英伟达公司近期推出其新一代人工智能（AI）芯片平台Jetson AGX Thor，旨在为机器人系统提供强大的“大脑”。该平台专为生成式推理模型设计，采用先进的Nvidia Jetson软件平台，能够让下一代物理AI代理在边缘实时运行，同时减少对云的依赖。Jetson AGX Thor拥有卓越的计算能力和能效，比前代产品Jetson Orin在AI计算方面提升了7.5倍，能效提升3.5倍。该平台支持多种主流的生成式AI和推理框架，包括Cosmos Reason、DeepSeek、Llama、Gemini等，并针对机器人应用优化。其售价为3,499美元起，已获Agility Robotics、Amazon Robotics、Boston Dynamics等众多行业领导者和顶尖研究机构的早期采用。

🚀 **Jetson AGX Thor：机器人AI的全新大脑** Nvidia推出的Jetson AGX Thor是一款革命性的物理AI和机器人平台，被誉为机器人系统的“新大脑”。它专为生成式推理模型设计，能够驱动下一代物理AI代理，如大型Transformer模型、视觉语言模型和视觉语言动作模型，实现边缘端的实时运行，显著降低对云的依赖。

💡 **性能飞跃与能效提升** 相较于前代产品Jetson Orin，Jetson AGX Thor带来了巨大的性能提升，AI计算能力提高了7.5倍，能源效率也提升了3.5倍。此外，它还拥有3.1倍的CPU性能和两倍的内存，能够更高效地运行复杂的生成式AI模型，处理高复杂度任务，如Agentic AI、高速传感器处理和通用机器人任务。

🌐 **广泛的框架支持与生态合作** 该平台全面支持业界主流的生成式AI和AI推理框架，包括Cosmos Reason、DeepSeek、Llama、Gemini、Qwen等，并针对机器人领域提供了Isaac GR00T N1.5等特定模型。Nvidia的CEO黄仁勋强调，Jetson Thor是推动物理AI和通用机器人时代发展的终极超级计算机，能够支持数百万机器人开发者。

🤝 **行业巨头与学术界的广泛采用** 众多行业领导者，包括Agility Robotics、Amazon Robotics、Boston Dynamics、Caterpillar、Figure、Hexagon、Medtronic和Meta，已成为Jetson Thor的早期用户。同时，斯坦福大学、卡内基梅隆大学和苏黎世大学等顶尖研究机构也正在利用Jetson Thor探索感知、规划和导航模型的边界，预示着其在未来机器人技术发展中的广阔应用前景。

TMTPOST -- Nvidia Corporation stock gained around 1% on Monday, outperforming the U.S. stock market as the broader benchmark S&P 500 index closed 0.43% lower. Shares extended their rally last Friday after the artificial intelligence (AI) chip behemoth released its new-generation platform that can serve as the brains for robotic systems.

Credit：Nvidia

Nvidia on Monday introduced the Jetson AGX Thor, a platform for  platform for physical AI and robotics that the company described as “a new brain” for robots across research and industry. The platform consists of a Jetson AGX Thor developer kit and production modules.

Built for generative reasoning models, Jetson Thor is powered by the full-stack Nvidia Jetson software platform andenables the next generation of physical AI agents — powered by large transformer models, vision language models and vision language action models — to run in real time at the edge while minimizing cloud dependency. 

Nvidia said the platform supports all popular generative AI frameworks and AI reasoning frameworks and AI reasoning models, including Cosmos Reason, DeepSeek, Llama, Gemini and Qwen models, as well as domain-specific models for robotics like Isaac GR00T N1.5

Nvidia described the platform as Blackwell-powered robotics computer delivering 2,070 FP4 teraflops to tackle complex applications including agentic AI, high-speed sensor processing and general robotics tasks.It unlocks  real-time reasoning inference, critical for highly performance physical AI applications.

Compared with its predecessor, the Jetson Orin, the Jetson Thor delivers 7.5 times more AI compute, 3.5 times greater energy efficiency to run any generative AI model. It also has 3.1 times more central processing unit (CPU) performance and two times more memory of its predecessor.

Jetson Thor is on sale on Monday, starting at $3,499. It includes a Jetson T5000 module, a reference carrier board with abundant connectivity, an active heatsink with a fan and a power supply.developer kit and production modules. The Jetson T5000 modules are available starting at $2,999 for 1,000 units.

“We’ve built Jetson Thor for the millions of developers working on robotic systems that interact with and increasingly shape the physical world,” said Nvidia CEO Jensen Huang. “With unmatched performance and energy efficiency, and the ability to run multiple generative AI models at the edge, Jetson Thor is the ultimate supercomputer to drive the age of physical AI and general robotics.”

Early adopters include industry leaders Agility Robotics, Amazon Robotics, Boston Dynamics, Caterpillar, Figure, Hexagon, Medtronic and Meta, while 1X, John Deere, OpenAI and Physical Intelligence are evaluating Jetson Thor to advance their physical AI capabilities. Research labs at Stanford University, Carnegie Mellon University and the University of Zurich are tapping Jetson Thor to push the boundaries of perception, planning and navigation models for a host of potential applications.

更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App