Every year, thousands gather in San Jose to watch CEO Jensen Huang give the company's keynote presentation, while thousands more tune in online for the livestream.
Early in, Huang emphasized the very clear momentum we're seeing in computing power. In a nutshell, demand and adoption continues to climb at a rapid pace. To keep up, compute power and innovation is more important than ever, whether that's for healthcare, autonomous driving, artificial intelligence or graphics.
As such, Huang teased a new acronym early in the presentation: PRADA. That is, PRogrammable Acceleration of Multiple Domains with one Architecture. Essentially, with its Cuda-X program, "software-acceleration libraries are integrated into all deep learning frameworks." This helps increase efficiencies and by doing so, developers and companies can drive down costs significantly.
A great example comes from the company's RTX platform, which allows for real-time ray tracing on its GPUs. "It's very clear that ray tracing has become the next step" in video game and graphic design, Huang said. As part of the next step forward, Nvidia has turned to partnering with other companies.
As such, Huang announced that Nvidia is partnering with LGU+ in South Korea and SoftBank (SFTBY) in Japan to bring datacenters to gamers in these markets. Nvidia plans to build and then maintain the servers.
As for autonomous driving, the company had some big announcements. First, Nvidia's Drive Constellation is now available.
Constellation could be a game-changer when it comes to autonomous driving. It allows self-driving developers to use virtual environments to test their self-driving system. To the vehicle's sensors, it's no different than running the vehicle in the real world. This product is vital to scaling autonomous driving, as it allows the testers to do all sorts of simulations. They can add or subtract traffic, operate during the day, at sunset or in pitch black, and they can change the weather to snow, rain or sun. Most importantly, they can run multiple simulations simultaneously and as many times as necessary.
"Autonomous driving is one of the greatest challenges," Huang said, whether that's from a computing, A.I. or systems integration standpoint.
However, Nvidia also announced a new end-to-end partnership with Toyota (TM - Get Report) . Toyota, one of the largest automakers in the world, "is partnering with Nvidia to create the future of autonomous vehicles," Huang said, saying it will range from deep learning, in-car systems, simulation, and more. Given the size and scope of Toyota, this is a big announcement for Nvidia, which has already partnered with Daimler (DDAIF) on multiple A.I. fronts.
Finally, Nvidia is launching Drive AP2X Release 9.0, which allows for high-function Level 2 driving capabilities. The company also demonstrated safety force field and real-time mapping capabilities.
When talking about A.I., Huang made several new announcements. The first is that the company will now automatically support tensor core mixed-precision in TensorFlow, PyTorch and MXNet. The second is that RAPIDS, a suite of software libraries, will now be supported by Microsoft (MSFT - Get Report) Azure and Google (GOOGL - Get Report) (GOOG - Get Report) Cloud, while Nvidia is partnering with Accenture (ACN - Get Report) to add RAPIDS to its applied intelligence platform.
Later in the presentation, Huang announced that the company has created its smallest computer ever built, the Jetson Nano. The computer, which starts at just $99, is a robotics computer which can be used for various applications. And of course, it's fully integrated with the Cuda-X platform. Nvidia also unveiled a new workstation for data scientists, which will be built in collaboration with some of the world's top computer makers like Dell (DELL) , Hewlett Packard Enterprise (HPE - Get Report) and Lenovo, among others.
Shares of Nvidia now up 0.47% in after-hours trading, after falling 0.51% during regular trade Monday.