Your new graphics card is installed, the fans may spin up, the RGB lighting glows — but the monitor shows nothing. No BIOS splash, no manufacturer logo, just a black screen or a “No Signal” message. Sometimes the display works fine with the old card but goes dark the moment the new GPU is seated, or the screen goes black at the exact moment Windows should load. Other times the system beeps in a pattern you have never heard before, or the debug LEDs on the motherboard stop at the white VGA light.
This is one of the most common GPU installation problems, and the good news is that it is rarely fatal. The overwhelming majority of no-display cases are caused by cabling, power delivery, or firmware mismatches rather than a dead card — which means most of them can be resolved in under an hour at your desk. Work through the fixes below in order, from the quickest checks to the more involved ones, and stop as soon as your display returns. Do power the system down fully before each fix so you are not hot-swapping components.
Likely Causes
| Cause | How to confirm | Fix difficulty |
|---|---|---|
| Monitor cable plugged into the motherboard instead of the GPU | Look at the rear I/O panel — the cable is in the video ports near your USB ports rather than lower down on the card | Very easy |
| Fully or partially unseated card | Check whether the PCIe slot’s retention latch clicked and whether the bracket screws are flush | Easy |
| Missing or loose PCIe power connectors | Admire the top edge of the card — unused 8-pin/16-pin sockets, or connectors seated at an angle | Easy |
| Power supply too weak or using daisy-chained cables | Add up your PSU wattage versus the card’s recommended PSU rating; check whether one cable feeds two 8-pin plugs | Moderate |
| BIOS set to legacy/CSM mode or primary display forced to iGPU | Swap in the old card (or use the motherboard’s video output) and check BIOS settings for CSM and Primary Display | Moderate |
| UEFI firmware or driver incompatibility | Card works in another PC but not yours, or display cuts out exactly when the Windows logo appears | Moderate |
Fix 1: Plug the Monitor Into the Graphics Card
What causes it
Almost every CPU has integrated graphics, and motherboards ship with those outputs live. If your monitor cable stayed in the same port it occupied before the upgrade, it is feeding from the motherboard’s video chipset, which has no idea a discrete card exists.
How to check
Trace your DisplayPort or HDMI cable from the monitor. If it terminates in the cluster of ports near your USB and Ethernet jacks — the motherboard I/O — that is the problem. A card-driven signal only ever comes from the ports on the card’s own bracket, lower down the case.
What to do
Power off the PC, move the cable to the DisplayPort or HDMI port on the graphics card itself, and boot again. If you are running a high-refresh monitor, prefer DisplayPort, since some cards limit HDMI to 4K at 60 Hz on certain outputs.
How to undo it
Nothing to undo — just move the cable back if you ever remove the card.
Fix 2: Reseat the Card and Check the Retention Latch
What causes it
A card that looks installed can still be sitting a few millimeters out of the slot. Shipping jostle, a tight case, or an unopened PCIe latch are typical culprits. Even partial seating breaks the gold-edge contact pattern the slot needs.
How to check
Power down and unplug the PSU. Look at the rear bracket: it should sit flush against the case with both screws tightened. Then press gently on the top of the card’s cooler — if there is any visible flex or movement, it is not fully home. Check that the plastic retention latch at the rear end of the PCIe slot has clicked over the card’s notch.
What to do
Release the latch (on many boards you press a small plastic tab or a push-button at the slot’s end), pull the card out, and reinsert it with firm, even pressure until the latch snaps audibly. Screw the bracket down before powering on.
How to undo it
There is nothing to undo. Reseating is reversible by design — just release the latch before pulling the card again so you do not crack the slot.
Fix 3: Connect All PCIe Power Cables Properly
What causes it
Modern cards draw 150 to 450 watts or more, delivered through one to four 8-pin PCIe connectors or a single 12V-2×6 (12VHPWR) plug. A missing connector, a partially seated plug, or a single PSU cable daisy-chaining two high-draw plugs can trip the card’s safety logic before it ever outputs a signal.
How to check
With the system off, inspect every power connector on the card’s top edge. Each socket must be filled on cards that require it. Tug gently on each plug — it should not lift. If you are using a 12V-2×6 connector, look at the plug from the side: it must sit fully flush, with no visible gap between the connector body and the card’s shroud.
