Nvidia unveils a new chip to integrate AI directly into personal computers.

Jensen Huang, CEO of Nvidia, presented the new superchip named RTX Spark during his keynote address at Computex 2026 in Taipei on Monday, June 1. This chip combines the capabilities of a central processing unit (CPU) and a graphics processing unit (GPU) into a single component designed to power a new generation of so-called “AI PCs.” Microsoft and Dell confirmed that the first models of machines utilizing this superchip will be available for sale in the fall of 2026. Nvidia and Microsoft “will reinvent the PC,” Huang stated in his speech. Nvidia’s stock rose by 6% during trading on Wall Street on Monday, while Dell jumped by 10%. AMD and Intel saw declines.

The main goal of the RTX Spark is to enable the execution of artificial intelligence agents locally—that is, directly on the user’s machine, without a connection to a remote server. According to Huang, these agents would be able to “watch you, analyze your files, conduct research, and perform complex tasks.” Microsoft clarified in a separate statement that PCs equipped with the RTX Spark would be capable of running “high-performance AI models” and handling demanding workloads. “Here is the new PC,” Huang summarized. Neil Shah, an analyst at Counterpoint Research, described the announcement as a movement “revolutionizing how PCs will evolve in the next ten years.”

The local presence of AI models—without dependence on data centers—represents a significant architectural shift from the current generation of AI functions embedded in PCs, which primarily rely on cloud servers. Cristiano Amon, CEO of Qualcomm, also speaking in Taipei before the official opening of Computex, referred to 2026 as the “year of agents,” suggesting that the industry is transitioning from AI as a tool for answering queries to fully autonomous systems.

Vera CPUs for Data Centers and Humanoid Robots

Huang also announced during his speech that Nvidia’s new Vera processors for data centers are in full production and will be “our new major growth engine” amidst the boom of agent-based AI. The first expected clients include Anthropic, OpenAI, and SpaceXAI.

He also presented a reference design for a humanoid robot named Isaac GR00T, standing 1.83 meters tall and based on the chassis of the Chinese manufacturer Unitree (model H2). This reference design is introduced as a guide for research, particularly in higher education.

Huang had previously announced his intention to invest about $150 billion per year in Taiwan, which he describes as the epicenter of the AI revolution. He had also accompanied American President Donald Trump on a visit to Beijing to meet Xi Jinping as part of a delegation of American businesses. The Computex event runs until June 5 in Taipei.

Targeting a $200 Billion Market

During a results call in May, Huang noted that the new Vera processors give Nvidia access to a new $200 billion market. The RTX Spark is part of a diversification strategy beyond high-end graphics cards for data centers—a segment that has made Nvidia prosperous since 2022—into the consumer PC market, which has so far been dominated by Qualcomm (on ARM processors) and Intel (x86 processors) for embedded AI.

Nvidia’s entry into the PC market with its own superchip poses a direct threat to these two players, whose stocks fell on Wall Street on Monday in response to the announcement.