Today, Jensen Huang unveiled NVIDIA’s Jetson AGX Thor, a major leap forward in robotics. This tiny computer is for humanoid robots, self-driving cars as well as other smart machines, and has the performance to change how they interact with the world.
Built on the Blackwell GPU architecture, it has 2,070 teraflops of AI processing power – enough to do complex tasks in real time and use 130 watts of power. If you like the idea of machines that can see, think and move like humans, check this out.
Sale
Acer Nitro V Gaming Laptop | Intel Core i5-13420H Processor | NVIDIA GeForce RTX 4050 Laptop GPU | 15.6"...
- Beyond Performance: The Intel Core i5-13420H processor goes beyond performance to let your PC do even more at once. With a first-of-its-kind design,...
- AI-Powered Graphics: The state-of-the-art GeForce RTX 4050 graphics (194 AI TOPS) provide stunning visuals and exceptional performance. DLSS 3.5...
- Visual Excellence: See your digital conquests unfold in vibrant Full HD on a 15.6" screen, perfectly timed at a quick 165Hz refresh rate and a wide...
Humanoid robots have been in science fiction for years but turning them into reality has been a slow go. The challenge is giving robots the ability to process vast amounts of data – like sensor inputs, visual cues and language instructions – and make split second decisions. Jetson AGX Thor tackles this head on. With 128GB of memory and 7x the AI compute of Jetson Orin, it can run multiple AI models at once. So robots can scan their environment, understand spoken orders and perform precise actions all without missing a beat. Companies like Agility Robotics and Amazon Robotics are already using it to create machines that can explore warehouses or aid in manufacturing with more autonomy.
Robots, especially those in dynamic environments like hospitals or farms can’t be tethered to a wall outlet. Jetson AGX Thor has 3.5x the energy efficiency of Jetson Orin so it can do demanding tasks without draining batteries in minutes. So robots can do longer shifts whether sorting items for Amazon or assisting surgeons in hospital settings as Medtronic is investigating. Robots can do more with less power not only makes them possible but also scalable across industries.
NVIDIA’s Jetson software stack which runs Thor has a wide range of AI frameworks including vision based models and language processing. This flexibility allows engineers to tailor robots to specific tasks – a humanoid in a retail store or a tractor in a field for John Deere. Over 2 million developers are using NVIDIA’s robotics tools and Thor fits into the company’s ecosystem – from cloud based training to edge computing – so it’s no wonder.
The Jetson AGX Thor dev kit is $3,499, which is not cheap, but a good value for what you get. Building a custom system with the same features would take much more time and resources. The kit, which includes a reference board, power supply and cooling system, is available now, while the Drive AGX Thor for self-driving cars is available to order now with delivery in September.
