Nvidia has never sold you a PC chip before. That changes this fall.
At the Computex trade show in Taipei, Nvidia's CEO Jensen Huang unveiled the RTX Spark, a single chip that combines a 20-core processor, powerful graphics, and AI acceleration into one package. It is designed for slim Windows laptops and small desktops, and it is being built into devices from Dell, HP, Lenovo, Microsoft Surface, ASUS, and MSI. Acer and Gigabyte follow later.
The chip is built in partnership with MediaTek and manufactured on TSMC's most advanced 3nm process. It can hold up to 128GB of memory, far more than a typical laptop. That memory capacity is the key to the whole pitch: running large AI models entirely on the device, without sending data to the cloud.
The use case Nvidia is selling is AI agents. Not chatbots. Agents are software that can take a goal, break it into steps, open applications, read files, draft outputs, and complete tasks with minimal hand-holding. Think of a junior assistant who can sort through a week of emails, pull the relevant contracts, and prepare a summary, without you clicking through each step yourself. Right now, most agents run on remote servers. Nvidia's argument is that doing this locally is faster, cheaper per task, and keeps sensitive company data off third-party infrastructure.
For businesses in sectors where data confidentiality matters, such as insurance, legal, finance, or healthcare administration, local agents are a different proposition than cloud-based ones. The file never leaves the building.
This is also a larger strategic play. Nvidia posted $81.6 billion in revenue in its most recent quarter and forecast $91 billion for the next one. Despite that scale, almost all of its revenue comes from selling chips to data centres. Huang is now pushing into CPUs, a market he estimates at $200 billion. He has already sold $20 billion worth of a server-focused CPU called Vera this year alone. The RTX Spark is the consumer-facing end of that same strategy.
The history here carries weight. In 2013, Microsoft launched a Windows device on an ARM-based Nvidia chip. It could not run standard Windows software, nobody bought it, and Microsoft wrote off $900 million. The lesson was simple: a Windows PC that cannot run Windows programs is not a Windows PC. This time, the situation is structurally different. Microsoft has spent several years improving its emulation layer so that older software runs on ARM-based chips. Over 100 software makers, including Adobe, have already signed on to support the RTX Spark. The surface-level similarity to 2013 hides a very different foundation.
Pricing remains the unknown. Nvidia sells a developer-targeted AI mini-computer with similar hardware for around $4,000. Add a premium OLED laptop chassis and the bill climbs. If RTX Spark laptops land above $3,000, adoption outside professional and creative markets will be slow. If Nvidia and its partners price aggressively, the picture changes fast.
For business operators, the practical question is not whether to buy one of these laptops today. It is whether the arrival of locally-run AI agents on standard business hardware, at whatever price these settle at, changes what your teams can do in the next one to two years. The answer is almost certainly yes. The question is just when it becomes affordable enough to matter for your specific situation.