Key Takeaways:
Nvidia's entry into the Arm-based PC processor market sent Arm Holdings shares to a record closing high of $408.85 on Monday, a 16% surge that extends the chip designer's year-to-date gain to 274%. The rally came after Nvidia unveiled its RTX Spark superchip at Computex in Taipei, integrating Arm's CPU architecture with Nvidia's Blackwell graphics technology in a single system-on-chip for Windows PCs.
"This is the first completely re-engineered, reinvented line of PCs that has happened in 40 years," Nvidia Chief Executive Officer Jensen Huang said during the keynote. The chip, developed with Microsoft and co-designed with MediaTek, will power more than 30 laptop models and 10 compact desktops from partners including Microsoft, Dell, HP, Asus, Lenovo and MSI starting this fall.
The top-end N1X processor features 20 Arm CPU cores — 10 high-performance Cortex-X925 and 10 medium-sized Cortex-A725 — paired with 6,144 Blackwell-based GPU cores, the same count as the desktop GeForce RTX 5070. The chip supports up to 128 gigabytes of unified LPDDR5x memory, a design choice that gives the GPU access to far more memory than the 12 gigabytes available on a discrete RTX 5070. Nvidia said the chip's power envelope maxes out at 80 watts, compared with 250 watts for a desktop RTX 5070. A lower-end N1 variant offers up to 12 CPU cores, 2,560 CUDA cores and 64 gigabytes of memory with a 45-watt TDP, targeting thin-and-light laptops. Both chips are manufactured on TSMC's 3-nanometer process.
Arm's architecture gains a powerful ally
Nvidia's endorsement represents the most significant validation yet of Arm's push into the PC market, a domain long dominated by Intel's x86 architecture. Arm shares have more than tripled this year as the company's licensing and royalty model benefits from a wave of chipmakers adopting its designs. Nvidia's RTX Spark joins Qualcomm's Snapdragon X series as the second major Arm-based platform for Windows, but with a critical difference: Nvidia brings its CUDA ecosystem and DLSS upscaling technology, potentially solving the gaming performance gap that has plagued Arm-based Windows machines.
"Fast CPUs have become essential to keeping the AI factory moving," said Ian Buck, Nvidia's vice president of hyperscale and high-performance computing, pointing to the chip's ability to run 120-billion-parameter AI models locally with up to 1 million tokens of context. The unified memory architecture allows developers to access more than 100 gigabytes of video memory for AI workloads, compared with the 8 to 12 gigabytes typical of discrete laptop GPUs.
Who wins, who loses
The competitive implications extend beyond the PC market. Nvidia also announced Monday that its Vera data center CPU is now in full production, with early customers including OpenAI, Anthropic and SpaceX's xAI. Huang said Nvidia is making "millions" of the CPUs for what he called "a market that never existed before."
Intel shares fell 4.7% on Monday, while AMD dropped 1.2%, as investors priced in the structural threat to x86's dominance. Mizuho analyst Vijay Rakesh maintained outperform ratings on both Arm and AMD while keeping a neutral rating on Intel, writing that "strong CPU demand" would drive tailwinds across both x86 and Arm platforms through 2027. Independent analyst Richard Windsor of Radio Free Mobile said the "reality will hit home" if AMD announces its own Arm-based CPU for PCs and data centers, signaling a broader industry shift.
Nvidia shares rose 6.3% on Monday, reflecting investor enthusiasm for the company's expansion beyond its core data center business. The RTX Spark machines are expected to carry premium pricing; the existing DGX Spark developer workstation, which uses the same silicon, currently sells for $4,699. Nvidia said it will release specific performance metrics closer to the fall launch window.
This article is for informational purposes only and does not constitute investment advice.
