➡ Intel and AMD are producing powerful chips that are compatible with both Linux and Windows, making them ideal for running local AI applications. System76, Dell, and Lenovo are recommended manufacturers for Linux-compatible computers, with specific models like Dell’s XPS line and Lenovo’s Legion series being particularly reliable. While these high-performance computers can be more expensive, they offer long-term compatibility and are a good investment for future-proofing against evolving technology. However, HP is not recommended due to issues with Linux compatibility.

➡ Brax III Privacy Fund, a joint project by several companies, is a key tool in protecting privacy from big tech’s location tracking. It’s one of many products on BraxTech.net, including Brax Virtual Phone for anonymous calls, BraxMail for multiple email identities, and BiceVPN for IP address protection. These services are discussed by over 100,000 users on our social media site, BraxMe.

If you watched my videos last year, you may recall that I told people to hold off buying a new laptop in 2024 because of so many potential issues, incompatibilities, and unsafe features. I did end up buying a new laptop in December 2024, which was a Lenovo Legion 5, but this one had a late 2023 Intel chip. So this did not have the issue that bothered me, and it’s still as powerful as newer 2025 computers. But now here we are in almost mid 2025, and if you’re looking at possible choices for laptops, it’s going to be massively confusing.

First, there are so many computers promoting their versions of AI like Windows Co-Pilot or Apple Intelligence. Now, there are ARM-based laptops, even for Windows, versus new types of x86-based laptops. The new units are all powered by NPU chips or neural processing units with capability for running local AI. And you’re all confused, I’m sure, if this is a good thing or a bad thing. Then, of course, there’s the choice of operating system. This is not a simple choice, as going with a particular ecosystem locks you out from application choices or even gaming options. Some of you may be power users who want to do more complex things like me, doing local AI and video editing.

Some of you have listened to me and understand that your choice needs Linux compatibility, either for now, dual boot, or for future conversion. Folks, the wrong choice will mean that you may end up with a dud computer that will be nothing but a surveillance machine that is watching you with no options available to you, or it may end up with application incompatibility with things you do and you’ll be forced to buy a replacement. This is an important guide to potential buyers of computers. You may only be lurking looking at what’s available, but still, you need to understand this new landscape that is confusing as heck, even for a tech-savvy person like me.

Stay right there. Let’s first make sure we know the laptop players. First, you have to make your choice regarding operating systems. If you choose an Apple macOS device, then you will be locked into the Apple ecosystem. This means you really will have no choices with alternate operating systems, and you will also be prevented from running many popular games. The main claim to fame for Apple is long battery life with great performance, but this is thrown against the potential negatives. As your privacy advisor, I would not recommend Apple at all, and I will go into discussion of AI features a little later for all manufacturers.

The other most dangerous choice is anything with Windows 11 Copilot. Well, this has big negatives again that I will discuss with AI features separately. But the new split is between Windows computers that are x86 based, the old traditional way, and the new Windows computers from Qualcomm called Snapdragon Elite. These are ARM-based CPUs. While both are Windows computers, the Snapdragon CPUs are currently not Linux compatible. So if you get one of these, then you’re kind of stuck with Windows. I’ll get into more about Linux compatibility a little later as well, and I’ll tell you why this is important.

Then even among x86 CPUs based on AMD and Intel, you have another fork in the road. AMD is the most confusing product naming conventions you can imagine, and it’s hard for a casual techie to understand what you could be buying. Intel doesn’t make it easy either. New CPUs called Lunar Lake for laptops and Aero Lake for desktops have their own issues. So there I mention all the players, and it’s time to get to the nitty gritty. The biggest push among all manufacturers of computers, and the OS makers Apple and Microsoft, is of course to sell you on buying a new computer because of AI.

