RTX 5090/5090D Bricked Issues: What Should I Do?
Detailed Answer
The RTX 5090 launch brought incredible performance — and unfortunately, some significant teething issues. Reports of bricked cards, melted connectors, and random crashes have affected a subset of users. Here’s what’s happening and how to troubleshoot.
Understanding the 5090/5090D Difference
The RTX 5090 is NVIDIA’s flagship consumer GPU released globally. The RTX 5090D is a China-specific variant with slightly reduced specifications to comply with export restrictions. Both share similar architectures and, unfortunately, similar early issues.
RTX 5090 Specs:
- 21,760 CUDA Cores
- 32GB GDDR7
- 575W TDP
- 12V-2×6 connector (600W capable)
RTX 5090D Specs:
- ~20,000 CUDA Cores (reduced)
- 32GB GDDR7
- ~500W TDP (reduced)
- Same 12V-2×6 connector
Common Issues and Causes
1. Power Connector Problems (Most Common)
The 12V-2×6 connector (replacing 12VHPWR) carries up to 600W through a compact connection. Issues include:
Melting/Burning Connectors: Improperly seated connectors create resistance, generating heat that melts plastic and damages pins. This is the #1 cause of “bricked” 5090s.
Signs of connector damage:
- Discoloration around power pins
- Burnt plastic smell
- Visible melting or deformation
- Intermittent power delivery
Prevention:
- Insert connector fully until both clips engage
- Ensure cable runs straight without bending at the connector
- Use high-quality PSU cables (avoid cheap adapters)
- Check the connection periodically
2. Driver Crashes and Black Screens
Some 5090 users experience:
- Black screens during gaming
- Driver timeout errors (TDR)
- System freezes requiring hard reboot
- Crashes after driver installation
Common causes:
- Early driver bugs (NVIDIA releases frequent hotfixes)
- Inadequate power delivery from PSU
- Motherboard BIOS incompatibility
- Overheating (thermal throttling misinterpreted as crashes)
3. Complete Failure (True Bricking)
A truly bricked card shows:
- No fan spin on power-up
- No display output whatsoever
- No detection in BIOS/UEFI
- No response to any input
This typically indicates hardware failure — either manufacturing defect or damage from power issues.
Step-by-Step Troubleshooting
Step 1: Power Connector Inspection
Shut down completely and unplug from wall. Remove the GPU and carefully inspect:
- Are all pins straight and undamaged?
- Any signs of melting, discoloration, or burning?
- Is the connector from your PSU in good condition?
If any damage is visible, do not reinstall. Proceed to RMA.
Step 2: Reseat the GPU
Remove and reinstall the graphics card:
- Clean PCIe slot contacts with isopropyl alcohol
- Ensure the card is fully seated (click sound)
- Reconnect power cable fully (both clips must engage)
- Verify all auxiliary power connections
Step 3: Test Power Delivery
The RTX 5090 requires a quality 850W+ PSU (1000W recommended). Verify:
- PSU meets 80+ Gold or better efficiency
- Using native 12V-2×6 cable (not adapter)
- No other components are overloading the PSU
Corsair RM1000x (2024) PSU
1000W | 80+ Gold | Fully Modular | Native 12V-2x6 | 10-Year Warranty
Reliable PSU with native 12V-2×6 connector for RTX 40/50 series cards. No adapters needed. Corsair’s proven reliability with excellent voltage regulation.
Step 4: Clean Driver Installation
Use Display Driver Uninstaller (DDU) in Safe Mode:
- Download latest NVIDIA drivers and DDU
- Boot to Safe Mode
- Run DDU, select “Clean and restart”
- Install fresh drivers after reboot
- Don’t use GeForce Experience initially (isolate variables)
Step 5: BIOS/UEFI Updates
Check your motherboard manufacturer’s website for:
- BIOS updates mentioning RTX 50 series compatibility
- PCIe settings (try Gen4 instead of Auto)
- Resizable BAR / Above 4G Decoding settings
Step 6: Test in Another System
If possible, test the GPU in a different PC. This isolates whether the issue is:
- The GPU itself (fails in both systems)
- Your specific system (works elsewhere)
When to RMA
RMA immediately if:
- Visible connector/PCB damage
- Burning smell from the card
- Complete failure (no signs of life)
- Crashes persist after all troubleshooting
Contact your retailer first if purchased recently. They often have faster replacement processes than manufacturer RMA.
NVIDIA/AIB Partner RMA: Keep all documentation, photos of any damage, and purchase receipts. Both NVIDIA and AIB partners (ASUS, MSI, EVGA, Gigabyte, etc.) have acknowledged early production issues and are generally honoring warranties without difficulty.
Known Issues and Patches
Driver Updates: NVIDIA has released multiple hotfix drivers addressing:
- Power state transition crashes
- Display timeout issues
- Memory overclocking instability
- Specific game crashes
Always run the latest drivers — many early issues have been resolved through software updates.
Firmware Updates: Some AIB cards have received firmware updates for:
- Fan curve optimization
- Power delivery improvements
- Thermal sensor calibration
Check your card manufacturer’s website for available updates.
Prevention for New Buyers
If you’re buying an RTX 5090 now:
- Buy from reputable retailers with good return policies
- Inspect the power connector before installation
- Use a quality 1000W+ PSU with native 12V-2×6
- Don’t bend the power cable at the connector
- Monitor temperatures during initial use
- Keep all packaging in case RMA is needed
Frequently Asked Questions
It’s improved but not eliminated. NVIDIA redesigned the 12V-2×6 connector with better tolerances than the 12VHPWR used on 4090. However, the fundamental challenge — 600W through a compact connector — remains. Proper installation technique is still critical.
We strongly recommend against it. Adapter cables (multiple 8-pin to 12V-2×6) add connection points that can fail. If your PSU lacks native 12V-2×6, consider upgrading the PSU rather than using adapters. The potential cost of a damaged GPU far exceeds a PSU upgrade.
No, this sounds like a driver or game-specific issue. True bricking means complete failure. Game-specific crashes often resolve with driver updates or game patches. Check NVIDIA’s known issues list and try the latest hotfix drivers.
Typically 2-4 weeks depending on stock availability and your location. AIB partner RMAs vary — EVGA (when active) was fastest, while others may take longer. Buying from retailers like Amazon or Micro Center often means faster replacements through their return process.
Consider it if offered at reasonable cost. The 5090’s complexity (power delivery, GDDR7, advanced cooling) means more potential failure points than simpler cards. Most manufacturer warranties are 3 years; extended warranties provide additional peace of mind for a high-value component.
The 5090 runs hot by design — up to 80-85°C is normal under load. However, thermal throttling (90°C+) indicates inadequate case cooling. Ensure your case has good airflow, consider adding fans, and verify the GPU’s fans are operating correctly. For storage upgrades that can work with your high-performance build, see our NVMe SSD guide
Related Articles
- NVMe SSDs — Fast storage for gaming builds
- Best SSD for Gaming — Complete gaming storage guide
- SSD vs HDD — Storage comparison
- Samsung 990 Pro — Popular gaming SSD
- 2TB Drives — Capacity options for game libraries
Summary
A “bricked” RTX 5090 is often recoverable through proper troubleshooting — power connector issues and driver problems are the most common culprits. Truly dead cards (hardware failure) should be RMA’d promptly; NVIDIA and AIB partners are handling warranty claims for affected units. If you’re experiencing issues, work through the troubleshooting steps methodically before assuming the worst. Many reported “bricks” turn out to be fixable software or connection issues.