You’ve already covered most of the big factors, but since it’s still overheating even after water cooling, it’s worth going through a few less obvious checks that often get missed:
Thermal Pads – On many GPUs, memory chips and VRMs use thermal pads (not paste). If these have degraded or shifted slightly when you installed the cooler, they can cause localized overheating even if the GPU core looks fine.
Mounting Pressure / Contact – Sometimes the cooler isn’t making perfect contact with the GPU die. Too much or too little mounting pressure can both cause poor heat transfer — check for even thermal paste spread when you remount.
Pump / Radiator Flow – Ensure the water pump is actually circulating properly and not airlocked. If your radiator fans are blowing in the wrong direction or the loop has trapped air, the heat won’t dissipate effectively.
VRM Temps – Many monitoring tools (like HWInfo64) can show separate VRM temperatures. These often overheat before the GPU core and can throttle performance.
Case Pressure Balance – Positive air pressure (slightly more intake than exhaust) helps keep dust out but can also trap hot air if the exhaust path isn’t direct.
Driver / Firmware Bugs – Some cards (especially newer NVIDIA or AMD models) can run hotter due to aggressive boost behavior. Try limiting the power target to 90% in MSI Afterburner — that often reduces temps by 5–10°C with minimal performance loss.
If you can share your GPU model, case type, and load temps (idle / gaming / stress test), it’ll be easier to pinpoint.