Microsoft Surface Laptop Ultra Review: The RTX Spark Chip Changes Everything … for AI Workers

Last week, graphics processor manufacturer Nvidia launched a foray into laptop processors with the RTX Spark, a powerful laptop chipset that combines a CPU and GPU into a single system-on-a-chip (SoC), similar to what you find in most phones and tablets. You might think the goal is efficiency, and it is, but Nvidia has also packed in enough power to support a top-flight productivity laptop and far better graphics than any system without a discrete graphics card. Between RTX Spark and Google’s new GoogleBook laptops, we have entered a new age of laptops made and optimized specifically for working with AI. If you’re bullish on AI and use it regularly at home or at work, it’s a brand-new day.

As influential as Nvidia is, it’s no surprise that many major brands have jumped on the RTX train early. Six Spark laptops are out or coming soon, but the most notable to me is from Microsoft, which slapped the “Ultra” moniker on a new high-end version of its Surface Laptop. Microsoft, having leaned heavily into AI and productivity, is a perfect poster child (and test subject) for what this chip can do, so I’ve been eager to get my hands on this so-called Surface Laptop Ultra since Nvidia first unveiled Spark a few months ago.

Mostly, I’m curious. These systems are built for AI, and that makes sense, but not everyone needs a laptop powerful enough to replace a $2,000-per-month Claude token habit: The question is, can the Surface Laptop Ultra (or any RTX Spark system) create a Windows-based alternative to the MacBook Pro, a do-anything machine for power users, no matter the output?

The reality is, it can be, but probably isn’t quite as appealing (at least for now): The Surface Laptop Ultra is an impressive flagship laptop and performs well in many contexts, including gaming, where it beats any non-gaming laptop I’ve tested. However, with a $2,600 minimum price tag, this is a “Pros only” machine, and the easiest way to justify that spend is a heavy AI workflow, so I’d recommend it as a productivity machine for programmers and other AI workers. Still, if you have the budget and want to see what an RTX Spark machine can do, there’s a lot to love here.

The RTX Spark isn't just the heart of this laptop; it's the reason it exists, and why I’m writing about it. So let's break down what makes the RTX Spark so special.

The SoC features an 18- or 20-core CPU, split evenly between performance and efficiency tiers, and up to 6,144 Nvidia Blackwell RTX graphics processor cores in the most powerful configuration. Technically, that’s more GPU power than an Nvidia RTX 5070 laptop GPU, so Spark laptops should perform at least as well as a midrange gaming laptop when you’re playing games or editing video.

Perhaps more importantly, it features a unified memory setup, giving both the processor and graphics chip access to a larger memory pool—up to 128 gigabytes (GB) of RAM on the Surface Laptop Ultra. That’s a huge boon for AI work, which relies heavily on RAM to generate and replicate data while coding or building datasets and may need to switch rapidly between compute and graphics.

The goal here, cramming all this power into a laptop, is to allow AI workers to process larger AI tasks locally, using the laptop’s power, rather than relying on cloud-based computing from companies like OpenAI and Anthropic, which can get expensive very quickly.

That said, while the novelty (and primary focus) is AI, the specs also make this machine an excellent choice for creators. It can handle 4K video editing in Adobe Premiere or DaVinci Resolve, or image editing in Photoshop or Lightroom. And, almost as an afterthought, it is also a capable gaming PC, able to run many games at high settings with advanced features like ray tracing and Deep Learning Super Sampling (DLSS). Given the cost of the RTX Spark laptops at launch, I wouldn’t recommend it specifically for gaming, but it’s impressive all the same, and a very nice perk if you need this kind of power.

Even after a week of use-testing, I couldn't really find the upper limit of the Surface Laptop Ultra’s power. I had the top configuration—20-core CPU/6,144 GPU cores, 32 GB of RAM—so everyday life doesn’t give you the opportunities you need to push this kind of machine to its limits. It handled my routine of running a web browser with 20-plus open tabs, fielding a constant stream of emails, messages, and calls via Microsoft Outlook and Teams, editing batches of large image files in Affinity, and streaming music from Qobuz.

