The ability to run 120-billion-parameter models locally changes everything.
Surface Laptop Ultra features NVIDIA RTX Spark SoC with 6,144 CUDA cores, 15-inch mini-LED display at 2,000 nits, weighs under 4.5 lbs, and promises 1 petaflop of AI compute performance. Estimated pricing ranges $3,000-$7,000 based on historical Surface pricing; device launches fall 2026 alongside RTX Spark machines from ASUS, Dell, HP, Lenovo, and MSI.
- NVIDIA RTX Spark SoC with 20-core Grace CPU and 6,144 CUDA cores
- 15-inch mini-LED display at 2,000 nits, under 4.5 lbs, under 18mm thick
- Up to 128GB unified memory, 1 petaflop AI compute, runs 120-billion-parameter models locally
- Estimated $3,000–$7,000 pricing; fall 2026 availability
Microsoft unveiled the Surface Laptop Ultra at Computex 2026, a flagship workstation laptop powered by NVIDIA's RTX Spark chip designed for AI professionals and developers. The device offers up to 128GB unified memory and claims to run 120-billion-parameter AI models locally, positioning it as a direct competitor to Apple's MacBook Pro M5 Max.
Microsoft walked on stage at Computex on June 1st and announced the Surface Laptop Ultra, a machine that represents the company's most serious attempt yet to build a workstation-class laptop. This is not an incremental refresh. The device was engineered from the ground up with NVIDIA, powered by a new chip called RTX Spark, and designed explicitly for the people building the systems and infrastructure the world depends on—creators, developers, AI professionals.
The hardware itself is striking. A 15-inch mini-LED display with 2,000 nits of peak brightness, the brightest screen Microsoft has ever shipped on a Surface device. The whole thing weighs under 4.5 pounds and measures less than 18 millimeters thick. Inside sits NVIDIA's RTX Spark system-on-a-chip: a 20-core ARM-based Grace CPU paired with a Blackwell RTX GPU containing 6,144 CUDA cores. The machine can be configured with up to 128 gigabytes of unified memory—the same maximum as Apple's M5 Max MacBook Pro—and Microsoft claims it delivers one petaflop of AI compute performance. The cooling system is beefier than previous Surface laptops, with 2.5 times the thermal capacity of the 15-inch Surface Laptop 7th Edition. There's a full-size SD card reader, USB-A, USB-C, HDMI, and a headphone jack. The touchpad is the largest haptic pad Microsoft has ever put on a Surface device.
What makes the RTX Spark chip significant is that it's NVIDIA's first system-on-a-chip designed specifically for Windows PCs. The Grace CPU and Blackwell GPU are connected through NVIDIA's NVLink interface, creating a unified platform where memory can be dynamically allocated between processor and graphics based on what the workload demands. In practical terms, this means the Surface Laptop Ultra can run AI models with up to 120 billion parameters entirely on the device, without touching the cloud. It can render 3D scenes exceeding 90 gigabytes, edit 12K video natively, and play AAA games at 1440p above 100 frames per second. NVIDIA is positioning this as a step toward what it calls an "agentic AI platform"—where local AI agents can handle tasks, manage workflows, and interact with applications using natural language.
The Surface Laptop Ultra is not arriving alone. ASUS, Dell, HP, Lenovo, and MSI have all announced their own RTX Spark machines, with Acer and Gigabyte following later. This makes the Surface Laptop Ultra the flagship of an entirely new category of Windows-on-ARM PCs, and it's clearly positioned as a direct competitor to Apple's MacBook Pro M5 Max. The comparison is instructive. Both machines max out at 128 gigabytes of memory. The Surface Laptop Ultra is lighter—under 4.5 pounds versus 4.8 pounds for the 16-inch MacBook Pro. The MacBook has a larger display at 16.2 inches with a 16:10 aspect ratio and 1,600 nits of brightness; the Surface has a smaller but brighter 15-inch screen with a 3:2 ratio. The MacBook Pro is available right now with proven benchmarks and verified battery life of up to 24 hours. The Surface Laptop Ultra remains a pre-release product with unverified performance claims and an "all day" battery life that Microsoft has not quantified in hours.
Pricing remains a mystery. Microsoft has not announced official numbers, but based on the company's historical Surface pricing and industry estimates, the Surface Laptop Ultra is expected to cost between $3,000 and $7,000 depending on configuration. That would put it in direct competition with the MacBook Pro M5 Max, which starts at $3,599 for the 14-inch model and $3,899 for the 16-inch. The device is scheduled to ship in fall 2026, though some industry observers have raised questions about the timeline. Tech leaker Moore's Law Is Dead has suggested that RTX Spark machines may not be widely available until 2027, despite Microsoft's official messaging pointing to a 2026 release.
Several critical questions remain unanswered. Real-world benchmarks have not been shared—NVIDIA claims RTX 5070 equivalent GPU performance, but independent testing is months away. Battery life specifics are missing. App compatibility on Windows on ARM has improved significantly but is still not universal, and professionals should verify their critical workflows run natively on ARM before committing. For developers and AI professionals who depend on NVIDIA's CUDA ecosystem, however, the Surface Laptop Ultra could be genuinely transformative. The ability to run 120-billion-parameter models locally with 128 gigabytes of unified memory in a chassis under two kilograms is something no other Windows laptop currently offers. Whether that advantage translates to real-world value depends on pricing, verified performance, and whether the software ecosystem catches up before the machine ships.
Citas Notables
Engineered with NVIDIA from the silicon up, designed for professionals whose work involves massive 3D scenes, long compile cycles, and large scale AI model training.— Microsoft's official announcement
Built for those building the systems, the breakthroughs and the infrastructure the world runs on.— Brett Ostrum, Corporate Vice President of Surface at Microsoft
La Conversación del Hearth Otra perspectiva de la historia
Why does Microsoft need to build this machine at all? Aren't MacBook Pros already dominating the creative professional space?
The MacBook Pro is excellent, but it's built on Apple's integrated GPU architecture. If you're doing serious AI work or GPU-accelerated development, you're locked into Apple's Neural Engine. The Surface Laptop Ultra gives you full NVIDIA CUDA support—that's the entire ecosystem of tools, libraries, and optimizations that AI professionals have built their workflows around.
So this is really about CUDA, not about being lighter or faster?
It's both, but CUDA is the unlock. You can run 120-billion-parameter models locally on this machine. That's not a marketing claim—that's a fundamental shift in what you can do without cloud connectivity. For someone training models or doing inference at scale, that changes everything.
But the MacBook Pro is available now. The Surface Laptop Ultra is vaporware until fall 2026. Why should anyone care?
Fair question. Right now, you shouldn't wait. But if you're someone who's been frustrated by Apple's GPU limitations or locked into cloud-based AI workflows, this is the first real alternative in years. The risk is that Microsoft oversells the performance or the battery life doesn't hold up, or Windows on ARM still has compatibility issues.
What's the real barrier to adoption here?
Price and proof. If the base model costs more than $3,500, it's hard to justify over a proven MacBook Pro. And until we see independent benchmarks, we're taking NVIDIA's word that this delivers RTX 5070 performance. That's a big ask for a pre-release product.