Business

Microsoft unveils advanced AI cooling which lowers heat, cuts energy use – and could lead to more powerful data centers

By Wayne Williams

Copyright techradar

Microsoft unveils advanced AI cooling which lowers heat, cuts energy use - and could lead to more powerful data centers

Skip to main content

Tech Radar Pro

Tech Radar Gaming

Close main menu

the business technology experts

België (Nederlands)

Deutschland

North America

US (English)

Australasia

New Zealand

View Profile

Search TechRadar

Expert Insights

Website builders

Web hosting

Best web hosting
Best office chairs
Best website builder
Best antivirus
Expert Insights

Don’t miss these

AWS is working on proprietary innovative cooling tech to use on Nvidia GPUs for now — I wouldn’t be surprised if Graviton chips get it as well

AI workloads are reshaping infrastructure – here’s what data centers need to know

Not as thirsty as we thought – average data center uses less water than a ‘typical leisure center’, study claims

Microsoft announces “world’s most powerful data center” in latest billion-dollar AI spending splurge

Remember the doomed AI nation ship? A shipping giant is now planning a real, moving, floating data center that could power thousands of AI GPUs

This chip designer you’ve never heard of reveals first thermodynamic silicon in a bid to reduce AI’s unsustainable energy consumption

How can we create a sustainable AI future?

Chinese vendor launches liquid-cooled mini PC powered by AMD’s most powerful AI processor, with a built-in 400W PSU

US startup set to test new CPU that could herald the era of Exaflop on a single chip – I still can’t believe Neurophos’s OPU claims to deliver 234 POPS with a 1000 Petaflops model planned

“A potential revolution” – Microsoft backs researchers working on potentially the fastest fiber optics yet

AI’s infrastructure problem is bigger than we think—here’s how to solve it

Google says it will switch off energy-heavy AI usage at critical points if needed

Google, AMD and Intel catching up on Nvidia? Survey shows almost a third of AI teams now use non-Nvidia hardware

Microsoft says it will spend billions this year as AI moves pay off

Ryzen AI Max+ 395 CPU could be AMD’s sleeper hit against Nvidia’s AI dominance, as nearly 30 Strix Halo mini AI workstation models hit the market – including some rather funky ones

Microsoft unveils advanced AI cooling which lowers heat, cuts energy use – and could lead to more powerful data centers

Wayne Williams

24 September 2025

Microfluidics channels coolant through tiny etched grooves in chips

When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.

(Image credit: Microsoft)

Microsoft lab tests microfluidics cooling that triples efficiency compared to cold plates
Researchers say new system reduces GPU heat rise by 65 percent in trials
Microfluidics cooling promises denser, faster data centers while reducing energy costs for AI workloads

Microsoft has revealed it is testing a new chip cooling approach called microfluidics, aiming to solve one of the biggest challenges for artificial intelligence hardware.

As AI chips grow more powerful, they generate more heat than previous generations, pushing traditional cooling methods close to their limits.
Most data centers today use cold plates, where liquid passes through a plate attached to the chip. That system is separated from the heat source by multiple layers that trap heat, limiting its effectiveness. Microsoft’s new design takes a different path.

You may like

AWS is working on proprietary innovative cooling tech to use on Nvidia GPUs for now — I wouldn’t be surprised if Graviton chips get it as well

AI workloads are reshaping infrastructure – here’s what data centers need to know

Not as thirsty as we thought – average data center uses less water than a ‘typical leisure center’, study claims

Reliability tests
With microfluidics, tiny grooves are etched directly into the back of the chip, allowing coolant to flow onto the silicon itself and remove heat more efficiently.

Judy Priest, Corporate VP and CTO for Cloud Operations and Innovation at Microsoft, believes the new approach could reshape how future chips are built.
“Microfluidics would allow for more power-dense designs that will enable more features that customers care about and give better performance in a smaller amount of space,” she said in a blog post announcing the news.
Priest added that having proved the technology and shown the design worked, the next step was to test its reliability.

Are you a pro? Subscribe to our newsletter
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
Contact me with news and offers from other Future brandsReceive email from us on behalf of our trusted partners or sponsorsBy submitting your information you agree to the Terms & Conditions and Privacy Policy and are aged 16 or over.
In lab experiments, microfluidics removed heat up to three times better than cold plates, depending on workload, and also reduced GPU temperature rise by 65%.
By combining the channel design with AI that maps unique hotspots on the chip, Microsoft was able to direct coolant with greater precision.
Sashi Majety, senior technical program manager at Microsoft, said heat is becoming a barrier. “If you’re still relying heavily on traditional cold plate technology, you’re stuck,” he said, adding that within five years chips may run too hot to be cooled effectively with today’s systems.

