Nvidia reportedly considers 2025 as ChatGPT moment for robots
Nvidia is said to train robots using generative AI models
The company is also reportedly planning to train robots in simulation
Earlier this year, Nvidia unveiled an AI platform for humanoid robots dubbed GR00T
Photo Credit: Reuters
Nvidia is reportedly shifting its focus from artificial intelligence (AI) chipsets to robotics and embodied AI. The Santa Clara-based tech giant is said to be looking to provide processing capabilities and training for robots due to the rising competition in the AI hardware space. Notably, an Nvidia executive recently highlighted the company's plans to start training AI agents in simulation to prepare them for real-world scenarios. Additionally, earlier this year, the tech giant unveiled an AI platform, dubbed Generalist Robot 00 Technology (GR00T), for humanoid robots.
Nvidia Looks Towards Robotics as an Untapped Space
According to a Financial Times report, the tech giant is now positioning itself to be a major platform for robotics processing and AI-powered simulation training. Deepu Talla, Nvidia's Vice-President of Robotics told the publication that the “ChatGPT moment” for physical AI and robotics is about to arrive.
However, this shift of focus could also mark a deprioritisation of AI chipsets and GPUs, a division which is the main driver of the company's $3.6 trillion (roughly Rs. 307.9 lakh crores) valuation. The report claimed that the decision to develop new focus areas was made as the AI semiconductor space has witnessed increased competition in recent times.
Chipmakers such as AMD, Intel, and Snapdragon have all launched AI chipsets powered by neural processing units (NPUs). While the GPU space is still dominated by Nvidia, the entry of cloud computing giants such as Amazon, Google, and Microsoft could further take a significant chunk away from its revenue.
As per the report, Nvidia is now planning to position itself as a “full stack” solution for robotics that would include hardware acceleration as well as software for training. Recently, Jim Fan, head of the Embodied AI division at Nvidia, revealed that the company was planning to train embodied AI agents (AI agents equipped with sensors and mechanical parts) in simulation to prepare them for real-world challenges.
Nvidia is also reportedly gearing up to launch Jetson Thor, a chipset built by the company's Isaac Robotics Platform, that powers its GR00T platform for simulation-based training. The latter also draws from the capabilities of generative AI to create a new way to train humanoid robots that do not require extensive coding.
Talla told the Financial Times that the decision to focus on the robotics market was based on the rise of generative AI models and the ability to train robots on these models in simulated environments.
“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. We provide the platform for enabling all of these companies to do any of those tasks,” he added.
Nvidia Releases Cosmos-Transfer1 AI Model That Can Be Used for Simulation-Based Training for Robots24 March 2025
Search
![Gadgets 360 With Technical Guruji: Ask TG [ March 22, 2025]](https://c.ndtvimg.com/2024-12/5s52e4k_-ask-tg_640x480_14_December_24.jpg?downsize=180:*)