Well, I’m going to tell you that in 2025, I’m happy to say that this concept is not selling. People are not getting why you need to run a local AI that is embedded in your operating system, meaning Windows or Mac OS. And in fact, people are realizing the truth about the privacy invasion. Look at the super stupid concept of Windows Copilot, for example, with the feature called Windows Recall. Windows Recall uses screenshots every few seconds to capture what you’re doing on screen. This then removes any chance of privacy because the computer basically uses a technology called See What You See.

So there will be no private communications or even private viewing of any media. The AI will know everything you do. The supposed benefit is that the AI will know you deeply and can function as your personal assistant. I’ll discuss this failed concept in another video a little more. But in the meantime, people are not understanding the application of AI in real use, privacy issues aside. And while we already know the technique Microsoft will be doing, Apple has already previewed the same approach of scanning media using their old module called Media Analysis D. So while the specifics of Apple Intelligence are not yet clear, the mechanism has to be the same.

They see what you see. Both Apple and Microsoft use the chat GPT based AI from OpenAI. So they use the same technologies, thus expect them to do similar things. The main takeaway here is that you must not make a choice of computer based on who you think has some safer AI, because I assure you they will be equally bad. And Apple was first in scanning your content or the term is client side scanning. New computers all have NPU chips built in, and many of you are avoiding computers with NPU chips, thinking you will be safe from this AI invasion of your device.

Well, you would be attacking this incorrectly. NPU chips are actually like math coprocessors. Their main task is called matrix multiplication. This particular kind of math is heavily used in AI. NPUs themselves don’t have any brains, though the term neural processing unit may make you think that. Not all AI is bad, so it’s important to distinguish. There are two kinds of AI you can run on your computer. There’s the AI that someone forces on you, like Windows Copilot and Apple Intelligence, and there’s the AI you control yourself and install yourself. For example, you can install Olama AI locally and use any open source model.

This is something I do. To run your own AI with complete privacy, you will need to have a powerful computer, one with a powerful NPU, or a discrete graphics card like NVIDIA that’s a newer model. This is the kind of computer that will do well in video editing and gaming. But if you have no interest in AI video editing or gaming, then you would fall in the average workload category, and you will have more choices, though you should still understand what I will present for future proofing. Linux compatibility is a big deal. You might think that you’re not currently a Linux user, so I would just matter.

It matters because if you’re running an M Silicon Mac or Windows 11, then the cloth of AI watching whatever you do will begin to grip you, and you will need to find a way to escape that. The only escape is to use an operating system without embedded AI, and the only practical choice at the moment for newer computers is to use Linux. Let me give you the Linux compatibility landscape in 2025. And mind you, some of this information is very new. I would have given a different answer in late 2024. If you’re using an Apple M Silicon computer, M1 to M4, then there is a single Linux distro you can use, and it is Asahi Linux.

While this is good enough as an occasionally used Linux computer, it is not full-featured, and there are many missing capabilities. First, it will not utilize the full capability for power saving that Mac OS provides, so this cancels out the main benefit of M Silicon chips. Then many basic features like hardware video decoders will not work. So for practical use, this is troublesome. This may improve in the future, but when it will be ready for prime time is anyone’s guess. The next product group with huge issues with Linux compatibility is Snapdragon Elite computers. Because Qualcomm is having compatibility issues with Windows apps and are basically having a hard time with their reputation, it is apparent that they basically paused their investment of time in Linux compatibility.

It’s almost like Linux development has come to a standstill. So at this point, do not buy Snapdragon laptops as you will be stuck with Windows co-pilot and many Windows app incompatibilities. This would be the most troublesome selection right now. I had high hopes for Snapdragon because of power saving options, but aside from what I mentioned, its GPU performance is also subpar. It’s really hard to recommend Snapdragon to any in my community, as I cannot see any good points. No way out of the privacy problems when you’re locked into Windows co-pilot and Windows Recall.

When it comes to Intel, there’s some signs of good news. For example, Ubuntu has now certified certain laptop models as compatible with the newest Lunar Lake chips. Intel Lunar Lake chips, which are the ones now called Intel Core Ultra Series 2, are AI-ready CPUs with a fairly capable GPU called ARC. This means they can run local AIs on Linux. So perfect for running Olama. They have low power use similar to Snapdragon while sticking to the same x86 architecture. And because they are true x86 with no emulation, they are compatible with standard Windows apps, including games.