What to do
Run a separate PCIe cable from the PSU to each 8-pin socket on the card, rather than using the pigtail ends of one cable for two sockets. For 12V-2×6 cards, use the native cable that came with your PSU if it has one; otherwise use the adapter included with the card, connecting two or three separate 8-pin cables as the adapter’s design requires. Confirm the PSU’s wattage meets or exceeds the card manufacturer’s recommended figure, printed on the box and spec sheet.
How to undo it
Unplug the added cables if you ever revert to the old card, and store the adapter in the card’s box.
Fix 4: Reseat or Swap the Power Cables at the PSU End
What causes it
Modular PSUs use removable cables, and it is remarkably easy to leave the PCIe cable loose at the power supply side, or to plug a cable into a socket labeled for CPU output instead of PCIe. The 8-pin EPS CPU cable and the 8-pin PCIe cable are physically similar and cause endless confusion.
How to check
With the PC unplugged, follow each cable from the graphics card back to the PSU. Confirm it enters a socket marked PCIe, VGA, or PCI-E — never CPU or EPS. Push each connector in until it clicks, and check that no pins are bent or backed out inside the housing.
What to do
Unplug and firmly reseat each cable at the PSU end. If you find a CPU cable in a PCIe socket or vice versa, swap it for the correct one — they are keyed differently in most designs, but older or third-party cables may fit incorrectly. Never force a connector that resists.
How to undo it
Simply disconnect the cables if you remove the card later. Keep CPU and PCIe cables labeled or in separate bags so the mix-up does not repeat.
Fix 5: Clear the CMOS and Reset BIOS Display Settings
What causes it
If your previous setup used integrated graphics or an older card, the BIOS may still be configured to initialize the iGPU first, or to use legacy CSM (Compatibility Support Module) boot, which some modern cards do not support. A stale CMOS can also hang the POST sequence entirely.
How to check
If you can get a display using your old card or the motherboard’s video output, enter the BIOS (usually by pressing Delete or F2 during boot) and look for settings named “Initiate Graphic Adapter,” “Primary Display,” or “iGPU Multi-Monitor” under the Advanced or Boot menus, and a “CSM” or “Launch CSM” option under the Boot menu. If you cannot get a display at all, a CMOS reset is still safe to try.
What to do
Power off, unplug the PSU, and press the power button for 10 seconds to drain residual charge. Then either press the CLR_CMOS button on the rear I/O or motherboard PCB (if present) for a few seconds, or remove the coin-cell battery from the motherboard for five minutes and reinstall it. On boards without either option, there is often a two-pin CLRTC jumper — short it with a screwdriver tip for 10 seconds with the PSU unplugged. Set “Primary Display” to PCI-E or Auto, and disable CSM so the card initializes in pure UEFI mode. Save and exit.
How to undo it
A CMOS reset returns every BIOS setting to factory defaults, so if you had custom fan curves, XMP/EXPO memory profiles, or boot-order overrides, re-enter the BIOS and reapply them.
Fix 6: Test With Minimal Hardware and One Stick of RAM
What causes it
Sometimes the card is fine but the POST process stalls before it ever initializes video, usually because of a marginal RAM stick, a loose 24-pin motherboard cable, or a short caused by a standoff behind the motherboard tray. Because POST failures often look identical to GPU failures, you need to isolate.
How to check
Watch the motherboard debug LEDs (CPU, DRAM, VGA, BOOT) or listen for beep codes if a system speaker is fitted. A DRAM LED that stays lit points at memory, while a persistent VGA LED suggests the card is not being detected. If the screen shows BIOS text or a logo before going black in Windows, the card is detected and your problem is driver-side rather than hardware.
What to do
Unplug the PSU, then remove all but one RAM stick (use the slot your motherboard manual identifies as A2 or the second slot from the CPU). Reseat the 24-pin ATX cable and the 8-pin EPS CPU cable at the motherboard. Verify no unused standoff sits under the motherboard. Boot with just the card, one stick, and the boot drive connected. If display appears, add components back one at a time until the failure returns — that component is your culprit.
How to undo it
Reinstall your remaining RAM sticks and reconnect any peripherals you removed, powering down between each addition.
