How did the NVIDIA research laboratory once help become a company worth $ 4 trillion

When Bill Dali joined the NVIDIA research laboratory in 2009, it was employed only about ten people and focused on beam tracking, a display technique used in computer graphics.

The research laboratory, which was once more than 400 people, helped and helped convert NVIDIA from starting GPU in the video game in the 1990s to 4 trillion dollars that feed the mutation of artificial intelligence.

Now, the company’s research laboratory has its attention to developing the technology needed for robots and AI. Some of these laboratory works already appear in the products. The company revealed the two a New models of artificial intelligence in the worldAnd libraries and other infrastructure for robotics developers.

Daly, the chief scientist at NVIDIA, began consulting for NVIDIA in 2003 while working in Stanford. When he was ready to step down from being the head of the PC department in Stanford a few years later, he intended to make a vacation. Nafidia had a different idea.

David Kirk, who was running the research laboratory at the time, and CEO of Nvidia, Jensen Huang, believed that a more permanent position in the research laboratory was a better idea. Dally Techcrunch that the husband put a “full court printing press” about the reason for joining the NVIDIA research laboratory in the end.

Dali said: “It has ended with the perfect suitability for my interests and talents.” “I think everyone is always looking for the place in life where they can make the biggest, as you know, a contribution in the world. I think for me, it is definitely Nvidia.”

When Dally seized the laboratory in 2009, the expansion was first and foremost. The researchers started working in areas outside the rays immediately, including the design of circles and VLSI, or very widely integration, a process that combines millions of transistors on one chip.

The research laboratory has not stopped expanding since then.

“We are trying to find out what will happen positively for the company because we constantly see sexy new areas, but some, as you know, are doing a great job, but we have a problem to say if we will be very successful in this,” Dali said.

For a while this was the construction of the graphics processing units better for artificial intelligence. NVIDIA was early in the future of artificial intelligence and began to tamper with the idea of Amnesty International in 2010 – more than a decade before the current crazy intelligence.

Dali said: “We said this is amazing, this will completely change the world.” “We have to start doubling this, and Jenson believed that when I told him that. We started specialists in our graphics processing units and develop a lot of programs to support them, and to participate with researchers around the world who were doing it, a long time before clearly.”

The focus of material artificial intelligence

Now, since NVIDIA occupies a leadership progress in the GPU AI market, the technology company has started searching for new areas of demand after artificial intelligence data centers. This research led to NVIDIA to artificial intelligence and physical robots.

“I think in the end that robots will be a huge player in the world and we want to be basically the brains of all robots,” said Daly. “To do this, we need to start, as you know, develop the main technologies.”

This is where Sanja Fidler, Vice President of AI Research at NVIDIA. Fidler joined the NVIDIA research laboratory in 2018. At that time, she was already working on robot simulation models with a team of students at the Massachusetts Institute of Technology. When Huang told what they were working on at the reception of researchers, he was interested.

“I couldn’t resist joining,” Fidler told Techcrunch in an interview. “It is just like this, it is just a great topic, and at the same time it was also such a wonderful culture. As you know, Jensen told me, come with me, not with us, and not ours, do you know?”

She joined NVIDIA and got a work to create a research laboratory in Toronto called omniverse, the NVIDIA platform, which focused on building simulations for material AI.

Fidler said the first challenge to build these simulation worlds is to find the necessary 3D data. This included finding the appropriate size for the potential images to use and build the technology needed to convert these images into 3D deportations that can be used by simulation.

“We have invested in this technology that is called a breakfast presentation, which makes a main amendment to artificial intelligence, right?” Fidler said. “(From) goes from the 3D means to the image or video, isn’t it? We want to go in the other direction.”

World models

Omniverse released the first version of its model, which converts the images into 3D models, Ganverse3dIn 2021, then I got the work to discover the same process. Fidler said they used videos of self -driving robots and cars to create these 3D models and simulation through Nervous nervous reconstruction engineThe company announced for the first time in 2022.

She added that these technologies were the backbone of the company Cosmos family ai world It was announced in CES in January.

Now, the laboratory focuses on making these models faster. When you play a video or simulation game, you want the technology to be able to respond in actual time, although the robots make the reaction time faster.

“The robot does not need to see the world at the same time, in the same way that the world works,” said Fidler. “It can see it like 100x faster. So if we can make this model much faster than it is today, they will be very useful for automatic or material artificial intelligence applications.”

The company continues to make progress in this goal. Nvidia has announced a fleet of Models of the new world of Amnesty International It is designed to create structural data that can be used to train robots at the Siggraph Computer Graphics Conference on Monday. NVIDIA has also announced new libraries and infrastructure programs also aimed at robot developers.

Despite progress – the current noise around robots, especially humans – the NVIDIA research team is still realistic.

Dally and Fidler said that the industry is still at least a few years of human presence in your home, as Fidler compared it to the noise and the schedule in terms of independent vehicles.

Dali said: “We are making tremendous progress and I think you know that artificial intelligence was really the empowerment of the empowerment here.” “Starting with the visual artificial intelligence of the robot’s perception, then you know the intrusive artificial intelligence, this is a great value for planning and manipulating the task and movement. We solve each of these individual small problems and how much data we have to train our networks, these robots will grow.”