Jensen Huang Just Delivered Incredible News for Nvidia Investors

Nvidia (NASDAQ: NVDA) supplies the world's champion graphics processing units (GPUs), which are the superior chips utilized to powerfulness the improvement of artificial quality (AI) software. Now, it's gearing up to vessel processors based connected its caller Rubin GPU architecture, which volition connection further increases successful show implicit its industry-leading Blackwell chips.

In different words, though its competitors are ramping up their efforts successful the AI accelerator space, Nvidia is chiefly competing with itself close now, and it's experiencing much request than it tin perchance meet. And during its Feb. 25 league telephone with investors pursuing the merchandise of its fiscal 2026 fourth-quarter results, CEO Jensen Huang made a bid of comments that suggest request could surge adjacent further from here.

Will AI make the world's archetypal trillionaire? Our squad conscionable released a study connected the 1 little-known company, called an "Indispensable Monopoly" providing the captious exertion Nvidia and Intel some need. Continue »

Nvidia's archetypal AI information halfway GPU, the H100, which launched successful 2022, was built connected the company's Hopper architecture. But its astir caller procreation of GPUs, based connected its Blackwell Ultra architecture, tin present 50 times amended show per watt than the H100 successful definite configurations. This highlights however accelerated the institution is progressing.

It's taking different leap guardant with the Vera Rubin platform, which includes the Rubin GPU, the Vera CPU, the latest NVLink 6 switches, and a scope of different networking components. The institution says the Rubin architecture is truthful almighty that AI models tin beryllium trained utilizing 75% less GPUs compared to Blackwell, portion reducing inference token costs by a whopping 90%.

Tokens are pieces of data, similar words and images, that an AI exemplary generates. These outputs outgo wealth to produce, which is wherefore galore AI companies complaint their customers based connected usage. When a idiosyncratic prompts a chatbot, for example, they are consuming tokens -- and the much analyzable the request, the much tokens it requires.

Bringing costs down volition promote much usage portion boosting the nett margins of AI companies, which is wherefore Nvidia Chief Financial Officer Colette Kress said, "We expect each unreality exemplary builder to deploy Vera Rubin." Samples person conscionable started shipping to customers, with wide accumulation scheduled to commencement successful the 2nd fractional of this year.

Nvidia generated $215.9 cardinal successful gross during its fiscal 2026, which ended Jan. 25, a 65% summation from the erstwhile year. Its information halfway sales, which roseate 68%, accounted for $193.7 cardinal of that total, reflecting the beardown request for AI GPUs.