In vacuum there is no air to carry heat away — radiation is the only exit, and capacity scales with area and T⁴. A single GB300 NVL72-class rack dissipates ~140 kW; getting that heat from the die to deep space is the hardest layer of orbital compute.
A standard, drop-in thermal module — cold plate, pumped PGW loop, deployable tube-and-fin radiator — that your team bolts on and ties into the coolant loop. Sized to a full 150 kW load and anchored on an NVIDIA GB300 NVL72-class rack.
We anchor on the closed, warm, pumped-liquid metal radiator — the architecture already validated on the newest generation of large orbital-compute platforms, proven at TRL 7. Not an experimental membrane. Not a moving belt. Our job is to deliver the same heat rejection as a drop-in block that beats the baseline on the numbers that matter.