Nvidia NVDA 0.39%increase; green up pointing triangle is set to bring its new technology for powering humanoid robots to market in the first half of 2025, aiming to stake its claim in the fast-growing robotics sector, a senior company executive said.
The chipmaking giant Nvidia is leaning more heavily into robotics in 2025. More specifically, in the first half of the new year, confirms the Financial Times, Nvidia is launching a new generation of compact computers for humanoid robots called Jetson Thor.
The move, which was expected, is part of an evolving, years-long strategy. Nvidia doesn’t plan to compete directly with manufacturers like Tesla, but rather provide the “hundreds of thousands” of robot makers in the world with a kind of underlying OEM, as its vice president of robotics, Deepu Talla, told reporters last month in Tokyo.
As for why now, Talla tells the FT that the shift owes to two technological breakthroughs: the explosion of generative AI models, and the ability to train robots on them using simulated environments.
What Nvidia isn’t saying but is also plainly the case: its push into AI-powered robots comes as major customers like Amazon and Google work to lessen their dependence on the outfit’s AI chips by developing their own.
Nvidia does not break out robotics product sales, but it currently represents a relatively small share of overall revenues. Data centre revenue, which includes its sought-after AI GPU chips, made up about 88 per cent of its overall sales of $35.1bn in the group’s third quarter.
But Talla said a shift in the robotics market is being driven by two technological breakthroughs: the explosion of generative AI models and the ability to train robots on these foundational models using simulated environments.
The latter has been a particularly significant development as it helps solve what roboticists call the “Sim-to-Real gap”, ensuring robots trained in virtual environments can operate effectively in the real world, he said. “In the past 12 months . . . [this gap] has matured sufficiently that we can now carry out experiments in simulation, combining with generative AI, that we could not do two years ago,” said Talla. “We provide the platform for enabling all of these companies to do any of those tasks.
Talla joined Nvidia in 2013 to work on its “Tegra” chip, which was initially aimed at the smartphone market. However, the company quickly pivoted, with Talla overseeing the redeployment of about 3,000 engineers into “AI and autonomous training [for vehicles, for example]”. This was the genesis of Jetson, Nvidia’s line of robotic “brain” modules that emerged in 2014. Nvidia offers tools at three stages of robotics development: software for training foundational models, which comes from Nvidia’s “DGX” system; simulations of real-world environments in its “Omniverse” platform; and the hardware to go inside the robots as its “brain”.
Apptronik, which uses Nvidia’s technology throughout its development of humanoid robots, in December also announced a strategic partnership with Google DeepMind to improve its products. The global robotics market is currently valued at about $78bn, according to US market researchers BCC, and is projected to reach $165bn by the end of 2029. Amazon has already deployed Nvidia’s robotics simulation technology for three of its warehouses in the US, and Toyota and Boston Dynamics are among other customers using Nvidia’s training software.
David Rosen, who leads the Robust Autonomy Lab at Northeastern University, said the robotics market still faced significant challenges, including training the models and verifying that they will be safe when deployed. “As of right now, we don’t have very effective tools for verifying the safety and reliability properties of machine learning systems, especially in robotics.
This is a major open scientific question in the field,” said Rosen.
Above: Tesla’s humanoid robot, Optimus; Nvidia has said it supplies Tesla with tech to build the machines.
