The "Microsoft Tax" is often cheaper than the "Linux Engineering Salary." While Linux alternatives exist, they require "assembly"—integrating LDAP, Kerberos, DNS, and config management (Ansible/Salt) to do what AD does out of the box.
Most businesses don't want to be in the business of maintaining their own identity infrastructure. They want a utility. Between Group Policy’s granular control over the endpoint and the tight integration with Exchange/M365, Microsoft has created a "sticky" ecosystem. I've tried the "DIY" route with Linux mail servers, and the friction of maintaining deliverability and security patches manually is a nightmare compared to the "it just works" nature of the Microsoft ecosystem.
I am not a system admin, so maybe this is a crappy take.
This guy is in for a disappointing future since he seems to be unaware that Windows is more than than the consumer editions. Revenue from Windows Enterprise (which has management tools like Active Directory and backwards compatibility with non-game apps needed for large corporate deployments) and Windows Server (needed for Active Directory, Exchange, SQL Server, etc.) is still in the billions and there's nothing on the horizon in the Linux ecosystem to replace those. Given that Microsoft is going to have to continue to develop Windows anyway, there's not much reason for them to throw in the towel on the consumer desktop.
I can't really envision what Microsoft stands to gain by doing this. If Windows were to become a Linux distribution, what reason would anyone have anymore to buy Windows, or Windows Server? I can run Linux for free. Any kernel modifications Microsoft makes would have to be open-sourced, due to the GPL, so I'm not missing out on those either. What's the unique value proposition at that point?
Prediction for the next 15 years: new operating system called Uni. AI will finally get to the point when it can actually reliably replicate the entire OS stack and it will be Windows. I mean Windows stack will get integrated into Linux, but not by Microsoft. It will be done by AI/Linux enthusiasts.
Won't happen in 2026, since AI coding is still dumb, and Cursor failed to produce a working version of a browser (despite claiming that). But soon binary files will be reverse engineered, and the whole NT kernel stack will be transferred to Linux.
At the same time there is a chance that new, AI-produced, fully Windows-compatible operating systems will start to emerge. OS similar to Windows 95.
As a Windows system and user-mode dev, I absolutely never understand these sorts of drive-by posts with nearly zero technical depth.
If there is one thing about Windows that is really good, it is its kernel and driver architecture, and absolute plethora of user-mode libraries that come with the OS, that can be programmed against with a variety of languages from ancient to brand-new, all maintained by the vendor. Doing the same thing on a given distro of Linux is a headache at best, and impossible at worst (which is partly why game developers don't target native Linux).
The problems with Windows have always been in the user-mode (with the notable exception of Vista, and I still maintain that Vista was OK; its problems were due to Intel strong-arming MS into certifying a broken version of Vista for its sub-par integrated GPUs of the time). Windows 11 control panel sort-of gone? There's still the god-mode menu introduced in Vista. Right-click menu gone, or too much Copilot? Go to Group Policy editor, switch off what you don't need; revert what you can. People complain you 'cannot create local user accounts any more'. Also not true, that feature is a fundamental part of Windows and probably won't ever be removed. There are workarounds. Any Windows user or sysadmin worth their salt will have a GPE fleet-wide policy, and registry settings.
Everything one sees on Windows can be stripped out and reverted to Windows 2000 mode. That grey boxy UI is literally still there. Compile a program for 32-bit, set the compatibility mode to Windows 2000, and bam, there you go. If you add in the manifests for UTF-8 and high pixel density, the UI is scaled pixel-perfect by the system.
Speaking of high pixel density, Windows is the only OS that does scaling properly. macOS just pretends non-'retina' displays don't exist, Linux distros are a minefield of Xorg, Wayland, a million different conf.d files, command-line arguments, and env variables.
Why would anyone want to replace their core product with something that a) they cannot control, and b) does not satisfy their business and customer needs?
In enterprise land, managing Windows endpoints is an exponentially larger PITA for the very reason that Microsoft can’t even secure their own OS by default or design, and spend more time shoehorning more surveillance and telemetry into OSes than actually improving them. As “traditional” enterprises increasingly move away from on-prem Active Directory and GPOs in favor of MDM policies and SSO providers, the traditional Microsoft central stack becomes more of a liability than an asset.
