Collaboration between emerging technology enterprises and research universities is one of the most consistently winning tactics for any nation building (or rebuilding) an industrial ecosystem.It’s an especially constructive approach in a handful of nations, including the US, with the highest concentrations of both viable deep-tech startups and world-class academic institutions.Fabric8Labs and the University of Illinois (UI) have just announced a textbook example of this kind of work, leveraging Fabric8Labs’ electrochemical additive manufacturing (ECAM) process to produce direct-to-chip (D2C) copper cold plates for data center thermal management.
As we pointed out at AM Research in our 2025 report on AM for the data center market—a report which includes coverage of Fabric8Labs—the rising power loads demanded by AI chips call for heat exchanger solutions that deploy liquid cooling methods, in addition to the air cooling methods that have been the standard for decades.AM can play a central role in the development of that new class of heat exchangers, thanks to the ability to use cooling designs, characterized by “tightly packed metal ‘fins'”, which are optimized for the surface area of chips.Fabric8Labs and UI have published the results of their initial work in the journal Cell Reports Physical Science.
Utilizing topology optimization methods, the collaborators iterated a series of different fin design possibilities with the objective of minimizing the power required to cool the relevant chips.According to UI, most existing methods for using such finned cold plates incorporate simple shapes like rectangles and cylinders.UI, on the other hand, designed cold plates “[w]ith pointed tops and jagged edges,” shapes that Fabric8Labs’ ECAM method is uniquely well-suited to produce.
In addition to the advantage in geometric complexity, Fabric8Labs also has advantages when it comes to material science.Since ECAM utilizes liquid metals, the technique is better for working with pure copper than are other AM methods, which tend to necessitate copper alloys, generally leading to weaker cooling performance.The UI researchers claim that their findings suggest the Fabric8Labs cold plates deliver improvements in data center cooling over other finned cold plates by 32 percent.
As the researchers note, most of the data out there involves work aiming to improve the cost efficiency of the manufacturing process.By instead focusing on maximizing the cooling performance of the cold plates, the UI researchers may have devised a superior method for lowering long-run data center operating costs, while simultaneously pointing to a path that implies a more sustainable carbon footprint.In a press release about UI engineers’ data center cooling research incorporating cold plates from Fabric8Labs, first author Behnood Bazmi said, “Cooling is the bottleneck in computer-chip design.
By bridging the gap between computational design and manufacturing capability, our approach provides a pathway for more energy-efficient liquid cooling of chips and other electronics.Our workflow can be applied to a wide range of cooling challenges across different length scales.” Senior author Nenad Milijkovic, a mechanical engineer at UI, said, “Topology optimization ends up converging on a design which is optimal in maximizing thermal performance and minimizing pumping power.…With our cold plates, data centers would only need to use 11 megawatts for cooling instead of 550 megawatts.” That potential is precisely why Fabric8Labs landed a $50 million investment round last November, only the latest big influx of funding for the San Diego company, and will be used largely to build up its manufacturing capacity in the US.
Working with institutions like UI is an excellent way to prime that same pump, as the company’s process has now undergone validation through a project supported by funding from the US Department of Energy (DOE).This project encapsulates what I’ve noted in recent posts about the role of defense spending in the US economy, and how AM may both impact and be impacted by changes in that broad dynamic.Bluntly, this is what the US government should be spending money on, as opposed to doubling down on the same defense procurement formula that has done such a disservice to readying US military personnel for duty, and has been a primary contributor to the accumulation of incomprehensibly large quantities of national debt.
The Pentagon is asking for $1.5 trillion for 2027.Can anyone seriously doubt that if even a tiny amount of effort was put into solving the problem, that the US could figure out a much better way to arm itself with a much smaller funding commitment? I say this because it absolutely mustn’t be overlooked that under the current arrangement, the Pentagon’s objective is in fact to figure out how to spend as much money as it possibly can.Shouldn’t we at least consider alternatives? I think the key to a starting point for strategizing how to spread the US federal budget more evenly across all its departments is to acknowledge how the current geopolitical era is demonstrating so convincingly that maintaining national security requires far more nuance than simply a plan to buy the most expensive weapons that the handful of largest defense contractors can come up with.
Cybersecurity and energy security, for instance, are much more relevant to everyone’s lives than the F-35.State-of-the-art data center hardware addresses both needs.Research projects like this one need to be prioritized.
Images courtesy of the University of Illinois Subscribe to Our Email Newsletter Stay up-to-date on all the latest news from the 3D printing industry and receive information and offers from third party vendors.Print Services Upload your 3D Models and get them printed quickly and efficiently.Powered by FacFox Powered by 3D Systems Powered by Craftcloud Powered by Xometry 3DPrinting Business Directory 3DPrinting Business Directory
Read More
