Researchers at the Massachusetts Institute of Technology (MIT) have implemented a new in-rack immersion cooling system created by Ferveret. This technology is being utilized in a high-performance computing platform to quicken AI-driven materials research for next-generation nuclear energy.
Ferveret’s method, based on techniques from the nuclear industry, is an immersion cooling approach where the server and heat exchanger are housed in a compact box that can be placed in a compute rack. In an earlier conversation this year, DCD spoke with Reza Azizian, the CEO and co-founder of the company. [Date: 03 Sep 2026] Ferveret claims that by quickly dissipating heat within the box, it can unlock up to 35% more usable compute from the same power envelope.
Azizian explains Ferveret’s approach to evaluating high-performance AI infrastructure: they focus on real scientific workloads instead of synthetic benchmarks or artificial stress tests, which provides a more meaningful measure of system performance and creates tangible value for researchers. MIT’s AI-driven materials research, which continuously pushes GPU infrastructure to its limits, is an ideal environment to demonstrate how advanced cooling enables greater computational performance and accelerates scientific discovery.
Ferveret’s platform allows us to conduct research that is practically impossible with the usual academic computational resources. The rapid advancement in our ability to train bigger models, improve iteration speed, and conduct experiments incessantly is significantly boosting our discovery process.
Today, compute is evolving into an essential component of scientific discovery, akin to the laboratory’s role. According to MIT professor Ericmoore Jossou, progress in AI will rely not just on improved algorithms, but also on the availability of the necessary infrastructure.
Both Azizian and his fellow founder, Matteo Bucci, are graduates of MIT. A project backed by the company aims to enhance the interpretation of congruent X-ray images obtained from nuclear substances.
