Nvidia Pushes Beyond Data Centers With AI PCs and Humanoid Robots

Nvidia is pushing beyond data centers and into personal computers and robotics. At Computex in Taipei, CEO Jensen Huang unveiled a new Arm-based PC processor, a humanoid robotics platform developed with Unitree, and confirmed that the Vera data center CPU has entered full production.

Together, the announcements highlight Nvidia’s strategy to build a full-stack AI ecosystem spanning cloud infrastructure, consumer devices, and physical AI.

Nvidia Unveils the Arm-Based N1X Processor and RTX Spark Superchip

Nvidia introduced its new N1X processor, integrated into the RTX Spark superchip, marking a direct challenge to Intel, AMD, Qualcomm, and Apple.

The platform combines a custom Arm-based CPU developed with MediaTek, Nvidia’s Blackwell GPU, and 128GB of unified memory. Built on TSMC’s 3-nanometer process, it is designed for advanced on-device AI workloads.

“This reinvention of the computer is as big of a deal as the reinvention of the phone into what we now know as the smartphone,” Huang said during his keynote.

Key Details of Nvidia’s PC Expansion

• Launch Partners: Microsoft, Dell, HP, ASUS, Lenovo, and MSI.
• Devices: Ultra-thin Windows laptops as slim as 14mm and compact desktops.
• Target Users: Content creators, AI developers, and gamers.
• Performance: Roughly comparable to an RTX 5070 laptop GPU.
• Architecture: Arm-based design aimed at improving efficiency for AI workloads.

Nvidia and Unitree Launch a Humanoid Robotics Platform

Nvidia is also expanding into physical AI through a new research platform built around Unitree’s upgraded H2 humanoid robot.

The system combines the H2 body with Nvidia’s Jetson Thor hardware, using a Blackwell GPU to run Isaac GR00T humanoid AI models and simulation software.

Standing nearly six feet tall and weighing 150 pounds, the robot features 31 degrees of freedom, while its Sharpa-built hands add another 25 degrees of freedom.

“We built this for higher education and university researchers, because for them to build this is insanely hard to do,” Huang said.

Early Adopters

The H2 Plus platform, scheduled for release in October, has already attracted interest from:

• Stanford Robotics Center
• ETH Zurich
• Ai2
• UC San Diego’s Advanced Robotics and Controls Laboratory

The announcement comes as Unitree seeks to raise 4.2 billion yuan ($620 million) through an IPO on Shanghai’s STAR board. More than 40% of its revenue already comes from outside China.

Vera CPU Enters Full Production

Nvidia also confirmed that its Vera CPU for data centers is now in full production and scheduled for release this fall.

According to Ian Buck, Nvidia’s VP of hyperscale and high-performance computing, Vera processes tokens 1.8 times faster than current x86 alternatives.

Early adopters include OpenAI, Anthropic, xAI, Oracle, and CoreWeave.

Looking Ahead

By bringing the N1X processor to PCs, launching Vera in data centers, and expanding Isaac GR00T through Unitree humanoids, Nvidia is building an AI ecosystem that spans cloud infrastructure, personal computing, and robotics.

The strategy positions the company to play a central role in the next phase of AI development, including what Huang describes as the rise of physical AI.