In the space of only a few years, Nvidia (Nvda) has become one of the most important flea companies in the world. Revenues has become soaked from $ 27 billion in the company 2023 of the company 2023 to $ 130.5 billion in its financial year 2025. Fig ratings also more than 680% since January 2023.
Not quite a familiar name like the other large technological companies, NVIDIA is at the center of the global push of AI thanks to its powerful chips, Including the ultra blackwellthat the company showed during its annual GTC event on Monday.
A certain number of technologies behind these processors, those which feed the PCs in the world and the software which both manage from the relatively small department of the research and development of Nvidia: Nvidia research named appropriately.
Created in 2006, the group is responsible for everything Nvidia rays tracing technologywhich creates realistic lighting for professional players and designers, in NVLink and NVSWITCHwhich allow graphic fleas and central processing units (CPU) to communicate to the type of speed necessary for AI systems.
Spectrum-X and Quantum-X photonics and NVIDIA photonic networking switches. (Image: nvidia) ·Nvidia
Currently, the organization is working on new flea architectures, quantum IT and software simulators that teach robots and autonomous cars How to sail in the real world.
Everything is supposed to continue to push Nvidia forward at a time when it already rises high. And to do this, the research team adopted the desire to fail most often while giving promising projects the time they need to succeed, no matter how long it takes.
“We have to realize that most of the things we start in research fail, and that’s actually a good thing,” said Bill Dally, Vice-President Director of Research and Chief Scientist at Nvidia. “I tell people, you know, if everything you do succeeds, you don’t swing for fences. You rush. “
While Nvidia has developed a number of impressive technologies over the years, the company’s research team is not as big as some of the other Silicon Valley companies.
“We are a tiny fraction of the size of competitive research laboratories,” said Dally. “We have 300 (people) and yet, I think that in things that matter, we hit well above our weight. And I think that the real measure of this, for me, is our impact over the years on improving things on a product (marketable). ”
According to Dally, the best researchers are those who offer an idea, test it and, if it does not work, abandon it without wasting resources.
But if a concept seems to take place, the company will continue to keep it away until it is a valid product or technology.
Spectrum-X and Quantum-X photonics and NVIDIA photonic networking switches. (Image: nvidia) ·Nvidia
Nvidia’s rays tracing is a perfect example. The product has taken 10 years to develop, but is now used in hundreds of major games and in design software.
“I think it is quite extraordinary that the company was able to follow a vision that has taken more than 10 years to implement,” said Bryan Catanzaro, vice-president of research applied in Deep Learning in Nvidia.
“AI is the most important example,” said Catanzaro, who joined Nvidia as an intern in 2008.
“The AI in 2011 was considered old and mute and dead. It’s like, people have been trying since the 1950s and it never worked, so why would it work now? But there were some of us who thought it was really an opportunity and therefore the company gave us space to continue trying things and then to produce increasingly better results, which then led to more placement, “added Catanzaro.
DLSS of Nvidia, or super sampling of Deep Learningis another example of a product that the company continued to continue despite the early difficulties. Introduced in 2019, the first iteration of DLSS improves the image quality of a game and performance using AI. But the software did not hit the brand out of the door. I remember trying it on my own computer and not seeing much improvement by playing games.
The NVIDIA DLSS4 improves game performance and graphic technology. (Image: nvidia) ·Nvidia
Quick advance for today, and the company now offers DLSS 4, which considerably improves game visuals for the most high intensity titles, including “Cyberpunk 2077”.
“DLSS 1.0 was not great, and many people thought it was a bad idea, it was bad technology. We believed in it,” said Catanzaro. “I think Nvidia just has this unshakable belief when he knows that something is true about the future, he continues to fight.”
Not all successful research projects are found as a product that directly generates income. However, they can help feed sales indirectly by stimulating GPU sales.
“I am perfectly satisfied with the people who develop … applications for GPUs that expand the market,” said Dally.
“Recently, our people have done this thing called Sana, which is this generative text network to (image). And so it is not going in a product, but it is always a great success because people outside use it, and therefore it feeds GPU demand.”
It is ultimately the goal. But the new Blackwell Ultra and Vera Rubin Superchip, recently unveiled, also came at a time when Nvidia faces increased competition. AMD offers its own AI chips designed to compete with Nvidia and the company’s customers develop or deploy their own specialized AI processors.
There are also bankrupt movements such as the release of the R1 AI model of Deepseek, which has sent the market capitalization of Nvidia Dive nearly $ 600 billion in Januaryand the unpredictability of government intervention, including prices and export controls, which continue to weigh the course of the company’s action.
And with technological companies like Amazon (Amzn), Google (Goog,, Googl), Meta (Meta), and Microsoft (Msft) For billions for AI infrastructure in the coming years, NVIDIA’s research efforts have become all the more important because it works to ensure its share of this premium.
He must simply continue to fail quickly and move forward.
Register for Yahoo Finance Week in Tech Newsletter. ·yahofinance
Send an email to Daniel Howley to dhowley@yahofinance.com. Follow him on Twitter at @Danielhowley.
