Microsoft’s Surface Laptop Ultra leads Tom’s Hardware’s local AI test with Nvidia RTX Spark
Tom’s Hardware says the Windows on Arm laptop is the fastest system it has tested for local agentic AI workloads.
Tom’s Hardware says Microsoft’s Surface Laptop Ultra with Nvidia’s RTX Spark Superchip is currently the fastest system it has tested for local agentic AI workloads, which matters because it points to a new class of Windows on Arm PCs aimed at running autonomous on-device agents. The test machine uses a 20-core CPU, 6144 CUDA cores and 128GB of LPDDR5X-9400 RAM, and the publication benchmarked it against a 14-inch MacBook Pro with an M5 Max and AMD’s Ryzen AI Halo. Using Artificial Analysis’s AgentPerf-local benchmark, the Surface beat the MacBook Pro by about 25% on Qwen 3.8-27B, and the AMD system took about 40 minutes on that same test. On the lighter Qwen 3.6 35B A3B workload, the Surface finished in eight minutes versus 10 minutes for the MacBook Pro and 13 minutes for the AMD system. Tom’s Hardware notes that the result is about end-to-end agent completion time rather than token speed, and says more RTX Spark laptop reviews are coming soon.
Why it matters
For people watching on-device AI on Windows on Arm, this result suggests the category can now compete at the top end of local agentic workloads instead of just matching on paper specs. It also sets a clear benchmark for how Microsoft, Nvidia and rival laptop makers will be judged on end-to-end agent completion time, not token speed alone.
Keep or strike?
Does this story matter, or is it hype? Mark it before you see what everyone else did.
Sources
- Tom's Hardware