A week with RTX Spark: everything we’ve learned using Nvidia’s new laptops

Nvidia’s RTX Spark N1X processor is eventually here, and we’ve been collectively evaluation a small fortune’s value of mighty laptops sporting this new “superchip.” RTX Spark is designed to be a high-performing Arm-based SoC for laptops and small desktop computers, rivaling the finest from Apple, Intel, AMD, and Qualcomm. It has loads of cores, Nvidia RTX Blackwell graphics, and adequate power and unified recollection to run sizable local AI models.

The flagship placement is inner Microsoft’s Surface Laptop Ultra alongside a staggering 128GB of unified memory, which our own Tom Warren completely reviewed. Sean Hollister has been evaluation a 64GB type of the Asus ProArt P16, which I’ve additionally been evaluation in its top-spec 128GB config. Since I’m the most spoiled of the lot, I additionally have a 64GB Lenovo Yoga 9n 2-in-1 and a 32GB HP OmniBook Ultra in for testing.

All of these laptops have the identical N1X part alongside 20 CPU cores and 6,144 GPU CUDA cores, and they’re all wildly costly — ranging from the HP OmniBook at $3,199.99 to the Asus ProArt’s $6,999.99. There’s an 18-core / 5,120 CUDA N1X part in lesser disbursal configs, but that lone gets as cheap as $2,599.99 for a Surface Laptop Ultra or a smaller Asus ProArt P14 alongside 24GB of RAM.

These are big and costly laptops built on new chips. They’re meant for developers, coders, and satisfied creators who power additionally desire the capability to use local AI and equal perform games using the latest Nvidia tech akin DLSS 5. There’s a lot to test, and we have not had a ton of period to test it all. So below you won’t discover an exhaustive scored assessment of any one laptop, but alternatively our impressions and thoughts from three writers using five laptops for a week.

Is Nvidia’s “superchip” really all that super? Are Apple, Intel, AMD, and Qualcomm officially on notice? Have any of us gotten any sleep?

Antonio Di Benedetto: Windows on Arm is no longer a entirely subpar cognition akin it was a few years ago. Thanks in ample part to the shove about Qualcomm’s Snapdragon X chips, many of betterment has been put into native Arm-based apps for Windows. And many x86 apps activity fine via Microsoft’s Prism emulation layer.

Adobe has had native Arm versions of Photoshop and Premiere for several period now, and equal although my precious Lightroom Classic is motionless an x86-based app I don’t notice any issues using it on Arm. At this point, you’ve have to be relying on specialized application that’s exclusively x86 based for Windows on Arm to be a non-starter — or you desire to openly perform any games out there without any concern of incompatibilities.

Sean Hollister: Every so often, there’s that instant whenever an app takes longer to burden than it should, and I amazement whether it’ll burden at all. Even then, it typically loads. Occasionally, it takes additional than one try. My Evernote notes disappeared a few times one day whenever I alt-tabbed away, but a power refresh fixed that. And apps don’t appear to regard my scheme penchant for dreary mode.

But I mostly concur alongside Antonio. I made an RTX Spark my regular driver, alongside all my representative activity and individual apps — Signal, Discord, Audacity, CapCut and Evernote, Irfanview, Slack, ScreenToGif and additional — and nothing has unsuccessful me so far. I never formerly needed to think whether I was operating an Arm or an x86 app. The genuine hitches came alongside gaming and AI, as you’ll peruse below.

Tom Warren: I’ve been using a Windows on Arm laptop alongside a Qualcomm part regular for the former two years and things have improved to the item that I can’t inform if an app is native or not anymore. That’s the identical narrative on the RTX Spark laptops, and I awareness akin Nvidia’s admission into this category volition lone enhance the narrative for imaginative apps and gaming.

TW: Windows on Arm never looked akin a grave gaming phase before now. While there motionless aren’t many native Arm games, Nvidia’s GPU helps bruteforce through the emulation layer. Your mileage in gaming is going to change depending on the title, but for the most part I’ve established that average settings, operating at 1440p and DLSS performance, volition get you to a stable 60fps in most games.

