Tuesday, June 2nd 2026

MSI Shows New Cooling and Power Designs for Future GPUs at Computex 2026

MSI has showcased several new cooling and power designs and innovations for future graphics cards at the Computex 2026 show, including spiral-groove heatpipes, diamond-composite thermal pad and diamond-copper composite baseplate, fan with ultra-thin metal blades, as well as all new MSI SafeGuard feature that automatically detects faults in both current and voltage and triggers several alerts and limits the power drawn by the graphics card.

MSI is currently showcasing these features on two graphics cards, the RTX 5090 SUPRIM Safeguard, that uses both the new server-grade eFUSE that adds another layer of short-circuit protection with fast response of around 200 ns, and also being resettable and reusable thanks to its internal gate-based reset mechanism, as well as all the MSI Safeguard features and software, like the smart 12 V power connector detection, real-time current monitoring, and several warning alerts.
The second graphics card is the RTX 5090 GAMING TRIO Next-Gen, which incorporates those next-gen cooling features like the spiral-groove heatpipes, which are meant to increase internal contact surface area and improve heat transfer efficiency, diamond-copper baseplate and diamond-composite thermal pad, both of which should improve heat transfer, and the new ultra-thin metal fan blades, which should reduce deformation at high speeds, increase effective airflow area, all thanks to 0.8 mm metal fan blades with higher rigidity.
Of course, it is highly unlikely that MSI will launch these two graphics cards but it is interesting to see what to expect from future next-gen flagship graphics cards from MSI.
Add your own comment

8 Comments on MSI Shows New Cooling and Power Designs for Future GPUs at Computex 2026

#1
Chaitanya
Thermal Grizzly also had a product featuring diamonds on display at this years Computex.
Posted on Reply
#2
EnviousFrog
Oh good. MSI graphics cards now have a fuse. I wonder if that would satisfy Northwest Repair.
Posted on Reply
#3
Legacy-ZA
The correct way to solve this little problem is for the PSU to monitor the current and back the hell off when the card requests too much on a certain pin, and then perhaps, some sort of handshake from the card to let the PSU know to supply more on another pin that isn't being heavily utilized.

But how about this, the best one yet? Stop selling us solutions to problems YOU imbeciles created. Just give us the old connectors back.
Posted on Reply
#4
Chomiq
Still polishing the turd in 2026.
Posted on Reply
#5
Gucky
That should have been built into the cards from day one... The problem is known since the release of RTX4000.
Legacy-ZAThe correct way to solve this little problem is for the PSU to monitor the current and back the hell off when the card requests too much on a certain pin, and then perhaps, some sort of handshake from the card to let the PSU know to supply more on another pin that isn't being heavily utilized.

But how about this, the best one yet? Stop selling us solutions to problems YOU imbeciles created. Just give us the old connectors back.
Loadbalancing is the way. By limiting the power per pin to ~100W or per 2 pins to 200W. Loadbalacing has been used on 8-Pin cards for years to limit the power to ~150W per 8-Pin.
Posted on Reply
#6
Grimling
Legacy-ZAsome sort of handshake from the card to let the PSU know to supply more on another pin that isn't being heavily utilized.
If a Pin is not heavily utilized, that means it has a higher resistance = more HEAT.
Supply more Power to that Pin would increase the Problem, because then you really trying to make a Fire.
Posted on Reply
#7
Random_User
Legacy-ZAThe correct way to solve this little problem is for the PSU to monitor the current and back the hell off when the card requests too much on a certain pin, and then perhaps, some sort of handshake from the card to let the PSU know to supply more on another pin that isn't being heavily utilized.

But how about this, the best one yet? Stop selling us solutions to problems YOU imbeciles created. Just give us the old connectors back.
The idea is not bad. But it's just the "crutches" and "duck tape" workaround, instead of fixing the problem itself. That would make unnecessary burden and overhead for the PSU design complexity, and add more cost.

Someone just have to kick the madlad in the... and rectify the soddin' "standard". And it bonkers, how it actually went that far, to become the standard, to begin with...
Posted on Reply
#8
_roman_
a power supply unit can not fix schematic part savings on a graphic card. i would not mislead others by claiming a power supply unit can fix parts of a schematics which are missing from a graphic card.
Posted on Reply
Nov 11th, 2026 20:32 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts