SemiValley
Two-Phase Jet-to-Chip Liquid Cooling
4x less flow. 16× the thermal performance. Cooling for AI datacenters at scale.
Performance That Speaks for Itself
<1.05
PUE
vs 1.3–1.6+ air
16×
thermal
vs single-phase
~$228M
annual savings
3 GW data center
Versus a modern single-phase liquid-cooled facility at PUE 1.15
10 kW+
TDP capacity
per chip
Research & Program Affiliations
Overview
Engineered for the highest heat-flux AI chips.
As accelerator TDPs push past 1 kW, air and single-phase liquid cooling can no longer keep junction temperatures in spec. Two-phase jet-to-chip cooling handles the load with a fraction of the pump work.
Hyperscalers
Scale liquid cooling to next-gen accelerator racks
Server & rack OEMs
Integrate into existing form factors
Colocation operators
Cut PUE and free stranded capacity