ASU expands AI research capabilities with Intel hardware donation
Arizona State University received a multimillion-dollar technology donation from Intel that will increase its AI computing power up to tenfold and give researchers and students access to large-scale processing capacity previously unavailable to most campus users.
The hardware donation supports ASU's effort to strengthen national competitiveness in AI and remove barriers to advanced computing across disciplines. ASU President Michael Crow said the university must make powerful technology accessible to lower entry barriers for researchers and students pursuing solutions to complex problems.
New research platform opens access
ASU has integrated Intel's hardware with its existing Sol supercomputer-ranked among the world's Top500-into a new system called the ASU AI Research Acceleration Platform, or AIR Platform. The system pairs Intel's AI accelerator chips with Google's Gemma and Meta's Llama large language models.
Researchers and staff access AIR through CreateAI Builder, the university's AI toolkit. Users start a new project and select the Gemma4 31B IT model designated for ASU AIR. The platform allows researchers to build custom AI experiences using their own data and instructions.
To date, the ASU community has created over 8,000 custom AI experiences using CreateAI Builder for academics, research, and operations.
Freeing resources for complex work
Before the donation, ASU operated hundreds of NVIDIA GPUs as its primary high-performance computing resource. The Intel accelerator chips expand and diversify that capacity without replacing existing infrastructure.
Sean Dudley, associate vice president at ASU's Knowledge Enterprise, said the new technology enables the university to support thousands of additional users developing generative AI models while shifting those workloads off existing resources. This frees up computing power for other intensive projects.
The expansion also addresses a practical constraint. Researchers previously had to use expensive national supercomputing systems for large-scale generative AI projects. Now they can run those projects in-house with full control over data privacy and security.
Real research gains traction
Jianming Liang, a professor in ASU's College of Health Solutions, developed an AI tool called Ark+ that helps physicians interpret chest X-rays more accurately. The model was trained on six public datasets of medical images paired with physician notes to identify common, rare, and emerging diseases.
With access to the new computing power, Liang plans to train a more ambitious model using over 1,000 datasets. The expanded model would identify diseases throughout the body, not just in chest imaging, and pinpoint their precise location and affected areas.
Suren Jayasuriya, an associate professor in ASU's School of Electrical, Computer and Energy Engineering, incorporated the AI accelerators into his deep learning class. Students benchmarked machine learning workloads on the new accelerators against traditional GPUs.
Long-term partnership focus
Intel and ASU have collaborated for years on addressing the U.S. semiconductor workforce shortage. The partnership supports graduate and undergraduate research, educator training, curriculum development, and experiential learning through equipment donations and mentorship.
By building a talent pipeline from K-12 through higher education, the two organizations demonstrate how industry and academia can coordinate on workforce solutions in advanced technology fields.
For researchers interested in using the platform, ASU's Research Computing Office offers consultations on incorporating AI into ongoing work.
Your membership also unlocks:
