3 hours ago
2
For roughly 40 years, the way humans talk to computers has looked the same: keyboard, mouse, screen, repeat. Jensen Huang just called time of death on that entire paradigm.
Speaking at GTC Taipei on June 1, Nvidia’s CEO laid out a future where AI agents handle the doing while users handle the asking. No more clicking through menus, no more memorizing keyboard shortcuts, no more wrestling with five different apps to accomplish one task. In Huang’s framing, the interface itself becomes intelligent. Or as he put it: “AI is the UX.”
The hardware to back it up
Bold visions need silicon to stand on, and Nvidia brought receipts. The centerpiece announcement was the RTX Spark superchip, built in collaboration with Microsoft and designed specifically for Windows laptops and desktops.
The headline spec: up to one petaflop of AI performance. For context, a petaflop is one quadrillion floating-point operations per second. Compact desktop configurations are targeted at approximately 200 TOPS (trillions of operations per second) for high-end setups. Devices featuring the RTX Spark are expected to ship in fall 2026, with Dell, ASUS, and Lenovo named as production partners.
Huang described this as “the largest opportunity for our partners” in the context of reinventing personal computing.
What agentic AI actually means for your computer
Here’s the thing about the current PC experience. When you want to, say, plan a trip, you open a browser, search for flights, compare prices across tabs, switch to a calendar app, check your budget in a spreadsheet, then book through yet another interface. You are the orchestrator. Every app is a dumb tool waiting for instructions.
Nvidia’s vision flips that. You tell an AI agent: “Plan a weekend trip to Taipei under $2,000 that doesn’t conflict with my calendar.” The agent handles the rest, navigating applications, pulling data, making decisions, and coming back with options.
This is what Nvidia calls “agentic AI,” and the company showcased several platforms built around the concept. Among them were Vera Rubin, a next-generation AI platform, and Nemotron 3 Ultra, a model designed to power these autonomous workflows.
By running AI agents locally on the RTX Spark chip rather than routing everything through cloud servers, Nvidia is making a deliberate pitch around data privacy and efficiency.
Why this matters beyond the tech industry
The event ran concurrently with COMPUTEX, the massive Taipei-based tech trade show, which underscored just how central this announcement was to the broader hardware ecosystem.
Microsoft’s collaboration on the RTX Spark chip suggests deep integration with the Windows operating system. Dell, ASUS, and Lenovo are named as production partners for devices expected to ship in fall 2026.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·