4:44
When I say I've been waiting for this launch for years, I'm not kidding. Rumors of a MacBook killer from Nvidia have been circulating since 2023, and the idea seems obvious. An energy-efficient ARM processor paired with a powerful GPU from Nvidia, connected to a huge amount of combined RAM. But there is a caveat: a regular creative laptop on Windows has significant limitations. A less efficient x86 processor and a discrete graphics card are usually a recipe for fairly low battery life, especially under load. With the brand new Nvidia RTX Spark, they hope to compete with the MacBook Pro. And they partially succeeded, but there is a catch.
The cheapest version at $2,599 loses almost all meaning of its existence. So I studied this issue from all sides. I have test samples from Microsoft and Lenovo. I've been in a lot of communication with everyone involved over the last few months, and have also been using RTX Spark systems every day for several weeks. At the same time, I am also someone who uses a MacBook Pro for video editing. So sit back , friends, and let's answer the question of who this is for First of all, about the basics. What is RTX Spark? Well, video laptops have been around for years, of course, but they've always been equipped with AMD or Intel processors. This is the first time
Nvidia has built everything on its own since, well, the original Surface RT. Oh, yes. We're not.we're not talking about Surface RT. Dude, he was ahead of his time. So, you get an 18- or 20- core ARM processor, a 50th generation RTX GPU, and up to 128GB of shared memory. The last part is the most important. If you have a regular laptop with Nvidia, you usually only get 8 or maybe 16GB of dedicated video memory. This is fine for games, but many tasks, especially those involving local AI, can use significantly more. For example, as much as 96 GB of RAM allocated to graphics. Yes, friend.
Oh, yes, of course! I immediately turned that slider to 96. Oh, you had to fight for that with a few people in the office. Yes. No, seriously, Matt's like, "Well, uh, 120 gigs of RAM, what are you going to do with it?" And I said, "I'm testing this. " Stay away." Alex and Kinsey say, "I only have 32 gigs of RAM." And I tell them: "Sit down. Yes, we are circling like kites. As soon as the video is finished, hand it over.
They're all just behind me because I don't let them out of my sight, okay? Just so you know. So, I have two Arctic Spark laptops for testing. The Surface Laptop Ultra, which really does convey the same MacBook vibe. There's a USB-C magnetic charger, and a unibody design. Then there's the Yoga Pro 9, which has a gorgeous PureSight Pro OLED display and the same classic Yoga keyboard with a touchpad that supports a stylus. But there is still a price. This, I'm sure, won't surprise you, these are not cheap laptops at all. When the price starts at $2,599 and can go over $6,000. And yes, I really need 120 GB of RAM in a $6,000 laptop. How else will I play Candy Crush? Do you know how much Facebook I could scroll through on this babe? But look, I think if
2026 and the RAM crisis have taught us anything, it's that technology is n't going to get cheaper anytime soon. Apple raised the prices of its laptops this summer, and the Arctic Spark systems coming out today are mostly pretty close in price to the 16- inch MacBook Pro. So if you were thinking of spending that kind of money, as much as I hate to say it, you should probably consider buying the exact configuration you need, because it will almost certainly cost more next year. And while the Surface is fairly user-repairable, the SSD is the only part that can be upgraded. Although I will say that if you are really concerned about SSD capacity, the Yoga actually has two SSD slots. But, despite everything, if you want 64 or 128 gigabytes of RAM in one of
these systems, it's better to order it right away, because the user won't upgrade it. Unless you are a real master of working with a soldering iron. And your balls don't contract with fear. It's probably worth seeing a doctor. I was going to say they're a little close to my face now, but you get the point. So what exactly can you do with all this memory and power? Well, it's almost like a whole series of demos that I made specifically for this video, spending weeks on them , and I hope you like them, because I should have stopped a long time ago. I just could n't. I have 128 gigabytes of RAM. What else can be done about this?
The answer is a bunch of things, as I found out. I don't know if you've noticed, but we make a lot of videos for different channels, short formats, it's really a lot. And these are actually great ideas that you offer in the comments. But I can't go through thousands of videos looking for new comments every day, so a few months ago we built a tool to solve this problem with the help of our sponsor, Zapier. This time, we've upgraded significantly, using the next generation of Zapier. Instead of relying on cloud-based AI, I can now use a local AI model on my laptop to scan hundreds of comments at a time and notify you of the ideas you suggest. The idea is that you can use AI where judgment is really needed-and no, three swimming pool emojis are not an idea
for a video-and use regular automation for everything else. Zapier found that 82% of the AI used by their largest customers is for things that are easily automated: moving fields, checking for new mail, retrieving records. So now my Zap for comment ideas watches for a new video on the channel, and then my laptop takes over At night, it receives a new video, loads comments, applies a few simple rules to filter out spam, and in one step uses local AI to determine which comments are real ideas. The notebook then sends these ideas back to Zapier via a webhook, and Zapier takes over. Each idea is entered into a separate Notion card, and the team receives notifications via Slack when new ideas arrive. Next- gen Zaps are currently in early access, so
check them out via the link in the description. So, I have a few demos of what you can do on Spark. Again, I'll turn on airplane mode. It seems I gave him access to the folder, let's check now. Okay, it sees the CSV file, giving my test results on the Surface Laptop, as well as on some gaming laptops. It may not be as impressive as ChatGPT, Gemini, Claude or something similar, but it works completely locally on the device. If I wanted to, I could ask, "Which one has the best processor?"
