JetCool to Showcase Cooling Demos with Flex at AI Infra Summit

As AI workloads accelerate compute density and power consumption, traditional cooling methods are no longer enough. At AI Infra Summit 2025, JetCool—a Flex company and leader in direct-to-chip liquid cooling—is showcasing how precision-engineered, end-to-end liquid cooling solutions are redefining performance, efficiency, and scalability in AI data centers.

Flex + JetCool: Your One-Stop Data Center Partner

JetCool’s single-phase, direct-to-chip liquid cooling solutions are now backed by the global scale and manufacturing excellence of Flex. By combining Flex’s manufacturing expertise and global footprint with JetCool’s cutting-edge technology, we empower hyperscalers to deploy AI infrastructure smarter, faster and more efficiently.

As part of Flex, we tightly integrate power, cooling and infrastructure with global operations to design, build, and deliver end-to-end liquid cooling solutions—all in-house. From hybrid cooling systems and precision cold plates to coolant distribution units, in-rack manifolds, and quick disconnect couplings, our portfolio supports customers at every stage of their liquid cooling journey. These solutions leverage JetCool’s patented microconvective cooling technology, which targets processor hotspots directly delivering precision cooling—even under the most demanding workloads.

Together, Flex and JetCool offer more than just technology—we’re a long-term partner supporting the entire data center value chain, enabling scalable, high-performance infrastructure anywhere in the world.

Products on Display at AI Infra Summit 2025

These demos offer a firsthand look at how liquid-first designs are solving the performance, density, and sustainability challenges that traditional air cooling can no longer meet.

Flex Liquid-Cooled 2U Server with JetCool’s SmartPlate ™ Technology

Explore Flex’s liquid-cooled compute reference platform, now integrated with JetCool’s SmartPlate™ cold plate technology. This demonstration highlights how customers can start with a proven reference design and tailor it to meet specific requirements— whether optimizing for power delivery, mechanical configuration, or advanced liquid cooling. With JetCool’s advanced cold plates built in, this platform offers a flexible foundation for rack-and-stack deployments, making it an ideal starting point for AI data center builds. Check out our OCP EMEA demo video to learn more.

Precision Cold Plates with Microconvective Cooling®

JetCool’s SmartPlate cold plates utilize patented microconvective cooling to directly target processor hotspots, delivering exceptional thermal performance. Validated by third-party testing, SmartPlate technology outperforms conventional microchannel solutions by up to 25%, enabling sustained high-power operation without throttling. On display will be JetCool’s cold plate for the AMD MI355 GPU, as well as the SmartLid™ direct-to-die cooling solution— engineered for high-TDP processors exceeding 4 kW per socket. These technologies demonstrate how precision cooling extends component lifespan and maintains headroom for denser compute.

SmartSense Coolant Distribution Unit

JetCool’s SmartSense CDU is a high-performance, rack-mounted liquid-to-liquid coolant distribution unit engineered to meet the rigorous demands of GPU-dense, high-power environments. Capable of cooling up to 300 kW per rack—or distributed across neighboring racks—it ensures reliable thermal management for next-generation workloads, keeping your infrastructure optimized and resilient.

AMD MI355 GPU SmartPlate Cold Plate Liquid Cooling

The JetCool Advantage

AI racks are densifying fast. As rack-level power climbs from tens to hundreds of kilowatts, cooling must evolve from room-level interventions into chip-level engineering. JetCool’s approach focuses on three practical outcomes:

Performance: Lower die temperatures let hardware run at higher sustained power without thermal throttling.
Efficiency: Warm-water, high-ΔT operation reduces chiller dependency and improves data-center energy metrics.
Scalability: Modular CDUs and factory-integrated rack builds let you replicate validated designs across sites without costly field rework.