For three decades, Intel basically owned the Windows PC. x86 architecture was so locked in that challenging it felt like trying to move a mountain. Then Apple proved you could actually do it—Apple Silicon showed the world that Arm could beat x86 at its own game. But Apple only moved Apple. Windows stayed stuck.
Until Nvidia decided to flip the entire board.
At Computex 2026, Jensen Huang just announced the RTX Spark Superchip—basically a data center crammed into a laptop. We’re talking 1 petaflop of AI compute, 6,144 CUDA cores, and 128 GB of unified memory in a thin chassis. Your next laptop won’t just run AI apps. It *will be* AI.
Here’s what makes this wild: Nvidia connected the CPU and GPU using NVLink C2C, the same ultra-high-speed interconnect they use to link chips in massive data center servers. They miniaturized data-center-grade engineering and stuffed it into a laptop. It’s ridiculous. It’s brilliant. And it’s coming to Dell, HP, Lenovo, Asus, MSI, and Microsoft’s Surface line this fall.
But here’s the thing—the hardware is almost beside the point.
The real weapon is CUDA. Over 15 years, every major AI model, every AI framework, every serious GPU-accelerated application got built on CUDA. It’s Nvidia’s deepest moat. And until now, it lived only in the cloud and high-end workstations. RTX Spark is the first Windows laptop chip to run the full CUDA stack natively. That means the entire AI ecosystem runs on your laptop, right out of the box, no compromises.
This is why Qualcomm’s Snapdragon X Elite flopped—good hardware, zero software gravity. Nvidia brings 30 years of software ecosystem with it from day one.
Nvidia’s also selling a vision: intelligence moves from the cloud to your device. With 128 GB of memory and 1 petaflop of compute, these machines run 120-billion-parameter models entirely on-device. No uploading your data to some server farm. No subscription. No third-party middleman. Nvidia and Microsoft are building OpenShell to turn Windows into a full agentic OS—persistent AI agents running locally, working on your behalf around the clock.
**So who wins?**
Obviously Nvidia. RTX Spark opens an entirely new revenue stream on top of its already-dominant data center business. The CPU market is exploding toward $200 billion. Even a modest slice of the premium PC segment means tens of billions in incremental annual revenue over the next decade.
But the cleanest, most underappreciated winner? **Arm Holdings**. It collects a royalty on every single RTX Spark chip sold. The RTX Spark’s CPU cores are based on Arm architecture. Arm bears zero execution risk, zero supply chain headaches, zero OEM relationship drama. It just clips a coupon on every premium PC sold as x86 gets displaced industry-wide. This is the purest expression of the trade.
Intel? Getting squeezed from above by Apple Silicon and now Nvidia on AI performance. That’s a real long-term problem for their client computing business.
The bottom line: the PC market just fundamentally changed. And the royalty collector wins.