This means that depending on the laptop they’re on, these would be recommended in general for both local AI use and Linux. If you’re interested in running more capable AI features in Linux, then I would advise sticking to those CPUs with NPUs or those with an NVIDIA GPU. So that limits you to models no older than 2023. The older Intel chips called Meteor Lake are typically in the 13th or 14th gen, and you will know this because the Intel CPU will start with 13 or 14. But I would only buy these models if the price is right and if they have an NVIDIA GPU.

Otherwise, I would buy a newer computer to ensure long-term compatibility. By the way, Intel is announcing a new generation of chips that moves Lunar Lake forward some more. The idea is to minimize the gap between x86 and Apple M Silicon. So that’s slated for end of the year, I think. Now let’s talk about AMD. In case you’ve lost track, AMD is now a much bigger company than Intel. And the reason this is the case is because their products have been performing a lot better. For example, power users have been shifting to the more powerful Ryzen line of CPU chips and they are really quite popular with desktop PC users.

The problem is that you don’t see a lot of AMD-equipped laptops. I don’t know why this is, but usually when I’m looking for AMD laptops, the one manufacturer that seems to have them is Asus. Asus also invested quite a bit in the Snapdragon line, which I told you to avoid. But you don’t have to avoid AMD. The neat thing about AMD is that it is mostly Linux compatible. The chip from 2024, which is the line called Zen 4, is not going to give you a problem with Windows or Linux. But there’s always minor glitches with new chips, like computers using the new Zen 5 architecture, which is from 2025.

So that may require a few Linux updates to resolve minor issues. If you go with a high-end Zen 4 and Zen 5 chips from AMD, you will be surprised because their GPUs perform just slightly below their equivalent Nvidia counterparts. This means you should have good AI processing performance while saving a few bucks. Recommended Model Families Understanding my focus here on more long-term power uses like local AI and Linux compatibility, I’m going to tell you my experience with certain model families of computers. The one that surprised me recently were computers from System76.

Usually, System76 has old CPUs. Since being a smaller manufacturer, they cannot get the latest and greatest. But the main benefit from System76 is absolute Linux compatibility. System76 makes the distro pop OS, which I also use. It is based off Ubuntu, and I was happy to go to their website and actually see that their new computers are running lunar-like CPUs. If you’re not an expert Linux user, this is the best option to get one with Linux straight out of the box. You can still dual boot with Windows if you wish, but at least you have the advantage of full compatibility with both.

Also, if you change the hardware on a System76, they do not invalidate your warranty. And while prices on System76 used to be on the high side, they are now quite competitively priced. The other manufacturer selling Linux computers is Starlabs. This is a good lower-end choice, but be aware that they are using older Intel chips, so these are not as suitable for the power user doing AI. Their computers have Linux pre-installed as well, just like System76. Maybe next year they’ll have the higher-end chips. I had a Starlabs in the past, though I broke it some time ago.

There are some particular manufacturers that make product families that seem to always be Linux compatible. The two companies that have a really good track record here are Dell and Lenovo. From Dell, the most reliable model family has been the XPS line. I personally had and still have a Dell XPS 13 and Dell XPS 15. Both of these are problem-free with Linux compatibility. The XPS 15 comes with an NVIDIA GPU option, so it is great for AI work. The new Dell XPS 13 with Intel Core Ultra Series II, meaning Lunar Lake, should also function well with AI.

You cannot change the memory on a Lunar Lake chip, so you would have to buy 16GB at the very least, and possibly 32GB for long term. The XPS 13 is specifically in the ultra-portable thin and light category, so this is a great all-arounder. While the consumer models of the Dell line likely also have good Linux compatibility, the XPS line is sold to businesses, so it is more foolproof. Even used, these are pretty solid. So a used Dell XPS 15 with an NVIDIA GPU would still function fine today. From Lenovo, I’ve had really good success.

