3D-Printed Copper Cold Plate Could Slash Data Center Cooling Energy to 1.1% of Facility Power
May 12, 2026
3D-Printed Copper Cold Plate Could Slash Data Center Cooling Energy to 1.1% of Facility Power
Researchers at the University of Illinois Urbana-Champaign have developed a 3D-printed cold plate made entirely of pure copper that could dramatically reduce the energy consumed by data center cooling systems. The innovation, published in the journal Cell Reports Physical Science and supported by the U.S. Department of Energy, addresses a critical bottleneck in high-performance computing: thermal management.
Cooling currently accounts for a substantial portion of data center energy use, with traditional air cooling systems consuming up to 30 percent of a facility's total electricity. The team behind the new cold plate claims that if deployed in a data center, their design would require just 1.1 percent of the facility's energy—a reduction that could translate into massive savings at scale. For a 1-gigawatt facility, for example, the cooling load would drop from 550 megawatts to only 11 megawatts, according to Nenad Miljkovic, a professor and the study's senior author.
The cold plate's performance stems from a design process called topology optimization. Starting from a simple rectangular fin shape, the team used a mathematical algorithm to iteratively refine the fin geometry, balancing thermal performance against the pumping power needed to circulate coolant. The resulting fins feature pointed tops and jagged edges—far more complex than conventional rectangular, conical, or cylindrical shapes. "Topology optimization ends up converging on a design which is optimal in maximizing thermal performance and minimizing pumping power," Miljkovic explained.
Manufacturing such intricate geometries required an advanced technique. The researchers partnered with Fabric8 to employ electrochemical additive manufacturing (ECAM), a process that deposits copper layer by layer through electrochemical plating. ECAM can produce pure copper parts with details as fine as 30 to 50 micrometers, less than the width of a human hair. This precision is essential because copper, while offering superior thermal conductivity compared to aluminum alloys, is notoriously difficult to work with using conventional methods.
When tested against cold plates with standard rectangular fins, the optimized copper plate delivered up to 32 percent better cooling performance. While the 1.1 percent energy claim is notably lower than the 5 to 15 percent typically consumed by conventional liquid cooling systems, the researchers emphasize that even incremental improvements in efficiency can have outsized impacts in the context of large-scale AI and hyperscale data centers. "Cooling is the bottleneck in chip design," said first author Behnood Bazmi, a mechanical engineering graduate student. "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."
The work highlights a growing convergence of computational design tools and advanced manufacturing, suggesting that future data center cooling infrastructure could be tailored to specific chip layouts and thermal loads. As the industry grapples with rising power demands from AI workloads, innovations like the copper cold plate offer a tangible path toward more sustainable operations.
Source: datacenterdynamics
