ERROR_REG_NAT_CONSUMPTION 0x000004ED Fix (BSOD)

Quick answer: Reset BIOS to optimized defaults, kill any overclock or XMP/EXPO profile, then check for faulty RAM and a bad driver — 0x000004ED almost never means "Windows is broken."

ERROR_REG_NAT_CONSUMPTION is a Windows bugcheck with code 0x000004ED. It fires when the kernel's native register state gets into a condition it can't safely consume — think of it as a corrupted CPU context passed from ring 0 to a driver and back. On modern Intel and AMD chips this is almost always hardware instability or a driver that's playing with native state it shouldn't. You'll see it most on machines pushed past spec — Ryzen 7000 with EXPO on, Intel 13th/14th gen with a "free" motherboard overclock, or X299/X399 boxes running four DIMMs at high speed. Kernel anti-cheat drivers (vgk.sys, EasyAntiCheat.sys, BEDaisy.sys), some VPN drivers, and older virtual machine hypervisors are repeat offenders. I've also seen it from a bad WHEA event that didn't recover.

If you've just changed RAM, added a second GPU, updated BIOS, or enabled a new overclock — that's your culprit. Revert it before doing anything else.

Step-by-step fix

Get the bugcheck parameters. Pull the minidump so you know which module crashed. Open an admin Command Prompt and run:
```
wevtutil qe System /q:"*[System[(EventID=1001)]]" /f:text /c:5
```
Or open C:\Windows\Minidump and load the newest .dmp in WinDbg. Run !analyze -v. Look at MODULE_NAME and IMAGE_NAME — that's the driver or module to blame.
Reset BIOS to defaults. Reboot into BIOS (Del or F2 on most boards). Load Setup Defaults. Save and exit. On AMD boards, disable EXPO/DOCP. On Intel 13th/14th gen, make sure you're on a BIOS with the 0x12B microcode — earlier revisions let the CPU cook itself and 0x4ED is one of the ways it manifests.
Run MemTest86. Not the Windows memory diagnostic — the real thing, bootable from USB. Let it run at least two full passes. Single-bit errors at high clocks are the #1 cause of 0x4ED I've seen in the field. If it throws errors at stock JEDEC speeds, RMA the sticks.
Check WHEA errors. Open Event Viewer, Windows Logs, System. Filter for source WHEA-Logger. Corrected machine-check errors, cache hierarchy errors, or a bus/interconnect error mean the CPU itself is unhappy — usually voltage or thermals. Note the APIC ID and bank.
Update chipset and GPU drivers. Intel Chipset Device Software, AMD Chipset Drivers, and the latest Adrenalin/Game Ready package from the vendor's site — not Windows Update. Windows Update ships months-old chipset drivers that misbehave with new silicon.
Test with Driver Verifier. Only if the dump doesn't point to anything obvious. From admin CMD:
```
verifier /standard /all
verifier /querysettings
```
Reboot. If you get a bugcheck on the next boot, the dump will name the offending driver. Turn it off with verifier /reset once you've found it. Do not leave it on — it slows the machine to a crawl.
Pull third-party kernel drivers. VPN clients (older NordVPN TAP driver, some Cisco AnyConnect builds), motherboard RGB software (Asus Aura, Gigabyte RGB Fusion), and third-party anti-cheat are common. Uninstall one at a time, reboot, see if it stops.

If that doesn't work

Drop to JEDEC speeds on RAM. Not "XMP off" — actually set DDR5-4800 or DDR4-2133 in BIOS. If the crash goes away, your memory controller can't handle four sticks at rated speed. Run two DIMMs in A2/B2 instead of four.
Revert to an older BIOS. Not every update is an improvement. If 0x4ED started right after flashing, go back one revision.
Swap the CPU if you can. On a 14900K that's been running with no power limits for a year, the chip may be permanently degraded. Intel's extended warranty covers this — RMA it.
Disable C-States and C1E as a band-aid. It masks an unstable chip, but if the machine's out of warranty and you need it running, it works. You'll burn more idle power.
Boot into Safe Mode with bcdedit /set {current} safeboot minimal. If 0x4ED stops in Safe Mode, it's a driver. If it still bluescreens, it's hardware.

Prevention

Stop treating "stable for two hours in Cinebench" as proof your overclock is fine. 0x4ED shows up days later under different loads. Run OCCT's CPU test for an hour, y-cruncher for another, and Prime95 Small FFTs for two. If any of those throw errors or the machine reboots, back the voltage or clock down one notch. Also: buy matched RAM kits, keep BIOS current on 13th/14th gen Intel, and don't install five different vendor utilities that all want kernel-level access. Every ring-0 driver is a chance to corrupt native state. Fewer is better.