Since they cater primarily to business, they almost always have Linux compatibility and have good power, though they’re more on the pricey side. I currently have a Lenovo X1 Carbon, which is an ultra-portable like the Dell XPS 13. The new model with Intel Core Ultra Series II will of course perform well for the long run. They are expensive as heck, though. I have a very powerful Lenovo Legion 5, which is actually marketed as a gaming laptop. The Legion series is highly recommended for AI work, and they’re all Linux compatible. I bought a model that was a year old already, so I got a good price on it, as it had an NVIDIA 4070 GPU on it.

I also saw that you can find it at Costco with a 4060 NVIDIA for even cheaper. In terms of capability, the 14th Gen Intel’s are just as powerful, if not more than the Intel Core Ultra Series II. But the new chips use less power and thus have a longer battery life. So buying a 2024 Intel laptop is not a bad idea from the likes of Dell, Lenovo, or Asus. There are other model families that will have AMD Zen 4 or Zen 5 architecture like the ones from Asus, and that would also be recommended.

Other manufacturers that I’ve not had a problem with Linux compatibility has been MSI, which is popular with gamers. One major manufacturer that I don’t mention is HP. I used to be a big user of HP, and I had a new HP laptop every year. But then I encountered Linux incompatibility and I never went back. If they put in new hardware and don’t invest time to send their drivers over to Linux for inclusion, then they’re a bad choice in my book. Likely older HPs, which will not be good for AI use, will have Linux compatibility.

But new HPs will always be a crapshoot. It’s really too bad as they are well built and perform nicely under Windows. But we are in a new age, and we need to be able to respond with solutions when OS makers forcibly want to see what we see. If you’re more of an average user and do only simple tasks on a computer like word processing, spreadsheets, browsing, and watching videos, then any cheap computer would work. But again, the important capability is Linux compatibility. So what I said about choices are still valid and still the same.

The difference, if you’re planning on doing more high-end tasks, is the need for an NPU and GPU even under Linux. While the need for this may not be apparent at this early stage, we are likely at a fork in the road where older computers without the processing power of AI may not be able to run apps in the future. This is kind of like the time we went to Windows from DOS, though perhaps not so extreme. Thus, my recommendations are more about future proofing and also the capability for privacy protection.

You don’t want to be locked into a surveillance device and have no options due to lack of funds. But understand that the extra capabilities also mean a higher price. While low-powered computers can be had for $500, the more AI-capable computers will hover closer to $1,500. They will, however, have quite a bit longer lifespan of use. Or you can buy them used next year to save money and they will likely be on sale at the end of the year. Again, I keep going back to Linux compatibility as a major requirement.

I’m warning you now that very soon neither Windows nor Mac OS will be safe even for casual use. Folks, as many of you know, this channel does not have sponsors. Instead, we rely completely on community support to keep us going. Hopefully, we give you enough information that you find good value. Some of you support us on Patreon, Locals, and YouTube memberships. And though I don’t publicly acknowledge you for privacy reasons, you are very appreciated. Thank you. For most of our support, we instead strive to provide products and services that help you in your privacy journey.

Our newest product is the Brax III Privacy Fund. This is a community project involving several companies and is sold on BraxTech.net. This is an important tool in the privacy battle as it is immune from things like location tracking from big tech. We have other products and services that can be found on our privacy social media site BraxMe. We have Brax Virtual Phone that gives you inexpensive, no-KYC phone numbers for better anonymity. We have BraxMail, which offers unlimited aliases and many domains to create different email identities. We have BiceVPN, which protects your IP addresses from being harvested by data brokers, as well as protecting you from ad spyware.

Join us on BraxMe and meet the way over 100,000 users who discuss privacy issues daily. Thank you for watching and see you next time. [tr:trw].

See more of Rob Braxman Tech on their Public Channel and the MPN Rob Braxman Tech channel.