Nvidia CEO Jensen Huang told a developer conference that computing is advancing at an “insane” rate.
Nvidia has announced its latest family of chips to power artificial intelligence, reinforcing its position as a leading supplier to the AI craze.
“We need a bigger GPU. So, ladies and gentlemen, we'd like to introduce you to a really big GPU,” CEO Jensen Huang said Monday at a developer conference in California. We have mentioned the essential graphics processor.
The event, dubbed “AI Woodstock” by Wedbush analyst Dan Ives, is a must-see on major tech companies' calendars due to Nvidia's unique role in the AI revolution that has swept the world since the introduction of ChatGPT. There are no scheduled dates. Late 2022.
“I want you to understand that this is not a concert, this is a developer conference,” joked Huang, who took the stage in a packed arena normally reserved for ice hockey games and concerts. Said.
Nvidia's powerful GPU chips and software are essential to creating generative AI, and rivals like AMD and Intel are struggling to match the power and efficiency of the company's blockbuster H100, which launches in 2022. Still struggling.
Apple, Microsoft, and Amazon are also developing chips with AI in mind, but so far they've struggled to get their hands on Nvidia's coveted products to deliver on their own AI promises. are doing.
This linchpin role in the AI revolution has driven Nvidia's stock price up nearly 250% over the past 12 months, putting the company behind Microsoft and Apple and ahead of Amazon when measured by market capitalization.
Not giving up, Nvidia told the audience of developers and technology executives that it would release even more powerful processors and accompanying software on a platform called Blackwell. Blackwell is named after David Blackwell, the first black scholar to be inducted into the National Academy of Sciences.
According to Nvidia, the Blackwell GPU was an AI “superchip” that was four times faster than previous generations when training AI models.
“The rate of advancement in computing is crazy,” Huang said.
It will also be 25 times more energy efficient, Nvidia says, a key claim when AI developments are criticized for their voracious demand for energy and natural resources when compared to traditional computing.
Unlike rivals Intel, Micron and Texas Instruments, Nvidia, like AMD, does not manufacture its own chips, relying primarily on subcontractors such as Taiwan Semiconductor Manufacturing Company.
This could be a potential weakness given geopolitical concerns between Taiwan and China, and the US has banned Nvidia from sending its most powerful chips to Chinese companies.
Nvidia also announced other AI developments, including a platform for training humanoid robots.
Project Gr00t is not named after the character Groot from the Guardians of the Galaxy movie, Nvidia said in a statement, describing it as “the world's first basic model of the human body.”
Nvidia says robots powered by Gr00t will be designed to understand what people say, imitate people's movements, and learn from experience how to interact with the world.
The model will allow robots to “learn from a small number of human demonstrations, allowing them to assist with daily tasks or emulate human movements simply by observing humans,” NVIDIA said. Stated.
Nvidia also said it is working with Apple to incorporate AI capabilities into its newly released Vision Pro spatial computing device.
The partnership comes amid pressure for Apple to show it's keeping pace with Amazon, Google, Meta and OpenAI when it comes to artificial intelligence.
Nvidia also announced its Earth-2 cloud platform for predicting climate change using AI supercomputer simulations.
