megabigd
I wondering if i really need to install the pads. I have a 1080 sc. Installed a evga hybrid cooler. Is it something i still have to worry about? the card wont go past 40C. Hybrid kicks ass

• Remove the card from the case and unhook every power connection
• Unscrew all screws on the backside (on the backplate)
• Remove Backplate
• Unplug LED and Fan-Connector
• Remove Heat Spreader (the metal cage where the fans also sit on)
• Remove "middle plate" where the heat spreader sits on. You will see several thermal pads on the underside of this plate already
• Put the "optional" thermal pads in place. These are adhesive plastik pads which you can install like power stripes to attach a picture frame on a wall without nails or screws. The exact positioning of these pads is what is actually of interest for the more advanced geeks around here.
• Check the TMI (the gooey paste between heat spreader and chip die), most more advanced geeks will want to renew it or replace it with something more geeky to bring temps down further*
• Put everything back together, making sure the previously installed thermal pads are also back where they belong and are touching both sides (the plate AND the chip) and making sure that the connectors are all plugged in
• Reinstall the card, giving it a hug or maybe a kiss beforehand, just to make sure it knows you still love it, even if it is... specially enabled.
• Boot up your rig and forget about this issue.

Angier_1985
Every card has a risk of blowing up once any of the components under stress gives in, be it a manufacturing error, a case of inapropriate usage or simple wear out. In fact, any GPU has a limited life span as the heat emitted by the die, the VRAM and the VRM is enough to slowly degrade the durability of the PCB and at some point in the future (however far that may be) it could cause the PCB to break apart, making room for a sparking discharge where none should be possible.
 
Fact: The thermal pads are to lower the temperatures of the VRMs on the card
Fact: The cooling solution EVGA is using in the FTW-series is deemed appropriate even after several tests by the company
Fact: There is just one single review with measuring methods that reports values that are irritating. In this review, the card did not give out. The measuring methods have been critizised to be not accurate by both EVGA and individuals
Fact: As part of this review a test was used that puts the card under stress not specified for (Furmark)
Fact: Every other report about this is using the data from said review to discuss this possible design flaw
Fact: There is not a single report publicly available that can clearly prove a failure of the VRM because of overheat
Fact: EVGA stated themselves that there is no overheat issue affecting cards that have been RMA that can be linked to the missing thermal pads
Fact: There has been a bad batch of cards with faulty VRMs. These are prone to failure even under light stress.
Speculation: As it is reported to be around 3-4% we are looking at a batch size of possibly several hundred cards. 
Fact: It is unknown to the public if the blackscreen & 100% fan issue and the combustion issue both stems from these faults.
Speculation: The bs + 100% issue and the combustion is caused by cards with the faulty VRMs. Either because the card is detecting the error and tries to prevent it or the faulty VRMs simply combust.
 
FINAL Fact: There have always been cards that gave out like this, in every generation and with every brand. People are confused and fear to be affected themselves because they read about a possible design flaw and combusting VRM and link these issues together even tho there is no proof of a relation. The fact that the 9xx FTW series with the ACX 2.0 cooler had also issues with heat makes people instinctively relate these. Understandable but it's simply a misconclusion.