate the system.
Nvidia also announced the N1X chip, an ARM-based system-on-chip developed with MediaTek that marks the company's return to the consumer PC market. The chip incorporates 20 custom ARM cores and an integrated GPU with performance equivalent to a standalone RTX 5070 graphics card, positioning Nvidia to capture a portion of the estimated $200 billion PC market as manufacturers race to build AI-capable laptops.
"The N1X represents Nvidia's strategic expansion beyond data centers into the massive installed base of Windows PCs," analysts noted. "By enabling local AI inference without cloud connectivity, Nvidia addresses both privacy concerns and the economics of running AI workloads at the edge."
In a move that extends Nvidia's reach beyond hardware, Huang introduced NemoClaw, an open-source platform for building enterprise AI agents. The software enables companies to create autonomous AI agents that can execute complex, multi-step workflows, connect to internal enterprise systems, and deploy locally on Nvidia hardware to reduce cloud dependency for sensitive workloads.
Huang outlined a three-generation roadmap extending through 2028. The Vera Rubin platform is currently in production, with a mid-cycle refresh called Vera Ultra anticipated in the second half of 2027. The next-generation Feynman platform, scheduled for 2028, will utilize TSMC's A16 1.6-nanometer process technology and introduce silicon photonics for inter-chip communication that could deliver a 10x bandwidth improvement over current electrical interconnects.
The announcements drew immediate support from Wall Street. Bank of America Securities reiterated a "Buy" rating with a $300 price target, while Wells Fargo analysts recommended buying the stock ahead of the conference. Of 61 analysts covering Nvidia, 58 rate it a buy or strong buy, with a consensus target price of $267.54, according to data compiled by Bloomberg.
The company's financial position remains robust, with trailing twelve-month revenue of $216 billion and a profit margin of 55.6%. Nvidia's quarterly earnings growth stands at 95.6% year-over-year, though that pace has slowed from the triple-digit growth rates of earlier AI boom periods.
Nvidia also announced a strategic partnership with Thinking Machines Lab, a multiyear deal to deploy at least one gigawatt of Vera Rubin systems for frontier model training. The partnership underscores how Nvidia is evolving from selling individual chips to providing complete AI infrastructure packages as competition intensifies.
"Nvidia is executing on a strategy that goes well beyond selling GPUs," said analysts at Morgan Stanley in a recent note. "The combination of hardware, software, and now enterprise agent platforms creates a moat that will be difficult for competitors to breach."
The GTC conference runs through March 19, with additional sessions focused on robotics, digital biology, and automotive applications expected throughout the week. Nvidia's ability to maintain its innovation lead while expanding into new markets will be crucial as the company faces increasing scrutiny from regulators and mounting competition from both traditional chipmakers and cloud providers developing their own silicon.