A laptop that runs 120-billion-parameter models without choking
At Computex 2026, Microsoft unveiled the Surface Laptop Ultra — a machine that marks a quiet but consequential shift in the company's identity, from maker of capable everyday tools to challenger of the professional workstation throne. Built around Nvidia's RTX Spark chip and capable of running vast AI models without a cloud in sight, it is less a product announcement than a philosophical declaration about where portable computing is headed. Whether it can unseat Apple's MacBook Pro among the creative class remains an open question, but the ambition itself is worth noting.
- Microsoft has never aimed a Surface device this directly at power users — the Surface Laptop Ultra is a deliberate break from the brand's reputation for safe, general-purpose machines.
- The RTX Spark chip at its core — packing 6,144 Blackwell cores and delivering one petaflop of AI performance — threatens to collapse the boundary between laptop and workstation.
- Apple's MacBook Pro has long held the loyalty of video editors, 3D artists, and developers; Microsoft is now knocking on that door with matching memory, a brighter display, and a full port suite.
- A global RAM shortage looms over the launch, threatening to price out the very professionals this machine is meant to serve.
- With a release still months away and no confirmed price, the Surface Laptop Ultra remains a bold promise awaiting the verdict of real-world workflows.
Microsoft took the Computex 2026 stage with a Surface device unlike any it had built before. The Surface Laptop Ultra, revealed during Nvidia's keynote, is a deliberate departure — not a tool for everyday productivity, but a direct challenge to Apple's MacBook Pro among creative professionals and AI developers.
At its heart is Nvidia's RTX Spark chip, formerly known by the codename N1X, which pairs 6,144 Blackwell RTX cores with 20 Arm-based CPU cores and up to 128GB of unified memory. That architecture delivers one petaflop of AI performance — enough to run 120-billion-parameter language models entirely on-device. Microsoft claims GPU performance on par with a mobile RTX 5070, opening the door to serious video rendering, 3D work, and gaming that Surface machines have never before attempted.
The display matches the ambition: a 15-inch mini-LED PixelSense Ultra screen reaching 2,000 nits of peak brightness, the most luminous panel Microsoft has ever shipped in a Surface. A full port selection, the brand's largest-ever haptic touchpad, and promised all-day battery life round out a spec sheet clearly aimed at professionals who have long looked elsewhere.
What the device signals may matter as much as what it contains. The RTX Spark chip reflects a broader industry bet that the future of mobile computing lies in machines that process AI locally, render media without cloud dependency, and absorb workloads once reserved for desktop towers. The unified memory pool — shared across GPU, CPU, and system — is designed to eliminate the bottlenecks that have historically frustrated creative work on laptops.
The launch is expected later in 2026, but pricing remains unresolved. A global RAM shortage has pushed costs upward, and a machine with 128GB of memory and a cutting-edge GPU will not come cheaply. For now, the Surface Laptop Ultra is a statement of intent — that Microsoft sees where computing is going, and is willing to build something that meets professionals there.
Microsoft walked onto the Computex 2026 stage with something it had never quite attempted before: a Surface device built unapologetically for power users. The Surface Laptop Ultra, announced during Nvidia's keynote presentation, represents a deliberate pivot away from the company's historical positioning of Surface machines as everyday productivity tools. This one is different. It's built to compete directly with Apple's MacBook Pro, and it's built around Nvidia's newly unveiled RTX Spark chip—a processor that had been circulating in tech circles under the codename N1X until its official reveal.
The hardware specifications alone signal Microsoft's ambitions. The RTX Spark chip contains 6,144 Blackwell RTX cores paired with 20 efficient Arm-based CPU cores, all backed by up to 128 gigabytes of unified memory. That configuration delivers one petaflop of AI performance—enough computational muscle to run 120-billion-parameter language models directly on the machine without relying on cloud servers. Microsoft claims the GPU performance matches what you'd find in a mobile RTX 5070, which translates to faster video rendering, smoother 3D work, and gaming capability that goes well beyond what Surface devices have historically offered. For a company that has spent years positioning Surface as a device for general users, this is a meaningful statement about where the market is heading.
