BAE Systems' research and development arm, FAST Labs, has completed Phase I of the U.S. Defense Advanced Research Projects Agency's (DARPA) high-frequency device cooling program, THREADS, and secured continued support to move into Phase II. The $12 million research effort, which began in 2024, is now shifting from exploring materials and manufacturing processes to further pushing output and heat-dissipation performance. However, what has been announced this time is merely the fact of selection—BAE's specific power density, thermal resistance, and any additional funding amount have not been disclosed. The achievements confirmed in Phase I and the goals set for the overall program need to be read separately.
Research That Began with $12 Million Moves to Phase II
In May 2024, BAE announced it had received a $12 million contract from DARPA for THREADS. Its formal name is "Technologies for Heat Removal in Electronics at the Device Scale." It is research aimed at dissipating heat generated inside transistors at the device scale, in high-power radio frequency (RF) circuits.
What's new in the August 12, 2026 announcement is that FAST Labs has completed Phase I and, with continued support, is advancing to Phase II. Principal investigator Isaac Wildeson explained that Phase I confirmed the validity of the materials and process improvement approaches. However, no figures were provided that would allow evaluation of differences between approaches—such as measured output power density, junction temperature, or device lifetime. The contract amount, duration, and completion conditions for Phase II were also not disclosed.
The research is being carried out at the Microelectronics Center in Nashua, New Hampshire. BAE describes this facility as developing and manufacturing GaN and gallium arsenide (GaAs) integrated circuits and as certified as a U.S. Department of Defense Category 1A Trusted Supplier. Modern Microsystems has joined the effort here, and Penn State University, Stanford University, University of Notre Dame, and University of Texas at Dallas are also part of the research team. This structure places a company with mass-production facilities alongside universities strong in materials and device research on the same team—though it does not mean that mass-production readiness for the devices developed under THREADS has been achieved.
What 81 W/mm and One-Eighth Thermal Resistance Are Aiming For
Gallium nitride (GaN) is a wide-bandgap semiconductor used in RF amplifiers that operate at high power density. When the program launched in 2022, DARPA explained that while GaN had achieved output density more than five times that of previous-generation transistor technology, electronically an additional order-of-magnitude increase was possible. What prevents sustained operation is waste heat. Localized temperature rises degrade performance and shorten device lifetime, so actual circuits are operated at lower output than the theoretical limit.
THREADS aims to solve two challenges simultaneously. One is reducing the internal thermal resistance of the device while preserving the characteristics of the current-carrying channel. The other is moving heat away from hotspots in high-power transistors and dissipating it externally without degrading RF performance. Since adding a material that cools well but disrupts electrical characteristics would render the device unusable as an amplifier, this is not a program that competes on heat-dissipation performance alone.
The ultimate goal is to create test devices of highly efficient, highly reliable X-band transistors and power amplifiers that achieve an output power density of 81 W/mm while reducing thermal resistance to one-eighth. The figures of 81 W/mm and one-eighth are, at this stage, the program goals set by DARPA—not measured values that BAE has achieved in Phase I. Losing sight of this distinction would reverse the research stage and the achievement stage.
DARPA had previously pursued Near Junction Thermal Transport (NJTT), which aimed to reduce thermal resistance within 100 micrometers of the junction. That effort examined replacing substrates with high-thermal-conductivity diamond, removing low-thermal-conductivity layers, and liquid cooling near the junction, with a goal of tripling the power-handling capability of GaN amplifiers. Compared to NJTT, THREADS combines materials, transistor structure, and cooling pathways, and extends the scope of evaluation to X-band amplifier test devices.
A 5x Power Density Does Not Guarantee 2x Radar Range
In June 2026, DARPA announced that Phase I had increased RF power density approximately fivefold compared to current state-of-the-art devices, while maintaining the device lifetime required for defense applications. Translated into operational terms, this performance improvement would roughly double radar range.
However, the approximately fivefold figure is DARPA's explanation for THREADS participants as a whole, not a number published as BAE's individual measurement result. The specifications of the "current state-of-the-art devices" used as the comparison baseline, as well as the materials and transistor structures used by each team, have also not been disclosed. It is not possible to determine which approach contributed what portion of the fivefold improvement.
Radar range, too, is not a figure uniquely determined by a transistor's power density. It depends on how much of the device's output can actually be utilized by the amplifier and transmitter, as well as on the antenna and frequency. Results also vary depending on the target, receiver system, and signal processing. DARPA's 2022 projection was 2–3x, its 2026 Phase I explanation was roughly 2x, and BAE's latest announcement describes it as "nearly 3x"—the premises and maturity levels are not aligned across these figures. Therefore, these should be treated not as detection ranges uniformly confirmed on actual radar systems, but as estimates showing system-level effects when thermal constraints are relaxed.
Phase II: Connecting Device-Level Achievements to Military Systems
In Phase II, DARPA will further increase output and thermal performance. At the same time, it has begun system-level research to examine the RF output thresholds that would be meaningful for current and future military platforms. The plan is to provide these results to government agencies and government contractors to accelerate application selection and technology transition.
The very need for separate system-level research clearly illustrates where Phase I stands. Even if a transistor achieves both high power density and long lifetime, if the efficiency or footprint advantages disappear once it becomes an amplifier, it cannot be integrated into existing platforms. Cooling of the power supply and peripheral circuitry must also be achieved simultaneously. DARPA's stated condition—increasing output without increasing the size, weight, or complexity of equipment—is a stricter bar than device-level records alone.
BAE has the advantage of being able to advance this research at its existing GaN/GaAs manufacturing facilities. However, the yield, reproducibility, power-amplifier efficiency, and manufacturing cost needed for an adoption decision have not been disclosed. What Phase II advances is not merely the demonstration of a cooler transistor, but the selection process of determining which military systems can actually capture the performance gains. When 81 W/mm, one-eighth thermal resistance, and long-term lifetime are all satisfied in the same test device—and when those advantages are confirmed to carry through to the amplifier and platform—THREADS will move closer to becoming a procurable technology rather than merely a research achievement.