MIT's Lincoln Laboratory Unveils TX-GAIN: The Most Powerful AI Supercomputer in U.S. Academia

On October 2, 2025, MIT's Lincoln Laboratory unveiled TX-GAIN (TX-Generative AI Next), the most powerful artificial intelligence supercomputer at any U.S. university. With a peak performance of two AI exaflops—equivalent to two quintillion floating-point operations per second—TX-GAIN is poised to significantly advance research in biodefense, materials discovery, cybersecurity, and other critical fields.

TX-GAIN is housed at the Lincoln Laboratory Supercomputing Center (LLSC) in Holyoke, Massachusetts. The system is powered by over 600 NVIDIA graphics processing unit (GPU) accelerators specifically designed for AI operations. This infrastructure enables researchers to perform complex simulations, data analyses, and model training at unprecedented speeds.

"TX-GAIN will enable our researchers to achieve scientific and engineering breakthroughs," said Jeremy Kepner, a Lincoln Laboratory Fellow and head of the LLSC. "The system will play a large role in supporting generative AI, physical simulation, and data analysis across all research areas."

Generative AI, which focuses on creating new data outputs rather than categorizing existing ones, is a primary application for TX-GAIN. At Lincoln Laboratory, teams are applying this technology to various domains, including evaluating radar signatures, supplementing weather data where coverage is missing, detecting anomalies in network traffic, and exploring chemical interactions to design new medicines and materials.

In the realm of biodefense, TX-GAIN's capabilities are particularly transformative. "TX-GAIN is allowing us to model not only significantly more protein interactions than ever before, but also much larger proteins with more atoms," said Rafael Jaimes, a researcher in Lincoln Laboratory's Counter–Weapons of Mass Destruction Systems Group. "This new computational capability is a game-changer for protein characterization efforts in biological defense."

The LLSC emphasizes user-friendly interfaces, allowing researchers to access supercomputing resources without needing expertise in parallel programming. This approach democratizes access to high-performance computing, enabling a broader range of scientists and engineers to leverage TX-GAIN's capabilities.

Energy efficiency is also a priority for the LLSC. The center has developed software tools that can reduce the energy required to train AI models by up to 80%, reflecting a commitment to sustainable computing practices.

The "TX" nomenclature of TX-GAIN pays homage to Lincoln Laboratory's Transistorized Experimental Computer Zero (TX-0) of 1956, one of the world's first transistor-based machines. Its successor, TX-2, played a pivotal role in pioneering human-computer interaction and artificial intelligence. TX-GAIN continues this legacy, representing a significant milestone in the evolution of supercomputing at MIT.

The deployment of TX-GAIN at MIT's Lincoln Laboratory represents a monumental advancement in AI supercomputing within the academic sphere. Its unparalleled computational power is set to drive significant progress in scientific research, technological innovation, and education, solidifying MIT's position at the forefront of artificial intelligence and high-performance computing.