The display tells part of the story too. Microsoft equipped the Surface Laptop Ultra with a 15-inch mini-LED PixelSense Ultra screen capable of reaching 2,000 nits of peak brightness in HDR mode—the brightest display the company has ever put in a Surface device. The company emphasizes color accuracy for creative professionals, which makes sense given the target audience. The machine also includes a full complement of ports: HDMI, USB-C, USB-A, an SD card reader, and a headphone jack. The haptic touchpad is the largest Microsoft has ever integrated into a Surface laptop. Battery life is promised to stretch across a full day of work, though that claim will require real-world testing to verify.
What makes the Surface Laptop Ultra genuinely interesting is not just what it can do, but what it signals about the direction of portable computing. The RTX Spark chip represents a fundamental shift in how manufacturers are thinking about laptop processors. Rather than simply scaling up traditional CPU performance, Nvidia and Microsoft are betting that the future of mobile computing belongs to machines that can run AI models locally, render complex media without cloud assistance, and handle workloads that previously required either a desktop workstation or expensive cloud infrastructure. The unified memory architecture—where the GPU, CPU, and system memory share the same pool—is designed to eliminate the bottlenecks that have historically slowed down creative work on laptops.
Microsoft is positioning this device as the answer to a specific problem: professionals who need genuine workstation power but also need portability. The MacBook Pro has owned much of that territory for years, particularly among video editors, 3D artists, and software developers. The Surface Laptop Ultra is an explicit challenge to that dominance. Whether it succeeds will depend partly on performance benchmarks and partly on price—and that's where uncertainty creeps in. The global RAM shortage has driven up the cost of high-memory systems significantly. A laptop with up to 128 gigabytes of memory, a cutting-edge GPU, and a premium display will not be inexpensive.
The device is scheduled to launch later in 2026, which means hands-on reviews and real-world performance data are still months away. Microsoft has made bold claims about what the RTX Spark chip can accomplish, but those claims will need to survive contact with actual creative workflows, actual AI models, and actual gaming scenarios. The company has also been careful to position this as a premium product for a specific audience—not a replacement for the standard Surface Laptop, but rather a new tier above it. For now, the Surface Laptop Ultra exists mostly as a promise: that Microsoft understands where computing is heading, and that it's willing to build hardware that meets professionals where they actually work.
Notable Quotes
The Surface Laptop Ultra is designed to handle creative workloads, development, local AI use, and gaming— Microsoft
The GPU performance matches a mobile RTX 5070, enabling smoother 3D rendering and faster video exports— Microsoft
The Hearth Conversation Another angle on the story
Why does Microsoft need a laptop that can run AI models locally? Isn't cloud computing supposed to be the future?
Cloud works great if you have reliable internet and don't mind latency. But a video editor working on set, or a developer iterating on code, or an artist testing a model—they need instant feedback. Running it locally means no network dependency, no waiting for a server response, no sending proprietary work to the cloud.
The specs mention 1 petaflop of AI performance. What does that actually mean for someone using the machine?
It means you can load and run very large language models—120-billion-parameter models—without the machine choking. That's the difference between a model responding in seconds versus minutes. It's the difference between usable and unusable.
Microsoft says this is the most powerful Surface ever. Why did it take them this long to build something like this?
Surface was designed as a consumer-first line. Premium, but accessible. This is a different bet—that there's a market willing to pay for genuine workstation power in a portable form. The RTX Spark chip made that possible in a way that wasn't feasible before.
The display reaches 2,000 nits. Is that just marketing, or does it actually matter?
For creative work—color grading, photo editing—brightness and color accuracy matter enormously. You're making decisions about how something will look. A dim screen or inaccurate colors means you're working blind. 2,000 nits is legitimately bright.
What's the real competition here?
The MacBook Pro. Apple has owned the professional portable computing space for years. This is Microsoft saying: we can build something that matches or exceeds what Apple offers, and we're building it around a chip architecture that's fundamentally different—one that prioritizes local AI capability.