I mostly equivalent on my desktop, anywhere I’m spoiled by using an RTX 5090 on a 32-inch 4K display all day alongside maxed-out equivalent settings, but I didn’t intellect the appearance of average settings on a 15-inch display. I was capable to run Black Myth: Wukong at 70fps on average at native resolution (3270 x 2180) alongside average settings and DLSS achievement manner enabled. I ran the identical average settings and DLSS achievement in Cyberpunk 2077 and averaged 66fps, and Forza Horizon 6 managed an average of 57fps.

In several titles you may need to autumn the resolution or settings lower, though. This certainly isn’t as fine as a dedicated gaming laptop, but for item that’s approximately on the flat of a MacBook Pro it’s additional than fine adequate for gaming on the go.

SH: Like Tom says, the GPU power is there — but the CPU power clasp rear several of your games. I had plentifulness of power for a gentle 60fps in Returnal (at 1440p alongside elevated settings), Expedition 33 (at a DLSS “Performance” upscaled 1200p alongside elevated settings), Jedi Survivor (DLSS “Performance” upscaled 1200p alongside average settings), and Shadow of the Tomb Raider (at native 1800p and highest spec).

But whenever I fired up Helldivers 2, equal dipping downward to a DLSS-upscaled 1080p was no justify of 60fps in the highest difficulty missions. There are fair too many enemies on display for the Spark to keep up. Star Wars Zero Company is equal additional CPU-bound — the Spark didn’t attention whether I was maximizing or minimizing the graphics and display resolution. No matter what I tried, I typically saw my framerate capped between 37fps and 39fps in battle.

Even ancient and low-powered games can have quirks. I tried the 2003 heading Max Payne 2 and saw situations anywhere framerate dipped to 50fps. I equal saw a few hitches in the outstanding UFO 50, a gathering of 2D games, although a brace another 2D games ran great.

I’m additionally not certain if RTX Spark is suitable for VR yet. The excellent Titanfall 2 VR mod worked, and it seemed to run fine at first, but misplaced a small stability following a few minutes. That could be my Wi-Fi, since the controller for Valve’s Steam Frame wireless adapter wouldn’t instal on the Spark. Neither did Steam’s streaming audio solution, so the audio came out of the Nvidia laptop alternatively of the headset.

Just so we’re clear, all of this is assuming you desire a laptop maxing out its fans and positively dumping warmth on your array or lap. That’s not different for lean gaming notebooks, but it’s way louder and hotter than a fine desktop or handheld.

AD: I’ve wanted a MacBook Pro-like laptop alongside complete PC gaming chops for as lengthy as I can remember, so I was hyped for these laptops to provision the products in graphics. The archetypal night, I set up the Asus ProArt P16, alongside all 128GB of its ridiculously ample pond of memory, and whenever I booted up Big Walk for my planned online evening alongside several allies my archetypal idea was, “Well, that looks okay. I guess.” I cognize Big Walk isn’t several triple-A graphical titan, but it has several fairly demanding graphics for the “friend slop” genre. I expected the N1X to max it out, or at smallest get close, but not really.

I had to bump the P16’s 4K display downward to 2560 x 1600, and alternatively of Big Walk’s Highest or Very elevated graphics presets, I brought it downward to High to keep a correctly firm 60-ish fps. Does any of that really matter for this silly co-op equivalent concerning solving puzzles alongside your buddies? Of way not. It looked fine, following all. But it was a sobering reminder that this “superchip” is no graphical magician.

As Tom and Sean item out, RTX Spark can do several decent gaming, but mood your expectations for average or equal low settings in several cases, and be prepared to depend on DLSS and example gen. There’s no denying that cheaper desktops and equal laptops alongside discrete Nvidia graphics cards can effortlessly attack out an N1X. Nvidia claims it’s akin having an RTX 5070 Laptop GPU in an SoC, but in my evaluation it was location between that and an RTX 5060 Laptop GPU — not a slouch, for sure, but if you’re a gamer archetypal your prosperity goes much additional alongside traditional Nvidia graphics cards.

