• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

xMEMS Extends µCooling Fan-on-a-Chip Technology to Data Centers SSDs

Nomad76

News Editor
Staff member
Joined
May 21, 2024
Messages
1,401 (3.64/day)
xMEMS Labs, Inc., the pioneer of monolithic MEMS-based solutions, today announced the expansion of its revolutionary µCooling fan-on-a-chip platform into AI data centers, bringing the industry's first in-module active thermal management solution to high-performance optical transceivers. Originally developed for compact mobile devices, xMEMS µCooling now provides targeted, hyper-localized active cooling for dense, thermally-challenged environments inside 400G, 800G, and 1.6T optical transceivers—a critical yet underserved category in next-gen AI infrastructure.

Unlike conventional cooling approaches that target high-power (kilowatt) processors and GPUs, µCooling focuses on smaller, thermally stressed components that large-scale cooling systems can't reach, such as optical transceiver DSPs, that operate at 18 W TDP or higher. These components introduce thermal challenges and increasingly limit transceiver performance and reliability as data rates scale.



xMEMS' monolithic MEMS fan, fabricated in standard silicon processes, pumps a continuous stream of silent, vibration-free high velocity air pulses and is the only active cooling solution small and thin enough to be embedded inside the transceiver module. Thermal modeling shows that µCooling can remove up to 5 W of localized heat, reducing DSP operating temperatures by over 15% and thermal resistance by more than 20%, enabling higher sustained throughput, improved signal integrity, and extended module lifetimes.

A key innovation in µCooling's system design is its implementation in a dedicated, isolated airflow channel that is thermally coupled to the transceiver's internal heat sources but physically separated from the optical path and core electronics. This architecture ensures that optical components remain protected from dust or contamination, preserving signal clarity and transceiver reliability while still delivering impactful cooling performance.

"As data center interconnect demands scale rapidly with AI workloads, thermal bottlenecks are emerging at the component level—especially in optical modules that are sealed, power-dense, and space-constrained," said Mike Housholder, VP of Marketing at xMEMS Labs. "µCooling is uniquely positioned to solve this by providing true in-module active cooling with no compromise to optics or form factor."

Market analysts forecast strong growth in high-speed optical connectivity, with Dell'Oro Group projecting 800G and 1.6T transceiver shipments to grow at over 35% CAGR through 2028. As these modules scale in performance and power, cooling challenges are becoming a critical barrier to adoption.

µCooling's solid-state, piezoMEMS design means no motors, no moving bearings, and no mechanical wear, enabling maintenance-free reliability and high-volume manufacturability. Its compact footprint, as small as 9.3 x 7.6 x 1.13 mm, and scalable architecture make it ideal for modular deployments across a wide range of interconnects, including QSFP-DD, OSFP, and future pluggable and co-packaged optics.

With µCooling now serving both mobile and data center markets, xMEMS is delivering on its vision of scalable, solid-state thermal innovation to unlock the next wave of high-performance electronics.

View at TechPowerUp Main Site | Source
 
I would argue that the piezo chip is in-fact a moving part... it vibrates much like a speaker does...just much faster and much shorter distances...it'll eventually wear out...How long, maybe 50-100k hours? Not really sure but its pretty innovative. Wonder if there's any noise generated by the unit since it's technically producing ultrasonic frequencies....
 
But why? There is already (a much stronger) forced airflow through servers, switches, routers and so on in a datacenter where these things are used and noise is the least of your concerns there.
What problem does this solve in a datacenter which a decently sized heatsink can't solve better?
This would be much better placed in devices where noise, size (in height), high TDPs and lack of airflow are equally important factors like handhelds, ultrabooks, etc.
 
I think it's targeted airflow, and it should help, cause heatsink now only remove the Heat from the top, and maybe the distance between the top and Heat source is still to big to efectively remove Heat with traditional heatsink.
Like You can lower temperatures in gpu with fan on the back side of gpu, and still gpu has massive heatsink, but with how much Heat it produce it is better to have more airflow in place that heat gathers.
 
On the subject of other uses I'd like to see piezoelectric cooling in, it'd be in wearables, such as helmets, headphones, VR/AR headsets, and even cooling vests, where even small amounts of airflow can make a big difference not only cooling electronics, but even the wearer, without having the same bulk as say, a 40mm-80mm fan or blower have.
 
Would be neat applying this to a tower cooler design in some manner for the individual fins to subtly vibrate to increase cooling.
 
Back
Top