DARPA's THREADS program has pushed radar transistors to about five times today's power density. That heat leaves the chip and has to cross a heat exchanger before it reaches seawater or outside air. On a ship or an aircraft, that exchanger sits in a fixed space, on a fixed power feed, and has to survive shock, vibration, and salt. Moving five times the heat with today's exchangers means a bigger box or much more pump power, and the platform has room for neither.
So we read The Defense Post's report on DARPA's THREADS program closely. BAE Systems' research arm, FAST Labs, has completed the first phase of the program and moved into the second, under a $12 million contract awarded in 2024. THREADS targets the heat limits of gallium nitride transistors, the chips at the heart of modern radar and communications electronics, the ones that send and receive radio signals. Those chips can only run as hard as their cooling allows. In the first phase, the teams reached about five times the power density of today's devices, which DARPA says corresponds to roughly doubling radar range. The work is at BAE's center in Nashua, New Hampshire, with Modern Microsystems and four universities.
Doubling radar range without a bigger antenna is a large deal for a program office: the same footprint, the same power feed, twice the reach. We agree the chip is the right place to start.
The heat has to go somewhere
Our question is what happens after the chip. Five times the power from the same chip area means five times the heat coming out of the same spot. THREADS gets that heat out of the transistor. The article doesn't follow it further, and on a ship or an aircraft, further is the hard part.
From the chip, heat runs through a cold plate, into a coolant loop, and finally through a heat exchanger that hands it off to seawater, outside air, or a vehicle's radiator. Moving five times the heat through that loop with today's equipment leaves two options. Pump the coolant much faster, which scales badly: in a typical loop, doubling the flow takes roughly eight times the pumping energy. Or fit a bigger heat exchanger, and there is no room for a bigger anything.
The field adds its own demands. The Coast Guard's recent request to industry on shipboard lasers asked about cooling requirements in the same breath as shock, vibration, sea spray, and airborne grit. A cooling loop that works in a lab and fails at sea is a program delay.
Where we come in
The way out is a heat exchanger that moves more heat in the same box, at the same weight, without a bigger pump. Conventional units are built from stamped plates and brazed fins, which fixes how much surface fits in a given envelope. We 3D print ours, so the internal channels are shaped to pack in surface while letting coolant flow freely. Our 1 MW heat exchanger, built for the Open Compute Project's Deschutes cooling design, is the data center version of the same idea. Printing puts more heat transfer surface into a given volume, and on a ship or an aircraft the volume is fixed. We're also in NATO DIANA's 2025 cohort, which is where we take this work for defense.
What program managers should ask
When a supplier promises more power from the same electronics, ask where the heat goes at the end of the loop, what that equipment weighs, and what it does to the pump. We're building our answer to fit the space that's already there.