Analyst(s): Ray Wang
Publication Date: August 27, 2025
NVIDIA has announced the general availability of its Jetson AGX Thor developer kit and production modules, a powerful new robotics computer designed for physical AI. Powered by the NVIDIA Blackwell architecture, Jetson Thor delivers a significant leap in performance, aiming to accelerate the development of general robotics across numerous industries.
What is Covered in this Article:
- NVIDIA announced the general availability of its Jetson AGX Thor developer kit and production modules, designed to accelerate the development of physical AI.
- The new system delivers a 7.5x increase in AI compute and 3.5x greater energy efficiency, enabling real-time generative AI processing at the edge.
- Leading technology and robotics companies, including Amazon Robotics, Boston Dynamics, and Figure, are early adopters of the Jetson Thor platform.
- The platform’s integration with NVIDIA’s full software stack, including Isaac and Holoscan, provides a comprehensive ecosystem for robotics development.
The News: On August 25, 2025, NVIDIA announced that the Jetson AGX Thor developer kit and production modules are now available. This new robotics computer supports physical AI and general robotics, allowing robots to interact with people and their surroundings in real time. The developer kit costs $3,499, and the production modules are ready for enterprise-level deployment.
Jetson Thor runs on NVIDIA’s Blackwell GPU architecture, delivering up to 2,070 FP4 teraflops of AI performance with 128GB of memory, all within a 130-watt power limit. Compared to its predecessor, Jetson Orin, it offers 7.5 times more AI power and is 3.5 times more energy efficient. This huge performance boost lets it handle complex generative AI models right on the device – something that is key for advanced robotics.
Is NVIDIA’s Jetson Thor the New Brain for General Robotics?
Analyst Take: The release of Jetson Thor is a significant step forward for the robotics world, tackling the big need for high-level compute in physical AI. By offering server-level performance in a small, energy-efficient form, NVIDIA gives developers the tools to build smarter autonomous machines. It also means large generative AI models can now run at the edge – no longer stuck in the cloud – making real-time thinking and reacting in changing environments possible. With backing from big names like Amazon Robotics, Boston Dynamics, and Figure, it’s clear there is emerging demand for an offering like this.
Zooming out, this marks the first time in NVIDIA’s history that its most advanced architecture is being deployed across every major platform. This milestone underscores the company’s strategic push toward broad compatibility, expanding its business portfolio, and demonstrating its ability to execute new technologies across diverse platforms and application domains.
Enabling a New Class of AI Models at the Edge
What sets Jetson Thor apart is its ability to run several generative AI models – including large language and vision-language models – directly on the device. This is especially important for tasks that cannot afford delays, like humanoid robots navigating unpredictable spaces or surgical robots processing sensor data on the fly. With 7.5 times more AI compute than before, it can handle multiple sensor inputs simultaneously without needing the cloud. This switch to edge computing solves a major problem in robotics – how to interact with the real world smoothly and intelligently. It is a big step toward making robots more capable and autonomous across different sectors.
Broad Industry Adoption Signals Market Acceleration
The fact that so many top companies are already using Jetson Thor shows how ready the market is. Agility Robotics and Figure are using it in humanoid robots, and Figure’s CEO Brett Adcock called it essential for perception, reasoning, and action in tough settings. In logistics, Amazon Robotics’ Tye Brady pointed to its role in helping scale AI robots worldwide. Caterpillar CEO Joe Creed also highlighted its use for real-time decisions in autonomous construction and mining gear. This quick, widespread adoption across sectors like manufacturing, logistics, and heavy machinery suggests Jetson Thor could become a key part of the future of physical AI.
An Integrated Ecosystem to Lower Barriers to Entry
More than just hardware, NVIDIA has built out a whole ecosystem that makes it easier for developers to get started. Jetson Thor works seamlessly with NVIDIA’s full software lineup, including Isaac for simulation, the Isaac GR00T humanoid models, and Holoscan for handling sensor data. With support from a developer base of over 2 million and more than 150 partners, this setup helps teams accelerate development cycles. In a field where building smart systems is tough, having a ready-made suite of tools gives NVIDIA a major edge.
A Strategic Play for a New Growth Pillar
Even though robotics is still a small slice—just 1%—of NVIDIA’s revenue, Jetson Thor is clearly part of a bigger plan to fuel the company’s next wave of growth. The robotics division is already growing fast, with 72% YoY growth, and CEO Jensen Huang sees robotics as the biggest opportunity outside of AI. By offering a powerful, easy-to-use platform, NVIDIA sets the stage for a whole new market that it can lead as it grows.
What to Watch:
- The commercialization of academic research using Jetson Thor will be a key validator of its long-term market success.
- NVIDIA’s delivery of promised software optimizations for FP4 and speculative decoding is crucial for maintaining its performance lead.
- The company must clearly differentiate its Jetson and DRIVE platforms to avoid market confusion as the robotics and automotive sectors converge.
- The entry of competitors with custom silicon could challenge NVIDIA’s market position if they offer more cost-effective or specialized alternatives.
See the complete news summary on the NVIDIA Jetson AGX Thor on the NVIDIA website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other insights from Futurum:
Is NVIDIA’s RTX PRO 6000 Blackwell the Tipping Point for Enterprise AI Acceleration?
Will NVIDIA and AMD Pass the 15% Fee to Chinese Buyers?
NVIDIA Q1 FY 2026 Revenue Jumps 69% Despite China Export Setback
Image Credit: NVIDIA
Author Information
Ray Wang is the Research Director for Semiconductors, Supply Chain, and Emerging Technology at Futurum. His coverage focuses on the global semiconductor industry and frontier technologies. He also advises clients on global compute distribution, deployment, and supply chain. In addition to his main coverage and expertise, Wang also specializes in global technology policy, supply chain dynamics, and U.S.-China relations.
He has been quoted or interviewed regularly by leading media outlets across the globe, including CNBC, CNN, MarketWatch, Nikkei Asia, South China Morning Post, Business Insider, Science, Al Jazeera, Fast Company, and TaiwanPlus.
Prior to joining Futurum, Wang worked as an independent semiconductor and technology analyst, advising technology firms and institutional investors on industry development, regulations, and geopolitics. He also held positions at leading consulting firms and think tanks in Washington, D.C., including DGA–Albright Stonebridge Group, the Center for Strategic and International Studies (CSIS), and the Carnegie Endowment for International Peace.