Luckily, we can artificially find the Surface Laptop Ultra’s limits with benchmarking tools that stress-test specific aspects of the system. The Surface Laptop Ultra scored 6,579 in the Cinebench multithread CPU test, ahead of the public average performance of the MacBook Pro with M5 (4,357), but behind last year’s M4 Max-equipped MacBook Pro (7,829). That’s not a surprise, given the high number of cores in the CPU, but a good technical indicator that this system delivers the performance its on-paper bona fides suggest.

In single-thread CPU testing, the RTX Spark received a 552 performance score, well behind the M5 (733) and M4 Max (676). Single-core performance relates to how your processor handles individual tasks, while multicore is a better indicator of multitasking within an app or across multiple apps. So this machine is made for multitasking.

I’m not an AI programmer, so I pushed the Surface Laptop Ultra as hard as I could in gaming. To my delight, the sleek notebook ran 007 First Light flawlessly on high settings at 1440p resolution (with some help from DLSS 5).

Gears of War E-Day, a Microsoft-made shooter, also ran well at 1440p with medium detail settings. Again, this is very impressive for a machine not labeled a “gaming laptop,” with performance comparable to a system with an RTX 5060 GPU. For all but the biggest spec-hounds, I think this is more than enough performance for a daily driver. That said, a dedicated gaming laptop with an RTX 5070 GPU would likely deliver much more in this department for less than an entry-level RTX Spark notebook.

In terms of power efficiency, the Surface Laptop Ultra performs decently, but not exceptionally. The battery lasted through a workday with my usual PC usage pattern, which didn’t surprise me. In high-usage tasks, it can deliver the same peak performance, but expect the battery to last far less—maybe a number of hours you can measure with the fingers of one hand. Like any high-powered laptop, it’s generally better to work while plugged in whenever possible.

Microsoft’s Surface line has always produced some of the cleanest, well-designed Windows machines of the last decade, and the Surface Laptop Ultra lives up to that reputation. Sleek and exceptionally well-crafted, this all-metal notebook is a pleasure to use and constantly reminds me I am using a top-shelf mobile PC. Its quiet keyboard offers excellent key travel, while the generously sized trackpad adds haptic feedback, making every interaction incredibly satisfying.

The Surface Laptop Ultra checks all the boxes on the hardware front, as expected for a product in its price category. Its Nvidia RTX Spark chip is paired with a sharp and superbright 15-inch PixelSense mini-LED touchscreen with Dolby Vision HDR support and up to a 120-Hz refresh rate. The display panel itself is gorgeous and great for touch interactions, though I wish it used a Surface Pen like the top-shelf Surface notebooks of yesteryear.

Storage-wise, my configuration featured a 1-terabyte SSD, which is more than enough for most consumers. I could fit anything—from my work files to the AAA games I mentioned above—and still have ample space left. For connectivity, the PC has everything a power user would need, including HDMI output and an SD card reader.

Lastly, for video calls, it features a 1080p Surface Studio camera, which supports facial recognition. The speaker and microphone sound quality are also decent. A Dolby Atmos-capable audio system supports simulated Spatial Audio, which is especially helpful for streaming and gaming.

Based on my time with the Microsoft Surface Laptop Ultra, it's clear the Nvidia RTX Spark chipset will change the Windows laptop landscape. It is very powerful, opening the door for AI workers to do more prompting on-device locally while still delivering strong performance in conventional creative workflows and gaming.

That said, all RTX Spark machines (especially at top specs) are prohibitively expensive. At $2,600, the base Surface Laptop Ultra is one of the most affordable RTX Spark laptops, but it's substantially costlier than the current base MacBook Pro ($2,000) or comparable gaming laptops like the Asus ROG Strix G16. The version I tested, a midrange configuration, costs $3,700, which is arguably a bargain compared to a fully kitted-out MacBook Pro, but obviously far more than I’d expect anyone to spend on a personal-use system. A fully specced-out system costs nearly $6,000, which is not unheard of in the laptop world, but is clearly reserved for professionals.

If you are looking for a great all-around PC laptop for everyday use, not including heavy AI and creative projects or high-performance gaming, save yourself a lot of money by ordering a Qualcomm Snapdragon X2 Elite laptop. You’ll get a lighter computer with longer battery life. Microsoft’s standard Surface Laptop is one of the first that comes to mind.

For now, RTX Spark still counts as “future tech” for most consumers, but we’re looking forward to testing more of these machines and seeing this tech trickle down to midrange and even budget-tier systems down the line.