NVIDIA CEO Jensen Huang officially announced the company’s entry into the PC processor market during the Computex Taipei 2026 keynote on June 1, launching the RTX Spark super chip and N1X processor in collaboration with Microsoft. The new products are built on TSMC’s 3nm process, with CPU cores co-developed by NVIDIA and MediaTek based on the Arm architecture. The chip integrates the Blackwell architecture GPU, N1X CPU, and AI computing unit into a single SoC, supporting up to 128GB of unified memory. GPU performance is roughly on par with the existing RTX 5070 laptop class. The first wave of Windows PCs powered by the new chips will include over 30 laptops and 10 desktops, set to be released by Microsoft, Dell, HP, ASUS, Lenovo, and MSI in the fall of this year. The initial thin-and-light laptops can be as thin as 14mm, and the target market is initially the high-end segment for creators, AI developers, and gamers. Analysts point out that NVIDIA’s move parallels Apple’s vertical integration path with Apple Silicon, but is aimed at the entire Windows ecosystem, marking a fundamental shift in the company’s position within the PC value chain.
On the data center front, Huang announced that the Vera CPU, designed for AI agent inference scenarios, has entered full-scale production, achieving 1.8x the performance efficiency of traditional x86 architectures. OpenAI, Anthropic, SpaceX’s xAI, and Oracle have become the first core customers. The concurrently released Vera BlueField-4 STX provides chip-level secure agent-based AI storage processing capabilities. Additionally, NVIDIA unveiled the new AI model Nemotron 3 Ultra and the DSX platform for AI factory infrastructure builders — a platform that supports digital twin simulation of an entire factory before any hardware installation to verify performance. Huang stated that “you can simulate the entire factory without spending a single dollar.”