Fix 7: Boot With Integrated Graphics and Install Drivers First
What causes it
On a fresh Windows install or a system jumping from a very old card, the display can cut out at the login screen because the driver stack loaded at boot conflicts with the new hardware. NVIDIA and AMD both recommend clean driver states during upgrades.
How to check
If your CPU has integrated graphics, connect the monitor to the motherboard’s video output and boot. If Windows loads normally and Device Manager (right-click Start, choose Device Manager) shows the new card under “Display adapters” with a yellow warning triangle, the hardware is detected but the driver is wrong or missing.
What to do
While on integrated graphics, download the latest driver package from the card vendor’s site, then run Display Driver Uninstaller (DDU) in Safe Mode (Shift-click Restart from the Start menu power icon, then Troubleshoot, Advanced Options, Startup Settings, Restart, then select Safe Mode). DDU’s “Clean and restart” option wipes old GPU drivers. Reboot, let Windows finish loading, then install the fresh driver package you downloaded. Power down, move the monitor cable back to the card, and boot normally.
How to undo it
DDU changes are permanent by design, but you can roll back a bad driver in Device Manager: right-click the GPU under Display adapters, choose Properties, then the Driver tab, and click Roll Back Driver if available.
Fix 8: Update the Motherboard BIOS
What causes it
Newer graphics cards occasionally need a motherboard UEFI update to initialize correctly, especially on platforms launched a year or two before the card. Symptoms include the VGA debug LED staying lit and no POST at all, while the same card works perfectly in a newer system.
How to check
With any working display path (old card, iGPU, or another PC), open System Information (type msinfo32 into the Start menu search) and note the “BIOS Version/Date” entry. Compare it to the latest version listed for your exact motherboard model on the manufacturer’s support page. Cards that ship today in 2026 commonly ask for UEFI revisions from the previous year or later.
What to do
Download the latest UEFI file from your board maker’s support page, extract it to a FAT32-formatted USB stick, and use the motherboard’s built-in flash utility — usually called Q-Flash, EZ Flash, M-Flash, or BIOS Flashback — accessible from the BIOS menu or, on boards with a Flashback button, without a CPU or GPU installed at all. Follow the on-screen prompt, do not interrupt power during the update, and reboot when it finishes.
How to undo it
Most boards let you flash an older UEFI version the same way, but there is rarely a reason to. Leave the update in place.
If Nothing Works
At this point the odds shift toward a hardware fault. Test the card in another system if at all possible — a working display in a friend’s PC confirms your motherboard or PSU is the problem, while the same black screen elsewhere means the card itself is defective. If the card is new, start an RMA with the retailer or manufacturer promptly, since return windows are typically 14 to 30 days and you want the clock on your side. If your own system is at fault, test the PSU with a replacement of adequate wattage, and contact your motherboard manufacturer’s support with your exact model and BIOS version. Document what you have tried — the elimination steps above are genuinely useful information for a support technician, and mentioning that you cleared CMOS, reseated power, and tested minimal hardware will speed up the exchange considerably.
FAQ
Why do the GPU fans spin but the screen stays black?
Fans spinning only proves the card is receiving standby and main power — it does not confirm video output. The card can power on while failing to initialize its display engine, usually because of a seating, power cable, or BIOS initialization issue. Work through the fixes above starting with Fix 1.
Can a weak power supply cause no display at all, or just crashes?
Both. An underpowered or failing PSU can prevent the card from completing its power-up sequence entirely, which looks identical to a dead card. Check that your PSU’s wattage meets the card manufacturer’s stated recommendation and that all connectors are native, undamaged, and fully seated at both ends.
Should I disable integrated graphics in the BIOS?
Usually not necessary. Leaving the iGPU enabled gives you a fallback display path if the card ever fails, which is exactly what you want during troubleshooting. Only disable it if your board’s documentation or the card vendor’s instructions specifically ask, or if you are experiencing resource conflicts.
How long should I troubleshoot before assuming the card is dead?
If you have worked through reseating, power connections, CMOS reset, minimal hardware, and a clean driver install without any POST display — and ideally tested the card in a second system — it is reasonable to conclude the card or another component is faulty. Start the RMA process rather than continuing to disassemble the system.