SH: Nvidia’s partners saw fit to equip their archetypal motion of RTX Spark laptops alongside several of the biggest batteries you can bring on an airplane alongside no questions asked — but equal alongside a 99.9 watt-hour group akin the one in my Asus ProArt P16, these laptops won’t be winning awards for longevity. While Tom and Antonio spent their days evaluation all use case under the sun, I made the P16 my regular controller for normal activity during observing the power division meter akin a hawk. That mostly method penning stories in a web browser, and it’s not fairly enduring the day.

My finest run was 5 hours and 40 minutes of real-world activity at 50 percent brightness, including an hour-long video call alongside Tom and Antonio to strategize our RTX Spark safety plan. The next day, I got lone concerning 5 hours from a complete accusation since I spent two of them editing this video in Capcut. The third period I drained the battery, it took 5 hours, too.

With web work, I’m typically seeing 13-17 watt power division drain on the laptop’s default settings, but that climbs to 20-30 watts alongside ray video work, up to 50 watts during the (brief) Capcut video export, and I made it jump all the way to 120 watts playing Indiana Jones and the Great Circle. The power division couldn’t keep feeding at that charge — it gradually went down, and at 30 percent accusation I saw a big achievement hit as it dropped dramatically to 60 watts.

The fine news is the Asus charges fast from a 140W USB-C PD power source. It took 20 minutes to go from deceased to 50 percent battery, another 15 minutes to get to 70 percent, and a total of 1 hour, 15 minutes to attain 95 percent charge.

AD: I’ve had to run many of benchmarks during plugged in, so I haven’t been capable to expend as much period evaluation power division existence as I’d normally like. Anecdotally, in many of fits and starts, power division existence seems solid. As Sean saw, you’re improbable to get through a complete eight hr activity day unless you’re going out of your way to be efficient. And if you blaze up item heavy, akin a local AI example (more on that in a minute), you won’t equal create it to lunch. I’ll eventually do additional thorough power division evaluation as I continue using these laptops, including our customary rundown tests for a bit of 1:1 comparisons.

TW: Not yet. It all feels akin a beta aimed at developers correct now: from the instal cognition of AI agents flashing up command prompts and Visual Studio, to the fans blasting distant for the most essential of queries.

I struggled to arrive up alongside many compelling use cases for the power of these chips for local AI work. A lot of tasks could merely be done far additional quickly and without the laptop fans spinning up by using a liberated haze model. The power current is really in leaving the delegate operating at complete throttle to build item overnight, operating tasks that developers would favor.

It’s additionally nice to have this all on a local device, without having to portion emails and records alongside a haze assistance and concern concerning the privacy aspects. Still, I would regularly near OpenClaw or Hermes and not let them run in the backdrop all day, particularly if I had stated them permissions to admission local files. I’m fair not prepared to rely AI fully, local or not.

Microsoft and Nvidia the two envision a forthcoming anywhere AI agents automate things on your PC, but I’m motionless not completely convinced. While there are several mundane tasks I could certainly automate, I akin using my PC to edit photos or videos, write, and perform games. I don’t desire to sit there and observe a robot do all of that for me. That’s part of the joy of computing to me.

SH: I walked out of Nvidia’s Editor’s Day thinking I power eventually have a logic to blaze up local AI. I asked Hermes, operating Qwen3.8-27B, to build me a regular news briefing much akin the one in Nvidia’s on-stage demo. It took 10 minutes, and 10 percent of my complete 99.9 watt-hour battery, to eventually create a catalog of stories that took the cloud-based Claude (Opus 5.5) fractional as lengthy — during utterly monopolizing my GPU and dramatically heating up the Spark laptop. The laptop stayed hot and a small sluggish as lengthy as I kept Hermes open.

Claude additionally produced improved results, alongside additional item and improved sourcing. I conjecture I got the local AI tokens for “free,” I wouldn’t rely Claude alongside my email, and I likely could have used a smaller, lighter model. But I went alongside what the program recommended, and that was the result. I tried a spotless reinstall of Hermes to create certain it wasn’t a misconfiguration, but this time, the immediate took equal longer and used 60 percent of my battery. Ouch.