From a manufacturing perspective, Microsoft is arguably one of the worst partners you could have - especially if your product has to be operated offline or in restricted modes. I’ve spent two weeks trying to debug kiosk mode on W11 creating wildly inconsistent logon times compared to W10, and this is just the latest wrinkle in a year of triage and wildfires directly caused by trying to use online-first Microsoft kit in offline-only products. I’ve spent my entire year banging on about how Linux solves much of our product line issues, but the old guard is coasting until retirement with no drive or incentive to change until after they’ve left - a cohort that’ll be 90% gone by 2030.
Then you add in the waves made by gaming companies and communities on the platform, and an increasing focus on the OS by developers worldwide seeking to free themselves of Microsoft and Apple taxes, and the memory shortage/AI bubble driving a need to operate with less capable machines, and at the very least it’s plausible that Linux does indeed become the de facto OS.
Really, the only things effectively holding back wider adoption are:
* User experience remaining wildly inconsistent between Linux distros and Windows machines. Enterprise distros don’t focus on bridging that gap at the moment (they’re more aligned to Mac or Unix users migrating to Linux), but I’d be shocked if there isn’t a direct Windows-alike by 2030 with Enterprise support options.
* Endpoint management remains a bugbear for MSPs and Enterprise teams precisely because Linux wasn’t engineered for non-technical usability so much as security. As more distros bake in support for Ansible or other endpoint management schemes out of the box, and more sweatshop-tier technical talent gain experience in Linux, this is going to gradually become a non-issue. The infinitely harder sell will be convincing businesses they don’t need stupid automated scores and algorithms like Microsoft shoehorns into M365, as those are privacy and security (and thus, legal) risks.
* Linux is software-secure but not hardware-secure, as in anti-theft or recovery mechanisms. Businesses want parts-pairing so we can better detect or identify intrusions, as well as remain compliant with the bevvy of frameworks and standards out there that mandate strict hardware controls. This is what mandates Windows and Microsoft tooling in a lot of environments, as they expose and utilize these controls by default. That said, Linux is also making major inroads in addressing these issues, and I expect them to be at or better than parity with Windows long before 2030; it’s also not fair to begrudge Linux about this, since a lot of it comes from Microsoft trying to kneecap competition.
Folks like to point to the gaming situation and say that’s why Microsoft will kill Windows, but I say the opposite: businesses want to kill Windows to save on costs, and will take the first affordable off-ramps they come across. A RHEL/SUSE/Ubuntu Enterprise distro that is immediately compatible with most Windows binaries and is backed with documentation and support will devour Microsoft’s lunch.
I would never predict this, but I think it's an idea which has occurred to many people over the years. the name WSL always hinted to me at a group inside MS who wanted Windows to BE the subsystem.
I think people who have run UNIX over non-traditional FS get this vibe too. We're used to thinking it has to be some linear progression from FS to VFS to UFS to "all the other FS" and the idea "nah, I can run on NTFS just fine thanks" never occurs to us. But DOSBOOT.EXE to boot unix from DOS...
I think they're going to do the least funny thing imaginable: make Windows software require a valid Windows install.
It might all be moot tho if nobody can buy RAM and we're all pushed to cloud computing (yay Azure...). Then your terminal's OS will be pretty irrelevant.
Because the "UX" of Windows sucks, Microsoft will switch the "Kernel" to Linux?
I don't get the argument. There are parts of Windows I don't like, so I have chosen a 3rd-party (often open-source) replacement. The exact same process as I do on Linux. I don't see why I have to switch to Linux to have that freedom.
(and to be honest, I don't care where the taskbar is)
Checks the bio Ahh, Zero Sysadmin experience so does not know about amount of legacy garbage being run at many companies and Enterprises.
Amount of software running on .Net Framework is mind boggling. If there is not 100% compatibility with .Net Framework on Windows running on Linux, forget it. I know of a company still using Visual FoxPro in 2020 and it was still being maintained.
Just like COBOL, across insane amount of businesses/enterprises/government, there are hordes of Windows machines, using technology that last saw updates in early 2000 computing away. Their last supported Windows Server was probably 2008 but somehow they still run on Windows Server 2019 and those licenses are not cheap.
