No pain, no TX-GAIN: MIT unveils the most powerful AI supercomputer at any US university

MIT Lincoln Laboratory has unveiled TX-GAIN (TX-Generative AI Next), the most powerful AI supercomputer at any U.S. university, with a peak performance of two AI exaflops. The system is optimized specifically for generative AI applications and is already accelerating research across biodefense, materials discovery, cybersecurity, and other critical domains for both Lincoln Laboratory and MIT campus collaborations.

What you should know: TX-GAIN represents a significant leap in university-based AI computing capabilities, powered by over 600 NVIDIA graphics processing unit accelerators designed specifically for AI operations.

• The system achieved recognition from TOP500, which biannually ranks the world’s top supercomputers across various categories.
• With two quintillion floating-point operations per second, TX-GAIN claims the top AI system ranking among universities and in the Northeast region.
• The supercomputer came online this summer and is housed in an energy-efficient data center in Holyoke, Massachusetts.

How it works: Unlike traditional AI systems focused on categorization tasks, TX-GAIN is engineered specifically for generative AI applications that create entirely new outputs.

• Lincoln Laboratory Fellow Jeremy Kepner, who heads the Lincoln Laboratory Supercomputing Center (LLSC), describes generative AI as “a mathematical combination of interpolation (filling in the gaps between known data points) and extrapolation (extending data beyond known points).”
• The LLSC pioneered user-friendly software that allows researchers to access powerful computing without needing expertise in parallel processing configuration.
• “The LLSC has always tried to make supercomputing feel like working on your laptop,” Kepner explains.

Real-world applications: Researchers are already leveraging TX-GAIN’s capabilities across diverse scientific domains beyond traditional large language models.

• Teams are using the technology to evaluate radar signatures, supplement missing weather data coverage, detect network traffic anomalies, and explore chemical interactions for designing new medicines and materials.
• Rafael Jaimes from Lincoln Laboratory’s Counter–Weapons of Mass Destruction Systems Group notes the system enables modeling “significantly more protein interactions than ever before, but also much larger proteins with more atoms.”
• Historical LLSC projects include simulating billions of aircraft encounters for Federal Aviation Administration collision-avoidance systems and training autonomous navigation models for the Department of Defense.

Strategic collaborations: TX-GAIN enhances research partnerships between Lincoln Laboratory and MIT’s broader campus ecosystem.

• Active collaborations include the Haystack Observatory, Center for Quantum Engineering, Beaver Works, and Department of Air Force–MIT AI Accelerator.
• The Air Force initiative is rapidly prototyping and scaling AI technologies for the U.S. Air Force and Space Force, with flight scheduling optimization for global operations as one deployed example.
• Thousands of researchers across federally funded projects tap into LLSC resources for data analysis, model training, and simulations.

Energy efficiency focus: The LLSC team is actively addressing AI’s substantial energy requirements through innovative power-reduction methods.

• Researchers developed software tools that can reduce AI model training energy consumption by up to 80 percent.
• The facility operates as an energy-efficient data center, balancing high-performance computing with environmental considerations.
• “The LLSC provides the capabilities needed to do leading-edge research, while in a cost-effective and energy-efficient manner,” Kepner says.

What they’re saying: Research leaders emphasize TX-GAIN’s transformative potential for scientific breakthroughs.

• “TX-GAIN will enable our researchers to achieve scientific and engineering breakthroughs. The system will play a large role in supporting generative AI, physical simulation, and data analysis across all research areas,” says Kepner.
• “TX-GAIN is allowing us to model not only significantly more protein interactions than ever before, but also much larger proteins with more atoms. This new computational capability is a game-changer for protein characterization efforts in biological defense,” explains Jaimes.

Historical context: The TX-GAIN system continues Lincoln Laboratory’s computing legacy dating back to pioneering transistor-based machines.

• All LLSC supercomputers use “TX” nomenclature honoring the 1956 Transistorized Experimental Computer Zero (TX-0), one of the world’s first transistor-based machines.
• TX-0’s 1958 successor, TX-2, played a foundational role in pioneering human-computer interaction and artificial intelligence development.