NVIDIA vs Intel: The Battle for the Future of Humanoid Robots

追加: 2026-06-05

[00:00:00]The robotics and physical AI industry has just witnessed one of the most important infrastructure developments of 2026.

[00:00:08]For the past several years, the AI race has been dominated by large language models. Companies competed to build smarter chatbots, larger models, and more powerful AI assistants.

[00:00:20]But the next phase of the industry is becoming increasingly clear. AI must move beyond screens [music] and begin interacting with the physical world. And over the last week, two major announcements from Nvidia and Intel showed that this transition is already underway.

[00:00:37]On May 31st, Nvidia officially unveiled a new [music] humanoid robot reference design as part of its Isaac GR00WT ecosystem.

[00:00:48]The company combined [music] several key technologies into a single platform. The design includes the Unitree [music] H2 Plus humanoid robot, advanced five-finger tactile robotic hands from Shaerpa featuring 22° of freedom, and Nvidia's latest Jetson Thor onboard computer designed specifically for robotics and [music] embodied AI applications.

[00:01:12]The significance of this announcement goes far beyond the robot itself. Nvidia is attempting to create a standardized foundation for humanoid robotics.

[00:01:21]Instead of forcing every university, startup, and research lab to build a complete robot from scratch, the company is providing a ready-made hardware [music] and software platform that researchers can immediately use for AI training, [music] manipulation research, and autonomous behavior development.

[00:01:39]Leading institutions, including Stanford University and ETH Zurich, [music] have already gained access to the platform.

[00:01:47]In practical terms, Nvidia [music] wants to do for robotics what Android did for smartphones, create a common ecosystem that accelerates development across the entire industry.

[00:01:58]But almost immediately [music] after Nvidia's announcement, Intel responded.

[00:02:02]The company introduced its new Core Ultra Series 3 processor family for robotics [music] and launched OpenVINO physical AI, an open framework designed to support modern robot intelligence [music] systems.

[00:02:16]Intel's strategy is notably different.

[00:02:19]While Nvidia focuses [music] on highly specialized hardware optimized for maximum AI performance, Intel is emphasizing affordability and accessibility.

[00:02:29]According to the company, its processors can efficiently run advanced [music] vision language action models, or VLA models, while significantly reducing deployment costs.

[00:02:39]VLA models are currently considered one of the most promising approaches to robot intelligence. Unlike traditional robotic software, where every task must be programmed separately, VLA systems allow robots to combine visual perception, >> [music] >> language understanding, and physical actions within a single AI framework. A robot can observe its environment, understand a natural language instruction, and determine how [music] to execute that task in real time.

[00:03:08]This is one of the key technologies behind the next generation of humanoid robots.

[00:03:14]What makes Intel's move particularly [music] interesting is the economics. In data centers, companies are often willing to spend tens [music] of thousands of dollars on high-performance AI hardware. Robotics is different.

[00:03:26]Every component directly [music] affects the final cost of the robot. If humanoid robots are ever [music] going to scale beyond research labs and pilot programs, hardware costs must come down significantly.

[00:03:39]That creates an opportunity for Intel.

[00:03:41]By offering lower-cost [music] computing platforms and open software tools, Intel is positioning itself as an alternative to Nvidia's increasingly [music] dominant ecosystem.

[00:03:52]At this stage, however, it would be premature to declare a winner. The physical AI sector today resembles the large [music] language model market of 2022. The technology is advancing rapidly, investment is pouring in, and companies are racing to establish standards, yet no single platform has achieved complete dominance.

[00:04:12]For researchers, [music] developers, and robotic startups, this competition is actually beneficial. More competition means lower costs, [music] faster innovation, broader access to development tools, and ultimately a larger ecosystem.

[00:04:27]The most important [music] takeaway is that robotics is beginning to mature into a true platform industry. We're no longer seeing only individual robots being unveiled. We're seeing operating systems, AI frameworks, [music] standardized hardware platforms, simulation environments, and dedicated computing architectures designed specifically [music] for physical AI.

[00:04:49]And historically, these infrastructure layers often matter more than any single product. The companies that define the platforms frequently shape the future of entire [music] industries.

[00:05:00]Whether Nvidia, Intel, or another player ultimately takes the lead remains to be seen, but one thing is becoming increasingly clear.

[00:05:08]>> [music] >> The next major AI battle will not be fought exclusively inside data centers.

[00:05:13]It will be fought inside the robots that interact with the real world.