Let’s play a game. I’m going to describe two computers, and you tell me which one you’d rather own.
Computer A:
- Intel Core i7 (4th generation)
- 64GB of RAM
- NVIDIA RTX 3070 GPU
- No TPM 2.0 module
Computer B:
- Some entry-level CPU
- 8GB of RAM
- No dedicated GPU (integrated graphics only)
- Has TPM 2.0
If you picked Computer A, congratulations — you have working eyes and a functioning brain. Computer A would demolish Computer B in literally every real-world task: video editing, gaming, running a dozen Chrome tabs without weeping, machine learning experiments, compiling code, you name it. 64GB of RAM alone puts it in “small workstation” territory. The RTX 3070 is still, years after release, a genuinely capable GPU that can run modern games at 1440p and handle CUDA workloads with ease.
And yet, according to Microsoft, Computer B is the “supported” machine and Computer A is not. Computer A can’t run Windows 11. Officially, it’s obsolete.
Wait, what?
This is the absurdity at the center of Windows 11’s system requirements. Microsoft’s minimum spec isn’t really about raw performance — it’s gated behind a security chip called TPM (Trusted Platform Module) 2.0, plus a CPU compatibility list that happens to exclude basically all 4th-gen Intel chips (Haswell, for the curious), even high-end ones like the i7-4790K that were flagship gaming and workstation processors in their day.
So a $150 office PC with a TPM 2.0 chip soldered to the motherboard passes Microsoft’s compatibility check with flying colors. A workstation with 64GB of RAM and a $500 GPU does not, because it happens to be one CPU generation too old and lacks a specific security chip — despite the fact that you can often buy a TPM 2.0 module for $15-20 and physically install it on many of these older boards, or in some cases just enable a firmware-based TPM (fTPM) that was sitting there disabled the whole time.
To be fair to Microsoft, TPM 2.0 isn’t nothing — it enables things like BitLocker encryption, Windows Hello, and better protection against firmware-level attacks. Security requirements aren’t inherently unreasonable. But the implementation here is a blunt instrument: it throws out machines based on a checkbox rather than actual capability, and it does so at a moment when e-waste is a legitimate global crisis. Perfectly good silicon — CPUs, RAM, GPUs that took real energy and real mining and real manufacturing to produce — gets marked “unsupported” and nudged toward a landfill, while a computer with a sixth of the RAM and no GPU at all gets the green light.
The absurd part, quantified
Let’s be blunt about what “unsupported” actually costs you if you follow Microsoft’s script:
- You retire a machine with 64GB RAM and an RTX 3070 (a GPU that still sells used for $250-350 today).
- You replace it with a machine that has 8GB RAM and no GPU.
- Your new “supported” computer will struggle to keep more than a browser and a spreadsheet open at once, let alone do anything the old one could do in its sleep.
- You did this because of a chip that costs less than a fast-food meal, or a CPU generation cutoff that has nothing to do with whether the machine can actually run Windows 11 (it can — Microsoft even lets you bypass the check with a registry edit or bootable installer for now).
This isn’t a security decision made in your interest. It’s obsolescence-by-policy, dressed up in a security justification, that happens to align nicely with new PC sales.
Here’s the good news: that “obsolete” machine is a beast under Linux
This is where the story gets fun instead of infuriating. That 4th-gen i7 with 64GB of RAM and an RTX 3070 doesn’t care one bit what Microsoft thinks of it. Install a modern Linux distribution and it comes roaring back to life as a genuinely high-performance machine — because Linux doesn’t have artificial CPU generation gatekeeping or a TPM 2.0 requirement at all.
A few reasons this pairing works so well:
The CPU is still fine. A 4th-gen i7 has four cores and eight threads, and for the vast majority of desktop computing — browsing, office work, coding, media, even a lot of gaming — that’s still perfectly adequate, especially when it’s not fighting Windows 11’s background service overhead.
64GB of RAM is enormous headroom. On Linux, that’s enough to run a dozen browser tabs, a full IDE, several VMs, and a Docker container stack simultaneously without ever touching swap. It’s the kind of RAM budget that lets you do real development or data work.
The RTX 3070 works great with NVIDIA’s Linux drivers. NVIDIA’s proprietary driver support on Linux is mature at this point. You’ll get full CUDA support for machine learning workloads, solid gaming performance via Proton/Steam Play (a huge percentage of the Steam library runs great, often within a few percent of native Windows performance), and full acceleration for video editing and 3D work.
No TPM required, no artificial cutoffs. Distributions like Ubuntu, Fedora, Linux Mint, or Pop!_OS (which has particularly good out-of-box NVIDIA support) will install and run without a single complaint about your “unsupported” hardware. It just works.
Practically speaking, this machine on Linux Mint or Fedora with the NVIDIA proprietary drivers is a faster, more capable daily driver than the “compliant” 8GB no-GPU box will ever be on Windows 11 — full stop. You’re not settling for a downgrade by switching; you’re unlocking hardware that was being held hostage by a compatibility checklist.
The takeaway
Windows 11’s hardware requirements conflate “security minimums” with “arbitrary generational cutoffs,” and the result is that genuinely powerful machines get labeled obsolete while genuinely weak machines get a stamp of approval. If you’ve got a machine like Computer A sitting around because Windows told you it’s not good enough — it’s not the machine that’s the problem. Install Linux, keep the RAM and the GPU doing real work, and let the “supported” 8GB potato do what it’s good at: reminding you why the label was never about performance in the first place.