AD: I’m motionless first in my journey of learning to use local AI, and these RTX Spark laptops are evidently a viable — albeit, costly — option. But equal if you have 128GB of recollection to use several chunky models, it’s motionless not a user-friendly cognition for beginners. Microsoft’s Get Started intro in Windows 11 gives you uncomplicated one-click installs for Hermes Agent, OpenClaw, and Perplexity. (Which all have ESRB ratings for several reason?) But following that, you’re benevolent of on your own. I stuck alongside Hermes, since that’s what I was already using on the Mac Studio, but if you’re using it on a scheme alongside 32GB of RAM or small it fair tells you there isn’t adequate recollection for its recommended local Qwen models — either routing you towards the haze or completely relying on you to go fetch a smaller example on your own.

Using local AI has important privacy advantages complete haze models, but it’s certainly small convenient since local LLMs are much slower to procedure and additional taxing on your system. Churning a local example through your laptop’s recollection tears through any of these laptops’ ample batteries, and the results from LLMs are motionless mixed. I burned through 30 percent of the Lenovo Yoga 9n’s power division in 36 minutes to get a Qwen 3.8 27B example to resize and add a pale backdrop to five merchandise images in PNG format — and it cropped the images incorrect and cut off the products. I got it to do it correct following a brace follow-up prompts, and grabbing the power cable to forestall Hermes from slaying the laptop in the process.

AD: These laptops are extremely expensive, but that additionally method the manufacturers are mostly bringing their A-games. Across the Asus, HP, Lenovo, and what I gathered from Tom’s thoughts on the Surface, the RTX Spark row is awash in great-to-excellent haptic trackpads, keyboards, screens, and speakers. The metal element builds of the three I’m evaluation awareness awesome to the touch. The OLED displays all appearance fantastic. And all of these are correctly lean for the benevolent of workstation-class laptops that they are.

If I had to choice one for myself according to hardware solitary I’d likely choose the Asus ProArt. It ticks the most boxes for me alongside its three USB-C ports, excellent brilliant display, better inverted-T arrow cluster, and complete SD cardstock slot. The HP really feels the nicest, alongside an impressively rock-solid chassis, but at nearly five pounds it’s the heaviest of the bunch, and its microSD cardstock slot is much small helpful to me.

As for the Lenovo, it’s really been my sleeper favorite. It’s the lone 2-in-1 in this crop I’ve been testing, and that fair makes it additional elastic (pun intended) and fun. You can use its ample stylus on either the display or the ample trackpad, giving it many additional possible for graphics and scheme work. And shelter manner alongside a ample 16-inch 120Hz OLED is a pleasance for the occasional gaming meeting or observing shows and movies — it’s fair not as brilliant as the Asus.

SH: I’ve lone really driven the Asus P16 so far, but I was impressed alongside the build norm of the entire spectrum of RTX Spark initiate laptops at Nvidia’s Editor’s Day. The Asus P16 likely wouldn’t be my choice personally since of its additional glossy display and oversensitive touchpad. I akin to use my laptops in rooms alongside many of ambient light, and discover I have to crank up the brightness of this machine’s alternatively excellent calibrated OLED screen. Meanwhile, my palms keep tapping my mouse cursor on things I didn’t average to tap, fair since they accidentally broom the touchpad during typing.

The speakers and webcam additionally depart item to be desired, but the Windows Hello visage login and proximity sensing are excellent. As shortly as I stroll away, the display dims, and I’m logged correct rear in fair by looking at my laptop. Despite the heat, Asus’ fans aren’t shrill — they’re extremely noisy and noticeable during gaming, but at smallest it’s whoosh fairly than whine.

TW: While I ran into several weird bugs alongside the Surface Laptop Ultra during testing, I can’t error the hardware. The appearance and awareness is extremely akin to a MacBook Pro, and the magnetic USB-C harbor really feels akin item Apple should have done — it’s that good. The ample haptic trackpad carries on the tradition of awesome Surface trackpads, alongside the added advantage of subtle haptic feedback whenever you move windows about or align objects in apps akin PowerPoint.

I would have preferred an OLED display on the Laptop Ultra, but I established switching distant from the default “Vivid” color choice to “Enhanced” improved the difference on the mini-LED panel. I additionally really value the harbor division on the Laptop Ultra, particularly that there’s a complete SD cardstock audience and equal a USB-A port. Check out my complete review for additional thoughts.