I guess I just keep saying, " Wait." Wait. Wait, let's check the numbers again. So, he tells me that the Surface Laptop has outpaced other devices, which is true. There are a lot of things you might want to run locally on your device instead of, I do n't know, uploading your tax returns to ChatGPT, but I think the more interesting demonstration is what you can actually create. Again, completely local to the device. Behold, my friends, an almost real mover. So, it's a copy of the Unreal Engine, the same engine that powers Fortnite and many other fantastic games. It will run on Spark, so it's mostly emulation. However, they now have a real MCP server
built into Unreal Engine, which means I have a local LLM. This is a separate application. For example, you can create a scene, move around, do all of that in Unreal. This is a little rough right now, but give me something you want me to generate. The interior of Gabe's yacht. Or just the yacht of an extremely wealthy person. A fancy yacht, cool, high-tech things. So what's happening now is, again, we're operating completely locally. Basically, it uses that 30B model to parse what it is, refer to Unreal Engine, and drive it. You can see that he is actively building a window. You can also hear the fan humming. It took 64 seconds. So, we have two sofas and an interestingly placed chair. If you're a real game developer, of course you wouldn't
do this, but if you want to do something simple like, I don't know, how about making a sunset? So, he should appear. Boom. I want to clarify that this is a demo, but the cool thing is that I, a person who doesn't know how to use Unreal Engine, can use a few very simple prompts to make it create different things. It also has, uh, you can let it generate everything on its own. It's essentially like "Hey, I'm lucky." That's all. Camp in a forest clearing at night with a tent and a bonfire. Okay, that's pretty quick. Perfectly. And the cool thing is that because it's Unreal Engine, you have support for everything. This thing has, like, full ray tracing. For example, you can
enable DLSS. Okay, let's do one last thing. Uh, make it rain laptops. I don't know if this will work. Oh, laptops, laptops, we have laptops. This engine is really cool, isn't it? I think the main advantage of the Arctic Spark is its use for creativity, because it is suitable for gaming, it is thin, light, you can carry it everywhere and use it all day. It handles all the usual laptop tasks, but you wouldn't spend three, four, or five thousand on a laptop just for everyday tasks, would you? If that's the case, you can find something that's almost as good but much cheaper, right? You want to buy it because it has a huge, simply huge
performance margin. But there are many things you can do simply by running on 120 gigabytes of RAM. So, I'm going to put it in airplane mode and show you a little app that I quickly wrote called Say It. The idea is that you take a video, drag and drop it, and it automatically transcribes the video and actually prepares it for upload to YouTube. For over 10 years, I have been paying the service to create subtitles for each of my videos. YouTube does automatic subtitles, but they are usually not very accurate. So I pay about $2 a minute, and it usually takes 6 to 12 hours for someone to review everything, send me the subtitle file, and then I upload it to YouTube, review the entire video, correct spelling, and make edits. This is a very laborious process.