You may like

AWS is working on proprietary innovative cooling tech to use on Nvidia GPUs for now — I wouldn’t be surprised if Graviton chips get it as well

AI workloads are reshaping infrastructure – here’s what data centers need to know

Not as thirsty as we thought – average data center uses less water than a ‘typical leisure center’, study claims

Microsoft believes its breakthrough shows how new approaches could enable faster, denser data centers while also lowering energy use for cooling.
The tech giant is collaborating with Swiss startup Corintis to refine the chip design.
In a bid to to massively reduce its water usage, Microsoft previously revealed it was working on a closed loop cooling system for data centers.
A number of other technology firms, including Lenovo, Dell, Supermicro and Giga Computing, are also racing to develop advanced cooling systems, as overheating risks putting a ceiling on the pace of AI progress.

Introducing microfluidic cooling: a breakthrough in chip cooling technology – YouTube

You might also like

Two cooling experts told me what AI’s biggest problem in the data center is
Researchers want to cool data centers using a common mechanism found in animals
Cooling high-density data centers with coolant distribution units

Wayne Williams

Social Links Navigation

Wayne Williams is a freelancer writing news for TechRadar Pro. He has been writing about computers, technology, and the web for 30 years. In that time he wrote for most of the UK’s PC magazines, and launched, edited and published a number of them too.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.

AWS is working on proprietary innovative cooling tech to use on Nvidia GPUs for now — I wouldn’t be surprised if Graviton chips get it as well

AI workloads are reshaping infrastructure – here’s what data centers need to know

Not as thirsty as we thought – average data center uses less water than a ‘typical leisure center’, study claims

Microsoft announces “world’s most powerful data center” in latest billion-dollar AI spending splurge

Remember the doomed AI nation ship? A shipping giant is now planning a real, moving, floating data center that could power thousands of AI GPUs

This chip designer you’ve never heard of reveals first thermodynamic silicon in a bid to reduce AI’s unsustainable energy consumption

Latest in Pro

Lovable alternatives

Libraseva urges users to patch now as it issues emergency fix following attacks

GitHub is finally tightening up security around npm following multiple attacks

Watch out – even small businesses are now facing threats from deepfake attacks

You’ve probably lost sight of your network – here’s why

Casino gaming giant hit by major cyberattack – employee information and more stolen, here’s what we know

Latest in News

Qualcomm’s new Snapdragon 8 Elite Gen 5 will power the next Galaxy and phones that will hear and see everything

‘I’ve seen it, it’s incredible’: Qualcomm CEO hypes new desktop Android OS that sounds like a genuine game-changer

Hisense’s new UST laser projector promises premium home theater up to 150 inches, for a very tempting price

Logitech’s new solar-powered keyboard can last 4 months in complete darkness on a single charge – and works with artificial light too

New Stranger Things season 5 trailer teases one last quest for the Hawkins crew – and lots of unseen footage for the hit Netflix show’s final hurrah

Proton VPN’s no-logs policy holds up under scrutiny of fourth independent audit

LATEST ARTICLES

This Blade Runner-inspired electric scooter can hit 65mph, travel for 60 miles and is getting a global launch soon

Are you an Apple Mac user? Cybercriminals are using this popular website to target you with malware and infostealers – here’s what you need to stay safe

“Making a great chip means nothing if we can’t do it the next year” – Qualcomm unveils powerful new Snapdragon X2 Elite chips for faster, better laptops

Qualcomm’s new Snapdragon 8 Elite Gen 5 will power the next Galaxy and phones that will hear and see everything

Terramaster uses the fastest Intel Celeron CPU ever in its new up-to-120TB 4-bay NAS – and I am curious about the new TRAID feature

TechRadar is part of Future US Inc, an international media group and leading digital publisher. Visit our corporate site.

Contact Future’s experts

Terms and conditions

Privacy policy

Cookies policy

Advertise with us

Web notifications

Accessibility Statement

Future US, Inc. Full 7th Floor, 130 West 42nd Street,

Please login or signup to comment

Please wait…