Sure, Windows Desktop is clearly becoming "Whatever" by Microsoft but it's also pretty cheap. NT Kernel and UI work has to be done for server side and until that cash cow is dead, shoving slop into Windows Desktop is cheap revenue stream on work they have to do anyways.
Really the only thing that will move this is market share.
With Claude Code and other tools making it super easy to interact with a computer, and Microsoft repeatedly shooting themselves in the foot adding AI everywhere to Windows 11, I've seen a noticeable amount of people actually switch to Linux, despite every year being the year people switch to Linux - AI in your OS that you can't control seems to be the actual tipping point.
First it would start with the personal / hobbyists, then very slowly the corporate world (they will always play things safely and slowly), assuming Windows continues to decline, which it almost certainly will.
I think stranger things have happened, but I don't really believe this is all that likely. Windows has sucked for 30 years now; tacking on another 15 probably won't change all that much about the current state of things.
Microsoft is an enterprise, and enterprises will continue to crank out enterprisey stuff. Linux is free and open source, developed by people with passion - some of it, I assume, is out of necessity. Unless the working world dramatically changes over the next 15 years, Microsoft is still going to Microsoft.
Windows sucks, Azure sucks, Office sucks. Microsoft is a corporation designed to make money, they have a deadlock on the market. From an investor's point of view, they're doing just fine. From a shareholder's point of view, uprooting the entire Windows base to make tech people happy isn't worth the investment. Microsoft hasn't been about making tech people happy since it went public. Microsoft makes money and employs people. People half-heartedly go to work to earn a living, they produce enterprise-grade software. Enterprise software makes money. That's all the investor cares about.
Actually, as a matter of fact, having Windows around to drive the continued development of Linux might be a good thing. I know Windows sucks, I know virtually anything technical is dramatically easier on Linux, but anything without competition eventually stagnates. Even if Windows exists simply as a "What not to do" in Linux, it's probably good that it remains around.
Currently typing this on a machine that dual-boots both Windows and Linux. Why? Because my laptop came installed with it.
Nope. The Windows NT kernel is fantastic. It's the Windows user mode apps that people complain about. Replacing the Windows NT kernel with the Linux kernel while keeping the user mode crust serves no purpose.
I don't disagree with the overall point. I do want to say one thing though.
> As a professional programmer, I no longer consider Windows a viable option for serious work.
Please get over yourself. There's plenty of actually serious programming work being done on Windows.
> If you’re a programmer who’s used to Windows and you think I’m being overly harsh, I encourage you to spend a couple weeks in any other operating system.
For the record, I've spent decades in many other operating systems. It's interesting because the OS used to matter. Now 90% of the apps we use are either on the web or are web apps repackaged as desktop apps. Of course I can still tell when I switch between OSes, but it makes much, much, much less of a difference than it used to.
To my knowledge, it's not possible to achieve anything like this layout on Windows 11, Linux or Mac. I did try it in various Linux distros a few times but frankly got sick of navigating the maze of window managers et cetera. I think something like XFCE came close to providing a Windows-like taskbar but it was still far, far behind what Windows NT can offer.
63 comments
[ 2.5 ms ] story [ 63.3 ms ] threadMost businesses don't want to be in the business of maintaining their own identity infrastructure. They want a utility. Between Group Policy’s granular control over the endpoint and the tight integration with Exchange/M365, Microsoft has created a "sticky" ecosystem. I've tried the "DIY" route with Linux mail servers, and the friction of maintaining deliverability and security patches manually is a nightmare compared to the "it just works" nature of the Microsoft ecosystem.
I am not a system admin, so maybe this is a crappy take.
Won't happen in 2026, since AI coding is still dumb, and Cursor failed to produce a working version of a browser (despite claiming that). But soon binary files will be reverse engineered, and the whole NT kernel stack will be transferred to Linux.
At the same time there is a chance that new, AI-produced, fully Windows-compatible operating systems will start to emerge. OS similar to Windows 95.
If there is one thing about Windows that is really good, it is its kernel and driver architecture, and absolute plethora of user-mode libraries that come with the OS, that can be programmed against with a variety of languages from ancient to brand-new, all maintained by the vendor. Doing the same thing on a given distro of Linux is a headache at best, and impossible at worst (which is partly why game developers don't target native Linux).
The problems with Windows have always been in the user-mode (with the notable exception of Vista, and I still maintain that Vista was OK; its problems were due to Intel strong-arming MS into certifying a broken version of Vista for its sub-par integrated GPUs of the time). Windows 11 control panel sort-of gone? There's still the god-mode menu introduced in Vista. Right-click menu gone, or too much Copilot? Go to Group Policy editor, switch off what you don't need; revert what you can. People complain you 'cannot create local user accounts any more'. Also not true, that feature is a fundamental part of Windows and probably won't ever be removed. There are workarounds. Any Windows user or sysadmin worth their salt will have a GPE fleet-wide policy, and registry settings.
Everything one sees on Windows can be stripped out and reverted to Windows 2000 mode. That grey boxy UI is literally still there. Compile a program for 32-bit, set the compatibility mode to Windows 2000, and bam, there you go. If you add in the manifests for UTF-8 and high pixel density, the UI is scaled pixel-perfect by the system.
Speaking of high pixel density, Windows is the only OS that does scaling properly. macOS just pretends non-'retina' displays don't exist, Linux distros are a minefield of Xorg, Wayland, a million different conf.d files, command-line arguments, and env variables.
Why would anyone want to replace their core product with something that a) they cannot control, and b) does not satisfy their business and customer needs?
In enterprise land, managing Windows endpoints is an exponentially larger PITA for the very reason that Microsoft can’t even secure their own OS by default or design, and spend more time shoehorning more surveillance and telemetry into OSes than actually improving them. As “traditional” enterprises increasingly move away from on-prem Active Directory and GPOs in favor of MDM policies and SSO providers, the traditional Microsoft central stack becomes more of a liability than an asset.
From a manufacturing perspective, Microsoft is arguably one of the worst partners you could have - especially if your product has to be operated offline or in restricted modes. I’ve spent two weeks trying to debug kiosk mode on W11 creating wildly inconsistent logon times compared to W10, and this is just the latest wrinkle in a year of triage and wildfires directly caused by trying to use online-first Microsoft kit in offline-only products. I’ve spent my entire year banging on about how Linux solves much of our product line issues, but the old guard is coasting until retirement with no drive or incentive to change until after they’ve left - a cohort that’ll be 90% gone by 2030.
Then you add in the waves made by gaming companies and communities on the platform, and an increasing focus on the OS by developers worldwide seeking to free themselves of Microsoft and Apple taxes, and the memory shortage/AI bubble driving a need to operate with less capable machines, and at the very least it’s plausible that Linux does indeed become the de facto OS.
Really, the only things effectively holding back wider adoption are:
* User experience remaining wildly inconsistent between Linux distros and Windows machines. Enterprise distros don’t focus on bridging that gap at the moment (they’re more aligned to Mac or Unix users migrating to Linux), but I’d be shocked if there isn’t a direct Windows-alike by 2030 with Enterprise support options.
* Endpoint management remains a bugbear for MSPs and Enterprise teams precisely because Linux wasn’t engineered for non-technical usability so much as security. As more distros bake in support for Ansible or other endpoint management schemes out of the box, and more sweatshop-tier technical talent gain experience in Linux, this is going to gradually become a non-issue. The infinitely harder sell will be convincing businesses they don’t need stupid automated scores and algorithms like Microsoft shoehorns into M365, as those are privacy and security (and thus, legal) risks.
* Linux is software-secure but not hardware-secure, as in anti-theft or recovery mechanisms. Businesses want parts-pairing so we can better detect or identify intrusions, as well as remain compliant with the bevvy of frameworks and standards out there that mandate strict hardware controls. This is what mandates Windows and Microsoft tooling in a lot of environments, as they expose and utilize these controls by default. That said, Linux is also making major inroads in addressing these issues, and I expect them to be at or better than parity with Windows long before 2030; it’s also not fair to begrudge Linux about this, since a lot of it comes from Microsoft trying to kneecap competition.
