At Computex 2026 in Taipei, Nvidia made a move few expected: CEO Jensen Huang unveiled RTX Spark, the company's first consumer PC processor. It's not just a new chip — it's a direct challenge to the $200 billion CPU market that Intel and AMD have dominated for decades.
What happened
RTX Spark is a 1-petaflop ARM-based superchip that combines CPU, GPU, and RAM in a single package. It's built to run AI agents and large language models locally on the device, without relying on cloud connectivity. Nvidia and Microsoft co-developed a secure execution sandbox that keeps agents running safely and in isolation on consumer hardware.
Who's on board
Six major PC manufacturers are already committed. RTX Spark-powered laptops are due in fall 2026 from:
- Microsoft — Surface Laptop Ultra, described as the most powerful Surface ever built
- Dell and HP
- ASUS, Lenovo, MSI
- Acer and Gigabyte models to follow shortly after
Why this matters
Until now, Nvidia's consumer business has been defined by GPUs for gaming and, more recently, AI workstations. RTX Spark is a bet on a much bigger opportunity: an estimated 500+ million aging PCs are due for replacement, and Nvidia wants the next generation to be agent-native. CEO Jensen Huang explicitly called CPUs the bottleneck for agentic AI workflows, framing Intel and AMD's traditional x86 designs as out of step with the new era.
Context and background
ARM-based Windows laptops have been tried before — Microsoft and Nvidia's Surface RT in 2013 was a notable failure. But the landscape has shifted dramatically. Apple's M-series proved that ARM dominates laptop performance; Microsoft has continuously improved Windows on ARM; and the rise of agentic AI is driving real demand for powerful local compute. Nvidia also debuted Nemotron 3 Ultra (550 billion parameters, open source) at the same Computex event — a strong signal that RTX Spark machines will have world-class models to run from launch day.
What's next
The first RTX Spark laptops arrive in stores in fall 2026. That's when the real test begins: will consumers pay a premium for an agent PC, and can Nvidia replicate in the CPU market what it achieved in GPUs? Analysts warn that if agentic workloads become mainstream, Intel and AMD could face the same kind of disruption they once delivered to others.
