NVIDIA unveils AI agent PCs

Nvidia used its Computex keynote in Taipei on June 1 to push a new idea: AI agents should run on personal computers, not only in cloud data centers. The company said Microsoft, Dell and HP are among the partners building what it called “AI agent PCs,” a category of Windows devices designed to run more capable local AI software. Jensen Huang, Nvidia’s chief executive, tied the move to a broader push beyond graphics processors and into CPUs, saying the company is chasing a $200 billion market. Reuters reported on June 2 that Huang also said Nvidia has enough supply to support strong growth in both CPUs and GPUs despite industry constraints. ### What is Nvidia actually launching in these PCs? Nvidia said the new systems are built around its RTX Spark platform, which the company described as a way to “reinvent” Windows PCs for personal AI agents. Nvidia’s GeForce unit said the platform is aimed at slim laptops and compact desktops and combines the company’s AI and graphics software stack, including CUDA and TensorRT, for local inference on PCs. (techcrunch.com) TechCrunch reported that Nvidia presented the machines as “AI agent PCs” with Microsoft, Dell and HP, framing them as a route into personal computing after years of AI demand centered on data-center chips. The company’s own Computex coverage said local AI agents on RTX systems are becoming faster and more capable across the Windows ecosystem. (nvidia.com) ### Why is Jensen Huang talking about CPUs now? Huang used the Computex push to connect personal AI devices with Nvidia’s CPU ambitions. TechCrunch reported that he described CPUs as a roughly $200 billion market, arguing that Nvidia can now sell both CPUs and GPUs as AI workloads spread beyond traditional servers. (techcrunch.com) Nvidia launched Vera in March as what it called its first CPU “purpose-built” for agentic AI and reinforcement learning. In separate company announcements, Nvidia said Vera is intended for AI systems that plan tasks, use tools, run code and validate results, and that major manufacturers including Dell Technologies, HPE, Lenovo and Supermicro plan to adopt it in standalone systems. (techcrunch.com) ### Are these consumer laptops, enterprise boxes, or something in between? Microsoft’s role points to Windows PCs, while Dell and HP point to mainstream computer makers that already sell business and consumer machines at scale. Nvidia’s language around “personal AI agents” suggests the company wants AI tasks to run closer to the user on a local device, even as the heaviest model training remains in data centers. TechCrunch said that pitch is aimed at expanding Nvidia’s reach from infrastructure into the devices people use directly. (investor.nvidia.com) Nvidia’s partner list also spans enterprise hardware makers and PC brands, which suggests the rollout will not be limited to one narrow device type. The company said RTX Spark products include laptops, desktops and partner-built systems, while Vera-related announcements named server and infrastructure vendors as early adopters. That places the launch across both edge and enterprise computing. (techcrunch.com) ### Does Nvidia have enough chips to support another product push? Reuters reported on June 2 that Huang said Nvidia has the capacity to supply “robust” growth in both CPUs and GPUs despite constraints. That comment addressed a central investor question as Nvidia adds new categories while demand for AI chips remains high across cloud and enterprise markets. (nvidianews.nvidia.com) Nvidia’s recent announcements show the company widening its product map at the same time. In the past two months it has introduced Vera for AI systems, highlighted early deliveries of Vera systems to companies including OpenAI and Oracle Cloud Infrastructure, and used Computex to promote RTX Spark for PCs. (techcrunch.com) ### What should readers watch next? Dell, HP and Microsoft are the named participants to watch because Nvidia tied the AI PC effort to those partners at launch. Nvidia has also said RTX Spark systems will come through its broader PC ecosystem, while Vera adoption is expected from manufacturers including Dell Technologies, HPE and Lenovo. (investor.nvidia.com) Computex 2026 runs this week in Taipei, and Nvidia’s partner announcements there are likely to be followed by product-specific disclosures on models, pricing and ship dates. Reuters’ June 2 interview with Huang and Nvidia’s Computex updates are the clearest markers for the next round of details. (blogs.nvidia.com) (techcrunch.com)