With Nature Article and $4M Grant, Schuman Advances Community-Level Neuromorphic Computing

Computers are generally built for small-scale tasks, like a laptop sending emails, or exascale tasks, like a supercomputer simulating weather patterns over the next 50 years.

But the world's first computers—brains—use the same neural building blocks whether they are inside a fruit fly or a human. What if all our technology were that versatile?

Neuromorphic computing (NMC) is an engineering philosophy that models computational systems on biological brains. Proof-of-concept studies have demonstrated that NMC hardware and software are more flexible, more energy efficient, and lower latency than conventionally designed deep learning (DL) software. The technology is promising for research in many scientific fields, including neuroscience, high-energy physics, and artificial intelligence (AI).

Unfortunately, large-scale NMC systems can cost millions of dollars, putting them beyond the reach of most research groups.

"The lack of availability of these systems has, in my opinion, significantly hindered the progress of the field," said Catherine Schuman, a neuromorphic computing researcher and assistant professor in the Department of Electrical Engineering and Computer Science (EECS). "Access is one of the biggest hurdles to people pursuing neuromorphic research."

In a review article published in Nature today, Schuman and 22 other NMC experts spanning academia, government, and the industry made recommendations for developing NMC to its fullest potential.

"We are hoping that these recommendations will inspire more institutions to invest in NMC hardware, software, and infrastructure, and inspire a wider community of scientists and researchers to use NMC systems," Schuman said. "Nature is a very exciting venue for this work, as it will bring neuromorphic computing to the attention of as much of the scientific community as possible."

While the article outlines many important components needed for NMC's advancement, the experts' overarching recommendation is to democratize access to the technology with a community-level NMC system.

In true Vol fashion, Schuman has already begun lighting the way forward. Along with two co-authors on the Nature article—lead author Dhireesha Kudithipudi of the University of Texas at San Antonio (UTSA) and Gert Cauwenberghs of the University of California San Diego—Schuman secured a $4 million grant from the National Science Foundation (NSF) last fall to create the much-needed community resource.

THOR: a Community-Level NMC Resource

Drawing parallels to the explosion in AI availability, Schuman and her co-authors believe that NMC will only achieve its true potential when it is accessible to users who do not know the mathematical processes behind it. (How many ChatGPT users can explain how machine learning works?)

Such widespread use requires the NMC community to create a large library of open-source code, tools, and training programs. For that, they need community-level access to NMC systems.

Last fall, Kudithipudi, Schuman, and Cauwenberghs began working on a large-scale NMC system called The Neuromorphic Commons (THOR). Though physically housed at UTSA, THOR will give researchers across the nation—and maybe, someday, the world—open access to a computer system that blends classical and NMC technology.

Schuman is the team's community outreach coordinator, developing workshops, trainings, and a groundwork for the eventual code library THOR's users will create.

"I'm trying to get as many people as possible playing with, developing, and researching this system, including researchers outside the neuromorphic community who previously wouldn't have had access to NMC," she said. "A lot of tools and development in NMC have been isolated one-offs in research groups. We're hoping to build a community-based toolset that is more usable and more accessible to the average person."

She is especially excited to bring THOR's capabilities to young scientists, creating a new generation of computer scientists steeped in NMC. In addition to the high school students and undergraduates who work in her lab, Schuman hopes to include undergraduates who take her course, Biologically-Inspired Computing, in the spring of 2026.

"Previously, it's been nearly impossible for students to access NMC systems," she said. "This will give them an opportunity to access these systems and build their own neuromorphic applications and software for the broader scientific community to use. They're going to get a sneak preview of THOR while helping me find out what's most useful and interesting for the community."

Provided by University of Tennessee at Knoxville