PC Powers Off During Startup: Troubleshooting Guide

A PC that briefly starts and then loses power is not automatically the same problem as a PC that stays powered but never completes POST. Watch what the machine does from the moment the power button is pressed: whether fans and lighting start, whether motherboard diagnostic indicators advance, whether a firmware logo ever appears, and whether the system becomes completely unpowered or immediately begins another startup attempt.

If the machine reliably reaches BIOS/UEFI or begins loading an operating system before it restarts, move to the random-restart or boot-failure workflow instead. This guide owns the earlier failure: power disappears or startup repeatedly collapses while the platform is still trying to initialize hardware. That boundary prevents Windows repair steps from being mixed into a pre-POST electrical or hardware problem.

Modern PC power supplies can include protection for abnormal voltage, current, temperature, short-circuit, and total-power conditions. Corsair’s current PSU documentation, for example, explicitly describes OVP, UVP, OCP, OTP, SCP, and OPP shutdown behavior. A system that loses power can therefore be consistent with a protection circuit responding to an abnormal condition elsewhere in the system as well as with a PSU fault.

Do not infer which protection fired from the symptom alone, and do not repeatedly force the machine on while an obvious short, damaged connector, burning smell, smoke, or abnormal heat is present. Never open a PSU or probe mains circuitry. Protection implementation and thresholds vary by PSU, so use the exact unit documentation where it exposes diagnostic behavior rather than assigning a universal shutdown threshold.

Check the wall or power-strip connection, the PSU rear switch, the motherboard 24-pin connection, and the CPU/EPS power connection required by the exact motherboard. ASUS current no-power/no-boot guidance treats AC input, the PSU switch, ATX power connections, and a known-good PSU test as separate checks. Fan or RGB activity only proves that some power reached some circuits; it does not prove that every required rail and connector is correct under startup load.

If the PSU was recently replaced, use only modular cables that the PSU manufacturer explicitly lists as compatible with that exact model or cable family. PSU-side modular pinouts are not universal. Reseat connections only with AC power removed and follow the component instructions; do not substitute improvised adapters, rewired cables, or a connector merely because it physically fits.

A new build or recent motherboard, CPU, cooler, PSU, GPU, memory, or case change gives you valuable evidence. Recheck the exact connectors and mounting hardware disturbed by that work before expanding the diagnosis. For a motherboard installation, inspect for misplaced case standoffs, loose conductive hardware, trapped or damaged cables, and connectors that are offset or incompletely seated. Do not run the board on conductive packaging or improvise insulation.

Front-panel wiring is another changed variable on a new case or motherboard. Confirm the power-switch lead against the exact motherboard front-panel header diagram. A damaged or mechanically stuck case switch can be isolated by disconnecting the case switch and using only the motherboard manufacturer’s documented power-on method where one exists. Never short arbitrary header pins from a generic diagram.

If shutdown began immediately after CPU or cooler installation, check the CPU/socket area, cooler mounting hardware, fan or pump connections required by the cooler, and CPU power. Use the exact socket and cooler instructions. An obvious mounting mistake or disconnected cooling component deserves correction, but a shutdown seconds after power-on does not establish a universal thermal diagnosis.

Do not invent a single CPU temperature or number of seconds at which every platform must shut down. Thermal limits and firmware behavior are product-specific, and a pre-POST power loss can have several unrelated causes. If the platform stays powered long enough to expose trustworthy firmware temperature data, that becomes useful evidence; if it does not, keep cooling installation as one branch rather than declaring it the cause.

If the motherboard remains powered long enough to show CPU, DRAM, VGA, BOOT, POST-code, or beep diagnostics, read the exact motherboard manual. Vendor indicators identify the startup stage or subsystem that needs investigation; they are not universal component-failure codes. A DRAM indication, for example, can lead to module seating, population, compatibility, CPU/socket, or firmware checks rather than automatically proving a dead DIMM.

For memory isolation, return to documented defaults and test a supported baseline population, commonly one module in the manufacturer-specified diagnostic or preferred slot. DDR5 training can also cause unusually long blank-screen startup after configuration changes. Do not repeatedly cut power merely because training takes longer than an ordinary boot, and do not apply one platform’s expected training time to every motherboard.

A CMOS reset can be useful after unstable memory profiles, overclocking, undervolting, or other firmware changes. It restores firmware configuration according to the motherboard’s supported procedure; it does not update BIOS, add CPU support, repair a short circuit, or fix a physically incorrect installation. Use the exact board manual because reset buttons, headers, battery procedures, and power-down requirements vary.

If a CPU change preceded the problem, separately verify the motherboard CPU-support list and the minimum BIOS revision for that processor. A physically compatible socket does not guarantee that the installed firmware can initialize the CPU. Use a vendor-supported firmware update or recovery method only when the exact motherboard provides it.

When the cause remains unclear, remove nonessential variables: storage devices, USB accessories, extra memory modules, add-in cards, RGB/controller hardware, and other peripherals that are not required to reach firmware. Keep the motherboard, supported CPU, correctly installed required CPU cooling, PSU, one supported memory module, and a valid display path. A discrete GPU is required when the processor/platform cannot provide integrated display output.

If the minimum configuration remains powered and reaches firmware, reconnect one category at a time until the failure returns. If it still powers off, the smaller configuration makes cross-testing a known-good compatible PSU, memory module, display path, or other documented branch more meaningful. Change one variable at a time so each result actually narrows the fault.

Once BIOS/UEFI is reliably usable, the original pre-POST power-off condition is no longer occurring in that test state. If shutdown appears only after reconnecting a particular device, under operating-system load, or after Windows begins loading, follow that evidence into the relevant PSU/load, temperature, driver, boot, or random-restart workflow instead of continuing to reseat unrelated POST hardware.

The useful order is observational: establish the failure boundary, inspect the recent change, verify required power, check documented diagnostics, return memory and firmware to a supported baseline, reduce to minimum hardware, then restore variables one at a time. There is no responsible universal replacement order. A shutdown is a symptom; the goal is to identify the condition that makes it repeat.