In the face of increasingly fierce competition in the chip business, NVIDIA is betting on robot technology as the main driving force for its next round of growth.

On December 29th, the Financial Times reported that NVIDIA will increase its investment in robot technology by 2025, launching a new generation of compact computers for humanoid robots, Jetson Thor, in the first half of next year.

Analysts stated that NVIDIA's move is not surprising and is part of a strategy that has been evolving over the years—NVIDIA does not intend to compete directly with manufacturers like Tesla, but plans to provide a basic original equipment manufacturer solution for "thousands" of Global robot manufacturers, offering a complete technology stack from software for training AI robots to robot chips.

According to reports, NVIDIA provides tools in three stages of robot development: software for training foundational models (from NVIDIA's DGX systems); capabilities to simulate real-world environments on NVIDIA's Omniverse platform; and hardware that serves as the "core" of the robot.

Last month, NVIDIA Vice President and Head of Robot Business Deepu Talla told the Financial Times:

The "ChatGPT moment" of physical AI and robotic technology is approaching, and the market has reached a tipping point.

Talla also added that this shift is attributed to two technological breakthroughs: the explosive development of generative AI models and the ability to use these models to train robots in simulated environments. The latter is particularly significant as it addresses the "Sim-to-Real gap" problem that robotics researchers refer to, ensuring that robots trained in virtual environments can operate effectively in the real world. Talla stated:

In the past 12 months... the maturity of this gap has been sufficient to support our experiments combining generative AI in simulated environments, something that was not possible two years ago... We have provided a platform that enables all these companies to complete any relevant tasks.

Analysts also mentioned that NVIDIA's advancements in robotics come at a time when it faces increasing pressure from competitors like AMD in the chip sector, and many Cloud Computing giants like Amazon, Microsoft, and Google are also striving to reduce their reliance on NVIDIA.

Robotics technology is still in its early stages of development.

Currently, robotics technology remains an emerging field that has yet to generate substantial returns. According to BCC, a USA market research company, the global robotics market is currently valued at approximately $78 billion and is expected to reach $165 billion by the end of 2029.

Analysts pointed out that many robotics startups are facing significant challenges in scaling up, reducing costs, and improving the accuracy of robotic products. David Rosen, head of the Robust Autonomy Lab at Northeastern University in the USA, stated that the robotics market still faces major challenges, including how to train models and verify their safety in real-world applications.

Currently, we have not developed very effective tools to verify the safety and reliability attributes of machine learning systems, especially in the robotics field, which is an important scientific challenge in this area.

Although NVIDIA has not disclosed sales figures for robotic products separately, it can be certain that their proportion of total revenue is relatively small, as Datacenter revenue, including AI GPU, accounts for about 88% of NVIDIA’s total sales in the third quarter.