The new chips should be able to train AI models even faster and more efficiently. Nvidia boss Jensen Huang is already dreaming of a world that is completely controlled by artificial intelligence.
Nvidia CEO Jensen Huang dreams of a world controlled by AI.
Eric Risberg / AP
No technology is currently developing faster than artificial intelligence. The developers present new systems or programs almost every week. And now things should go even faster: The American company Nvidia, the undisputed market leader in the production of AI chips, presented its latest system on Monday evening in San Jose, California, which is supposed to train the models even faster.
No expense or effort was spared for the event. Nvidia invited analysts, developers and tech enthusiasts to present its new products in the SAP Arena, which has space for 16,000 people. “I hope you are aware that this is not a concert,” said CEO Jensen Huang at the beginning of his appearance, which he wore, as usual, in a black leather jacket.
The new chips are faster, bigger and more expensive
According to the Wall Street Journal, an analyst is said to have described the event as “AI-Woodstock”. And the star of the evening, alongside Huang, was the Blackwell chip system, which the CEO described as “driving a new industrial revolution” through AI. The system is four times more powerful than the current generation Grace Hopper when it comes to training artificial intelligence, said Huang.
This was already considered revolutionary. The Grace Hopper chips, which began shipping this year, can perform 8 quadrillion calculations in one second. Huang calculated that the Chat-GPT chatbot could be trained within three months with 8,000 Nvidia chips and a power consumption of 15 megawatts. With Blackwell you can do this in the same time with 2000 chips and 4 megawatts of electricity.
The chips are not only faster and more efficient than their predecessors, but also larger and more expensive. UBS analysts estimate they could cost up to $50,000, about twice as much as the previous generation. They are rarely used alone, but rather as part of a larger system. According to Huang, 72 chips will be bundled in computer systems with around 600,000 components and a weight of more than 1.3 tons. The systems should be available later this year and, according to Huang, customers include Amazon, Alphabet, Meta, Microsoft, Open AI and Tesla.
Nvidia is valued at over $2 trillion
It is the first time since 2019 that the Nvidia developer conference, officially known as GTC, will take place in person and not in digital form. Before the pandemic, Huang gave his big speech at the nearby San Jose State University gymnasium. A lot has happened since then: Nvidia has become the leading provider of chips for AI training, and the company’s valuation is now more than $2 trillion. This makes Nvidia the third largest listed US company, behind Apple and Microsoft.
Nvidia’s computer systems already dominate in data centers for AI training; their market share there is estimated at 80 percent. Now the group also wants to expand its role in generating content through artificial intelligence. The Blackwell system is 30 times as good as “Hopper,” emphasized Huang. Nvidia also has new software for this that can also be used via interfaces via the cloud.
At the conference, Jensen Huang was convinced that in the future most content would not be retrieved pre-prepared from memory, but that AI software would generate it fresh based on the respective situation. Nvidia developed the computer system for this future. For example, you will be able to talk to buildings via chatbot instead of looking at data somewhere.
Rockstar der KI
Nvidia’s meteoric rise also made Jensen Huang something of a rock star, with thousands of spectators coming to a company event. The Taiwanese, who came to the USA as a boy, is considered an ambitious and disciplined CEO who does not rest on his success. He is one of the co-founders of Nvidia and has led the company since 1993.
Due to the rise in Nvidia shares, Huang is sitting on a fortune of 69 billion dollars – but it was not foreseeable for a long time that it would come to this. In its early years, Nvidia was on the verge of failure, and before the big AI breakthrough, the company was repeatedly hit by fluctuations in the crypto markets, in which the chips also play a major role.
Even now, success cannot be taken for granted. The announcement of the new chip came as no surprise to analysts. The Nvidia share price did not make any big jumps after the event. Many analysts also expect Nvidia’s market share to decline this year as new products from competitors come onto the market.
But Jensen Huang is known for his attitude that a company should always be run as if it could fail at any time. According to the Wall Street Journal, he doesn’t have one-on-one conversations with his employees to avoid his messages being passed on incorrectly within the company.
Huang also likes to have everything under control at the GTC – and ventures into big territory. Nvidia wants to use its technologies to generate weather forecasts, said Huang. For this purpose, the company is developing a world simulation called Earth 2.
Huang also relies on robots, nine of which accompanied him during his presentation. “Everything that moves will be robotic,” said the company boss. The goal is for robots to be able to learn simply by observing people. “The Chat GPT moment for robotics may be just around the corner.”