AD: Yes, the RTX Spark N1X is a mighty chip. These high-end laptop configurations we’ve been evaluation alongside many of unified recollection are meant to vie alongside the Pro and Max chips in Apple’s MacBook Pros. And during they can suspend in several GPU achievement tests, akin tying the top spec M5 Max in a Blender render, Apple’s CPUs are much additional powerful.

Apple has put many of betterment into its in-house processors and the extra-fast “super cores” of its M6, M5 Pro, and M5 Max chips. And, frankly, nothing competes alongside those chips in sole center achievement — not equal Nvidia’s “superchip.” (Seriously, can we halt alongside the term super?) But the part that stings most concerning the RTX Spark’s 20-core CPU is that equal a lowly 12-core M6 Mac Mini competes alongside it in multicore achievement — and a Mac Mini starts at fair $900.

It’s notable to see this benevolent of graphics achievement without the need for a discrete GPU, allowing for thinner scheme designs alongside small noisy fans. I’m into it, fair as I was alongside AMD’s Strix Halo APU — but akin Strix Halo, Nvidia’s RTX Spark is a pricey part. And there’s no denying that Apple’s offerings are additional well-rounded performers, so lengthy as you don’t attention as much concerning games.

TW: It’s certainly up there, but I think Qualcomm comes nearer to Apple’s M-series chips on the CPU side. Qualcomm additionally had years of experimentation alongside Windows on Arm before it acquired Nuvia and managed to create a rivalrous part alongside the Snapdragon X series.

I think Nvidia volition need a akin amount of activity to capture up on the CPU flank for forthcoming RTX Spark chips. The betterment of the RTX Spark took at smallest three years, and I had heard from sources acquainted alongside Nvidia’s plans that the chips were expected to initiate final year. So activity has been ongoing, but the CPU flank is motionless a small disappointing.

Nvidia picked MediaTek for its Arm-based CPU, a chipmaker that is improved known for the Wi-Fi hardware in laptops, not the CPUs. I certain anticipation the next-generation is additional powerful. The entire period I’ve been evaluation the Surface Laptop Ultra I couldn’t assistance but think the forbidden aspiration would be a blend of a Qualcomm CPU and Nvidia GPU. I can’t see that always happening in reality, but it would certainly be engaging to see.

TW: I think it volition elevate involvement in Windows on Arm for imaginative and gaming tasks, merely since we haven’t had Arm-based Windows laptops that are really capable of operating games until now. Nvidia is targeting creators and developers alongside its archetypal circular of RTX Spark laptops, but there’s certainly a gap current for a additional affordable type focused on gaming. Perhaps that’s the N1 chips we’ve heard gossip about.

SH: Nice, but too pricey. Too hot and noisy whenever you’re doing the things that would theoretically validate that price. I attention what they remain for, and the direction they’re pointed, far additional than I attention to own one.

AD: I affection that there is now equal additional competition in the laptop space, but the excellent elevated prices makes RTX Spark laptops mostly unattainable. I desire to see lesser disbursal options arrive around, as fine as an eventual second generation alongside higher CPU achievement to hopefully near the gap additional alongside Apple. I additionally anticipation this helps shove betterment for Arm-based and Arm-compatible games forward. Maybe it could equal one day average easier porting of games to macOS. Wouldn’t that be a fun outcome?

As chill as it volition be to keep evaluation these excellent high-end RTX Spark models, I’m most hyped to get my hands on a basis Surface Laptop Ultra or Asus ProArt P14 alongside 24GB of RAM and see how they stack up to the entry-level 14-inch M5 MacBook Pro. That’s my individual saccharine place for Apple’s lineup — the people’s MacBook Pro. (Well, okay, small these days now that it starts at two-grand.)

It’s an breathtaking period in computing whenever it comes to part choice, but it’s additionally an entirely cursed period gratitude to RAMageddon and sky elevated prices. I’m continually wondering what laptops akin these RTX Sparks or equal Googlebooks power have disbursal if AI demands didn’t run up the prices of RAM and storage. I have no idea who is going to expend up to $7,000 on these laptops whenever PC revenue as a entire seem on a downward trend, but it’s chill they exist.