Nvidia’s RTX Spark laptops run local AI models faster than RTX 5090 desktops
Nvidia’s RTX Spark laptops, including Microsoft’s Surface Laptop Ultra and HP’s OmniBook Ultra, combine an Arm CPU, Blackwell GPU, and up to 128GB of unified memory to run large local AI models efficiently. PCWorld tested these laptops, including the $7,000 Asus ProArt P16, against an RTX 5090 desktop, finding the laptop processed 472,084 tokens in 33 minutes and 13 seconds—nearly 20 minutes…
Key points
- RTX Spark laptops use unified memory (up to 128GB) for CPU/GPU access, enabling faster local AI workloads than RTX 5090 desktops
- PCWorld’s tests show a 120-billion-parameter model processed 472,084 tokens in 33m13s on an RTX Spark laptop vs 52m51s on an RTX 5090 desktop
- Gaming performance matches RTX 5070 but isn’t the primary focus; battery life drops to 10 hours under normal use, plummeting under heavy AI loads
The key advantage is unified memory, allowing both CPU and GPU to access a shared pool, unlike traditional laptops with separate VRAM. While gaming performance matches an RTX 5070 discrete GPU, it’s not a priority use case. Battery life drops sharply under heavy AI workloads, consuming 46% of charge in 33 minutes. Windows on Arm compatibility is improving but still requires checking software compatibility, especially for creative or specialized tools. Nvidia positions RTX Spark for AI developers and researchers needing local inference, though the high cost limits mainstream appeal.
The story so far
2 episodes →- Nvidia’s RTX Spark laptops run local AI models faster than RTX 5090 desktopsthis story
Nvidia’s RTX Spark laptops make no sense — until they do
pcworld.com · 4 November 2026
Loading the full article…
This text was published by pcworld.com and written by Brad Chacos. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- AMD reports 45% revenue growth and 64% data center share in Q3 · 3 src
- Marvell agrees to buy Lumentum in $115 billion all-stock deal for AI connectivity · 4 src
- PCMag introduces AI PC benchmarking tests for local AI hardware · 1 src
- OpenAI proposes renaming data centers to ‘AI compute campuses’ amid backlash · 5 src
- Chinese AI firms ordered 6M Nvidia RTX Pro 5500 chips worth $70B, expect ~1% delivery in four months, sources say · 4 src
Comments
via GitHub Discussions