Folks like to point to the gaming situation and say that’s why Microsoft will kill Windows, but I say the opposite: businesses want to kill Windows to save on costs, and will take the first affordable off-ramps they come across. A RHEL/SUSE/Ubuntu Enterprise distro that is immediately compatible with most Windows binaries and is backed with documentation and support will devour Microsoft’s lunch.
I think people who have run UNIX over non-traditional FS get this vibe too. We're used to thinking it has to be some linear progression from FS to VFS to UFS to "all the other FS" and the idea "nah, I can run on NTFS just fine thanks" never occurs to us. But DOSBOOT.EXE to boot unix from DOS...
The first clue will be a version of the Xbox running an OS with this model.
It might all be moot tho if nobody can buy RAM and we're all pushed to cloud computing (yay Azure...). Then your terminal's OS will be pretty irrelevant.
I don't get the argument. There are parts of Windows I don't like, so I have chosen a 3rd-party (often open-source) replacement. The exact same process as I do on Linux. I don't see why I have to switch to Linux to have that freedom.
(and to be honest, I don't care where the taskbar is)
Amount of software running on .Net Framework is mind boggling. If there is not 100% compatibility with .Net Framework on Windows running on Linux, forget it. I know of a company still using Visual FoxPro in 2020 and it was still being maintained.
Just like COBOL, across insane amount of businesses/enterprises/government, there are hordes of Windows machines, using technology that last saw updates in early 2000 computing away. Their last supported Windows Server was probably 2008 but somehow they still run on Windows Server 2019 and those licenses are not cheap.
Sure, Windows Desktop is clearly becoming "Whatever" by Microsoft but it's also pretty cheap. NT Kernel and UI work has to be done for server side and until that cash cow is dead, shoving slop into Windows Desktop is cheap revenue stream on work they have to do anyways.
ReactOS developers use Copilot to extract and copyright launder Windows source code, and then rather than fight it, Microsoft starts shipping ReactOS.
Microsoft is an enterprise, and enterprises will continue to crank out enterprisey stuff. Linux is free and open source, developed by people with passion - some of it, I assume, is out of necessity. Unless the working world dramatically changes over the next 15 years, Microsoft is still going to Microsoft.
Windows sucks, Azure sucks, Office sucks. Microsoft is a corporation designed to make money, they have a deadlock on the market. From an investor's point of view, they're doing just fine. From a shareholder's point of view, uprooting the entire Windows base to make tech people happy isn't worth the investment. Microsoft hasn't been about making tech people happy since it went public. Microsoft makes money and employs people. People half-heartedly go to work to earn a living, they produce enterprise-grade software. Enterprise software makes money. That's all the investor cares about.
Actually, as a matter of fact, having Windows around to drive the continued development of Linux might be a good thing. I know Windows sucks, I know virtually anything technical is dramatically easier on Linux, but anything without competition eventually stagnates. Even if Windows exists simply as a "What not to do" in Linux, it's probably good that it remains around.
Currently typing this on a machine that dual-boots both Windows and Linux. Why? Because my laptop came installed with it.
> As a professional programmer, I no longer consider Windows a viable option for serious work.
Please get over yourself. There's plenty of actually serious programming work being done on Windows.
> If you’re a programmer who’s used to Windows and you think I’m being overly harsh, I encourage you to spend a couple weeks in any other operating system.
For the record, I've spent decades in many other operating systems. It's interesting because the OS used to matter. Now 90% of the apps we use are either on the web or are web apps repackaged as desktop apps. Of course I can still tell when I switch between OSes, but it makes much, much, much less of a difference than it used to.
For example, this is my taskbar layout: https://i.ibb.co/1GqKH27L/taskbar-layout.png
To my knowledge, it's not possible to achieve anything like this layout on Windows 11, Linux or Mac. I did try it in various Linux distros a few times but frankly got sick of navigating the maze of window managers et cetera. I think something like XFCE came close to providing a Windows-like taskbar but it was still far, far behind what Windows NT can offer.
even if written in retrospect - this would've been interesting to see. since likely some of the reasons wouldn't include A.I
1: https://news.ycombinator.com/item?id=41923011