Jensen Huang: Vera Rubin is in full-scale production, AI Agents are a key focus, challenging Intel and aiming for the next-generation AI PC entry point

Author: Wall Street Insights

On June 1, Huang Renxun, founder and CEO of NVIDIA, delivered a keynote speech on the topic of NVIDIA GTC Taipei 2026 in Taipei, China.

In this speech, Huang Renxun focused on unveiling multiple new products and platforms, including the Vera Rubin architecture, the AI Agent-oriented CPU Vera, the AI model Nemotron 3 Ultra, and the AI factory platform DSX, among others.

At the same time, NVIDIA also sent further signals about its move into the next-generation AI PC market, targeting the entire Windows camp.

NVIDIA disclosed at the press conference that major China-based automakers and autonomous driving companies—including BYD, Geely, Zeekr, Xiaomi, and Pony.ai—have already adopted or are building intelligent driving solutions based on the NVIDIA Hyperion platform.

NVIDIA’s stock price rose 2.7% in after-hours trading in the US market.

Vera Rubin has entered full-scale mass production

Huang Renxun said that the Vera Rubin architecture has already fully moved into the mass-production stage, with OpenAI, Anthropic, and SpaceX becoming NVIDIA Vera chip’s first customers.

He also noted that AI Agents will become a key area of focus going forward, and NVIDIA will launch CPU Vera for AI intelligences.

Huang Renxun further emphasized that NVIDIA Vera BlueField-4 STX brings proxy AI storage processing capabilities with chip-level security.

Aiming at the next-generation AI PC gateway

Notably, NVIDIA announced a new processor for personal computers running WINDOWS systems, challenging Intel.

NVIDIA will collaborate with MediaTek to develop the RTX SPARK personal computer chip, which will be manufactured by TSMC. Huang Renxun revealed that NVIDIA’s PC chips will hit the market this fall and will be used in both laptops and desktop computers.

Huang Renxun said NVIDIA will release a new generation of personal computer (PC) chips alongside each generation of AI processors.

Previously, NVIDIA’s official social media account issued a three-character teaser—"A new era of PC"—aiming at the next-generation AI PC gateway.

Analysts say NVIDIA intends to integrate the CPU, GPU, and AI units into a single SoC, directly delivering a complete core computing platform to OEMs such as Dell and Lenovo. Its position in the value chain will undergo a fundamental change. And this is exactly the vertical integration path taken by Apple’s Apple Silicon, except the target is the entire Windows camp.

NVIDIA: To collaborate with Unitree Technology and other global humanoid robot manufacturers to develop robots

In addition to collaborating with China’s Unitree Technology, NVIDIA plans to also work with humanoid robot manufacturers in the United States, Europe, and South Korea to conduct research robot projects.

Launching a new AI model + AI factory “action guide”

In addition, NVIDIA announced the launch of a new AI model, Nemotron 3 Ultra.

NVIDIA announced the NVIDIA DSX platform, providing infrastructure builders with a complete action guide to create AI factories. Huang Renxun said:

With the DSX platform, you can simulate the entire factory without spending a penny, verify performance before installing a rack, and operate with the reliability required for production-grade AI.

The Hyperion ecosystem for autonomous driving taxis emerges

NVIDIA DRIVE Hyperion has become a global autonomous driving taxi platform.

At the press conference, NVIDIA disclosed that major China-based automakers and autonomous driving companies—including BYD, Geely, Zeekr, Xiaomi, and Pony.ai—have adopted or are developing intelligent driving based on the NVIDIA Hyperion platform.

NVIDIA also launched Alpamayo 2, a super open inference model for autonomous driving taxis.

Building a development ecosystem for the robot era

In the humanoid robotics sector, NVIDIA is trying to play the role of an “infrastructure provider.”

At this event, NVIDIA introduced the NVIDIA Isaac GR00T humanoid robot reference platform for academic research, and also released large open-source intelligent agent toolkits and skill sets for Physical AI.

At the event, NVIDIA selected Unitree Technology’s H2 Plus as the reference model and introduced the Isaac GR00T reference humanoid robot platform. This platform integrates the Jetson Thor computing chip, the GR00T software stack, and dexterous hands as well as a full-body control system, providing a one-stop hardware-to-software development framework for robot manufacturers.