Artificial intelligence (AI) is getting smarter, but with its increasing capacity comes a growing energy footprint.<\/p>\n
As modern AI models expand, massive data centers are needed \u2014 technology that consumes levels of energy and finite natural resources that some experts warn is unsustainable.<\/p>\n
To tackle this challenge, the MATRIX AI Consortium for Human Well-Being at UT San Antonio plans to launch a new initiative that establishes a national hub for \u201cneuromorphic\u201d computing available for public use.<\/p>\n
Neuromorphic computing is a revolutionary approach that mimics the human brain\u2019s structure to process information with a fraction of the energy used by traditional computers. Unlike standard processors that crunch data in a fixed sequence, neuromorphic chips operate like biological neurons. They are event-based, meaning that they activate only when there is something new to process, saving energy between events.<\/p>\n
The initiative, called THOR: The Neuromorphic Commons,<\/a> is funded by the National Science Foundation.<\/a> THOR will make the promising technology available for researchers nationwide to explore and conduct experiments, serving as the largest-ever full-stack neuromorphic platforms to be open to the public.<\/p>\n \u201cTHOR is the US national hub for neuromorphic computing. We are democratizing the technology, expanding industry-academia partnerships and serving as a catalyst for bringing neuromorphic computing closer to real-world applications,\u201d said Dhireesha Kudithipudi, PhD, founding director of MATRIX AI Consortium and lead scientific PI for THOR. Kudithipudi also serves as the Robert F. McDermott Chair in Engineering at UT San Antonio.<\/p>\n The national collaboration brings together research partners across the country, including Co-PIs Catherine Schuman, PhD (UT Knoxville) and Gert Cauwenberghs, PhD (UC San Diego). Tej Pandit, PhD, (UT San Antonio) serves as the project\u2019s AI Scientist, with Vijay Janapa Reddi, PhD, (Harvard University) supporting benchmarking efforts and William Severa, PhD, (UT San Antonio) contributing as a Community Leader. A global scientific advisory board and additional community leaders will further support outreach and training efforts.<\/p>\n Library for supercomputing<\/p>\n Historically, access to large-scale neuromorphic hardware has been exclusive to well-funded industry labs or select universities.<\/p>\n THOR upends that paradigm by making these powerful systems accessible to researchers and students across the United States, said Pandit.<\/p>\n \u201cOur goal is to host neuromorphic hardware developed by the THOR team and our partners, building a specialized center where the community can learn and develop the next generation of neuromorphic systems,\u201d he said.<\/p>\n Much like a public library, anyone can apply for access, and the resources will be free to use. Researchers will be able to enter a queue to run their experiments on the hardware. Once a user finishes his or her work, the system becomes available for the next person, allowing for high utilization of the resource.<\/p>\n The team estimates that around 50 researchers could use the resource at once, but the precise number will vary depending on the complexity of the programs they are testing.<\/p>\n Powering the future<\/p>\n At the core of THOR is the SpiNNaker2 system, developed in partnership with SpiNNcloud<\/a>.<\/p>\n SpiNNcloud is a massive computing platform that utilizes roughly 400,000 highly parallel processing elements, making it one of the largest neuromorphic systems available for public research. Parallelism refers to a system\u2019s capacity to complete a complex task quickly by breaking it down into smaller tasks and executing them simultaneously.<\/p>\n The system uses ARM-based cores \u2014 the same energy-efficient technology found in smartphones \u2014 specifically designed to simulate neurons and synapses. The architecture allows researchers to build \u201cspiking neural networks\u201d that process information in pulses, similar to how the brain signals muscles to move or eyes to process light.<\/p>\n Recent coverage from the publication HPCwire<\/a> notes that the deployment places UT San Antonio among the top tier of institutions globally hosting such large-scale neuromorphic capacity.<\/p>\n From theory to practice<\/p>\n While the technology is complex, the potential applications are practical and wide-ranging.<\/p>\n \u201cThere are so many devices right now that are limited by resources, in particular by a battery,\u201d Severa said.<\/p>\n He points to potential future applications, including wearables like smarter pacemakers that could adapt to a patient\u2019s changing physiology and hearing aids that could filter out unwanted noise.<\/p>\n \u201cMy mom uses a pacemaker, and sometimes when she goes in for a checkup, she can get really generic information,\u201d Severa said. \u201cBut it\u2019d be really nice if she could have a pacemaker that was constantly monitoring for signs of distress, and then it could alert the doctor. It\u2019d be great if it could adapt based on changes in her physiology.\u201d<\/p>\n Neuromorphic computing could also extend the life of mobile devices, such as drones, or devices used in remote environments, including wildlife trackers that can process data locally in forests without rapidly draining the batteries.<\/p>\n Beyond hardware efficiency, researchers, including those in the NUAI Lab,<\/a> will use the hub to explore how neuromorphic systems could transform how AI learns by addressing the challenge of \u201clifelong learning.\u201d Current AI systems tend to forget previous tasks when they learn new ones, a phenomenon known as \u201ccatastrophic forgetting.\u201d<\/p>\n The SpiNNaker2 system offers specialized mechanisms that may help address the issue and could one day allow AI to learn continuously over time.<\/p>\n MATRIX scientists are leading several major neuromorphic computing and AI semiconductor initiatives that are helping position San Antonio as a growing center for next-generation computing, including large-scale federally funded efforts such as DARPA\u2019s Next-Generation Microelectronics Manufacturing program, the Air Force Office of Scientific Research\u2019s CONCRETE Center of Excellence and the Department of Energy\u2019s MEERCAT Center focused on advanced computing and energy-efficient technologies. Kudithipudi noted that neuromorphic systems developed through these initiatives will help populate THOR, creating a pathway that moves semiconductor and AI research from laboratory prototypes to scalable, real-world applications.<\/p>\n Building a community of innovators<\/p>\n The official launch of THOR is slated for February 23 in the UT San Antonio San Pedro I building, with plans for a live demonstration to showcase the hardware capabilities. Beyond the hardware, the initiative focuses heavily on training and education.<\/p>\n \u201cTHOR will provide an opportunity to grow access and usability of neuromorphic systems,\u201d said Schuman. \u201cI believe broader adoption of the technology will lead to major innovations in the field.\u201d<\/p>\n Researchers and students interested in accessing the system or attending upcoming workshops can sign up for updates via the project\u2019s website<\/a>.<\/p>\n