With this laptop, it can chew through this 23-minute video in about 2 minutes, process the entire video, use a couple of local AI models to transcribe, clean up the audio, etc., and then give me an SRT file that I can immediately upload to YouTube and get high-quality subtitles. I'm not mentioning this because "oh, look how cool this is." It's more about the fact that, in my opinion, these laptops have no equal in such tasks. So, that's the point. It's not just that he's fast. What I mean is that you can do locally on the Spark system what previously required cloud services. This is a very real thing that was talked about in the presentation: you can do more on your device and stop paying for tokens. And that's why, if you really want to
buy an RTX Spark system, you should get at least 48 or 64 gigabytes of memory. Of course, you're spending four thousand not just to run AI. How does Spark stack up against everything else? The processor is not bad. It is on par with Intel and basic Snapdragon chips, and performs better in multi-threaded tasks because it has 20 cores. However, he is not impressive. And to be clear, the top-end Snapdragon X2 Elite Extreme completely outperforms it in all processor metrics. Unfortunately, I don't have a MacBook Pro with the M5 Pro on hand for testing, but judging by public benchmarks, it has a significant advantage in computing power. As for
the graphics, they're not bad. Of course, the usual caveats apply. I have pre-production hardware and a driver update that came in the middle of testing, so I have less time than usual to run all the tests. But overall it falls somewhere between a well-cooled RTX 560 and RTX 570. Quite decent and significantly ahead of the graphics in Snapdragon chips. However, in real-world games like Cyberpunk, Spark lags far behind. To be fair, this game is emulated on the CPU, so there is a significant performance gap depending on the game's optimization. This works for unoptimized games and apps, of course, but you definitely lose out on performance. But, admittedly, you still have much more gaming capabilities than on any MacBook, and
you can get quite acceptable performance, since the situation with Windows on ARM has improved significantly. You may still encounter anti-cheat issues, but for the most part everything works or will be fixed in the near future. Regarding the battery, I have good news: unlike most gaming laptops, as far as I can tell, Spark systems do not lose performance when running on battery power. Microsoft proudly announced this from the stage, and I can confirm: just like with the MacBook, you get the same performance whether you're plugged in or not. As for real-world battery life, at best, when browsing the web, you can expect
around 10 hours, and it's worth noting that the standby time is also quite good. I recorded only 4% battery drain overnight in sleep mode, which is about in line with the MacBook's performance. But there is also a disadvantage of the lack of performance limitations. If you really want to squeeze the most out of these Spark systems, for example by running a huge AI model non-stop on the battery, you can drain it in less than an hour. Once my battery died in 46 minutes. I think most people should expect about 6-8 hours of work. Of course, it won't compare to a MacBook or a Snapdragon laptop, but it's significantly better than a gaming laptop. And a good reason to buy one of these is the form factor. I mean, if you put a Yoga or Surface next to a
gaming laptop, you'll quickly see and feel the difference. But of course, even if you spend three, four, five or six thousand dollars, you still get Windows with all its pros and cons. The first setup still forces a Microsoft 365 subscription, OneDrive, and Yoga keeps pushing McAfee, which I, of course, immediately uninstall. The Windows team has made some progress this year. I have to admit, they tried to make Windows a little more peaceful, without so many pop-ups and junk, but there's still a lot of work to be done. I will say that I've been really liking the Going back to 16x10 feels a bit cramped, but it's a
trade-off. Microsoft has installed a very decent mini LED display, which is incredibly bright. I used it in the car in the sun and had no problem seeing the screen, but it still does n't look as good as OLED. And I really wish they would find a way to fit a tandem OLED into this case. The rest of the Surface hardware, however, is simply fantastic. The touchpad is huge and supports Windows' new haptic feedback, so you actually feel the vibration when you do things like snap a window. Uh, the keyboard is actually surprisingly good. This is a real step forward compared to previous Surfaces. You get a magnetic USB-C adapter, which is incredibly cool.
The idea is that you don't have to sacrifice an entire port like with MagSafe, because magnetic USB-C is also a regular USB-C port. In addition, it has very good speakers. The Yoga, however, clearly wins when it comes to the display. He is really, really good. This is 165 hertz. This is OLED, extremely bright. It's clear to me that Surface is the choice if you want a MacBook-like feel. And Yoga is what you need if you want a better display. But the main question is: is it worth buying any RTX Spark system? That is, I can say how I feel about it. The Surface Laptop Ultra will be my new primary work device. But I'll be honest, there are still times when I need to edit
videos, and then I'll still take my trusty old M1 MacBook Pro. For 95% of what I do every day, the Ultra does just fine. But those 5%, you know, when I go to an event and I need to urgently edit a video, I only know how to work in Final Cut. I haven't used any other editors in over 10 years. So I admit, this is more of a personal issue for me, but it applies to almost everyone who is going to buy one of these RTX Spark systems. Make sure the applications you rely on actually work well on Spark. For the most part, they work, but there are still some gaps. There are unoptimized plugins, and I encountered some of them when I was creating the AI demo earlier in this video.
Many things work really well with an RTX GPU, but tasks that should be done on the CPU sometimes require emulation or simply don't give you the full performance of Intel or AMD processors. You just need to consider what actually works today, not what you hope will work in the future. Look, I want to make it clear that they're not perfect, right? These Spark systems are expensive, and in the end you're not just running Windows, you're running Windows on ARM. That's fine, as long as it's just " normal." This is truly a versatile tool. A Windows laptop that you can play games on, create content on, run huge local models on, and it mimics a MacBook pretty well. All I can say is, finally, [__] it
. If you liked it, be sure to subscribe to the channel and click that bell, friends. Let us know how this ignited your deep, passionate love for Windows. Actually, on second thought, maybe it's not worth sharing. I don't really need to hear this.
