Unlock Editor’s Digest for free
FT editor Roula Khalaf has chosen her favorite stories in this weekly newsletter.
Nvidia is revolutionizing robotics through artificial intelligence, CEO Jensen Huang said Monday, outlining a vision for the next phase of the company’s phenomenal growth and calling it a “multi-trillion dollar” opportunity. I predicted that you would visit.
In his keynote speech at the annual Consumer Electronics Show in Las Vegas, Huang talked about “physical AI,” including AI models for humanoid robots and a major deal with Toyota to use Nvidia’s self-driving car technology. announced a series of new products and partnerships in the field. Vegas.
Nvidia has become one of the world’s most valuable companies, leapfrogging its $3 trillion market cap on the back of demand for AI chips. Mr. Huang has become widely known more than 30 years after founding Nvidia, a video game graphics chip company.
Lines formed outside the Mandalay Bay Convention Center long before the keynote began, with fans appearing on stage wearing glittery versions of his trademark leather jackets and exclaiming, “We’re in Las Vegas after all.” ” some people were still lining up.
Beyond semiconductors, Nvidia builds software that enables companies to train and deploy robots, from robots used in smart factories and warehouses to self-driving cars and humanoids, that run on its own chips. We are driving the expansion of AI use cases.
Resolving the technical challenges associated with large-scale deployment of robots will pave the way for “the biggest technology industry the world has ever seen,” Huang said.
Nvidia believes the field of robotics is reaching a technological tipping point by accelerating and fine-tuning the process by which AI simulates the physical world and generates the vast amounts of data needed to train robots. He said there was. According to the company, the humanoid robot market alone is expected to reach $38 billion over the next 20 years.
On Monday, Nvidia announced a set of basic AI models on its new Cosmos platform that developers can use for free to generate data and build their own models.
Nvidia said the underlying model, which it says was trained on 20 million hours of video data, is a fundamental technology development similar to the large-scale language models that power apps such as OpenAI’s ChatGPT. It works with Nvidia’s Omniverse platform, which is used to run simulations of the physical world.
“What[these models]are doing for language, they can now do for understanding the physical world,” said Rev Lebaredian, vice president of Omniverse and Simulation Technologies at Nvidia. told the Times. Although data in the physical world is much more difficult to collect and process than text, Lebaredian said it is a “necessary part” of the company’s mission.
“The big takeaway[from Huang’s CES speech]is that this moment is going to be special,” he added. “I think this year will be a tipping point where we will see an acceleration of physical AI and robotics.”
Currently, the Omniverse platform and robotics represent a small portion of the company’s overall revenue. In Nvidia’s quarter ending October, “professional virtualization” accounted for $486 million of revenue, while automotive and robotics revenue totaled $449 million.
That’s only a fraction of its overall revenue, as the company generated $30.8 billion in revenue during the same period selling data center chips that power its AI models.
Nvidia is exploring new markets as it faces increasing pressure from its biggest customers, including Amazon and Microsoft, which are rushing to build out their own AI data center chips.
Analysts at Bank of America said NVIDIA’s decision to double down on “physical AI” is “the next logical step.” The challenge, they added, will be to “make the product reliable enough, cheap enough, and widespread enough to create a reliable business model.”
At CES, Nvidia also announced a collection of basic humanoid robot models called “GR00T Blueprints.” It says this is a new tool for developing and testing fleets of factory and warehouse robots, as well as “significantly enhancing” robot development. Self-driving car training.
Recommended
Toyota announced it will build its next generation of self-driving cars on Nvidia hardware and software, known as Drive AGX. Self-driving vehicle group Aurora and auto parts maker Continental will use NVIDIA hardware and software to power thousands of driverless trucks in a long-term strategic partnership with the chipmaker.
Nvidia said it expects its automotive business to grow to $6 billion in fiscal 2026. Self-driving cars “will be the first multi-trillion dollar robotics industry,” Huang told the CES audience.
Separately, Nvidia announced it would release a “personal AI supercomputer” powered by its latest and most powerful AI chip, Blackwell. This allows researchers and students to run multi-billion parameter AI models locally rather than via the cloud. It will be available in May for an initial price of $3,000.
