Purdue ECE professor Haitong Li earns NSF CAREER Award to build a new class of AI hardware
Purdue University’s Haitong Li has received a National Science Foundation CAREER Award to develop new computing chips and systems that could help future artificial intelligence (AI) reason, solve complex problems, and respond more effectively in real time while consuming less energy.
Li, an assistant professor in Purdue’s Elmore Family School of Electrical and Computer Engineering, will lead a project focused on neuro-symbolic AI, an emerging approach that combines the strengths of neural networks, which power today’s large language models and other generative AI systems, and symbolic reasoning, which applies rules and logic to draw conclusions and make decisions.
The goal is to make AI systems better at multi-step problem solving in complex, real-world environments. But that kind of advanced reasoning pushes today’s computing platforms to their limits, demanding hardware that is faster and more energy efficient yet remains programmable, versatile and scalable across AI workloads.
Li’s project, “Efficient and Scalable Neuro-Symbolic Cognitive Computing on Three-Dimensional Integrated Circuits and Systems,” aims to address that challenge by developing new computing foundations for neuro-symbolic AI through cross-stack co-design, specialized memory technologies, and advanced three-dimensional integration.
“What excites me most is that we get to rethink the entire computing stack, from memory devices and circuit architectures all the way up to how AI algorithms run on hardware,” Li said. “Neuro-symbolic AI needs a fundamentally different kind of chip. This project lets us build it from the ground up so that future AI can reason, respond and assist more effectively and efficiently.”
The project will focus on four connected research directions. Li’s team will develop a co-design framework that bridges neuro-symbolic AI models, memory-centric chip architectures and emerging semiconductor technologies. They will then build programmable accelerator chips that use compute-in-memory technology to run neuro-symbolic AI workloads more efficiently. In addition, the researchers will develop tailored three-dimensional integrated systems enabled by beyond-silicon logic and reconfigurable memories stacked atop silicon CMOS. The team will also produce tools that generate chiplet hardware components, enabling a closed-loop workflow for continued algorithm-hardware co-design.
In practical terms, this new class of hardware could equip future AI systems with two capabilities at once: reasoning through complex, multi-step problems and operating at far lower energy. Both matter across applications ranging from smart devices and autonomous systems to data centers and scientific computing.
The CAREER Award also includes an education component. Li will develop new course materials and hands-on learning experiences in neuro-symbolic AI and semiconductor technologies for Purdue students and K-12 educators. The goal is to expand understanding of next-generation computing while helping prepare students for careers in the growing semiconductor workforce.
The NSF CAREER Award is among the foundation’s most prestigious awards for early-career faculty members, supporting researchers who have the potential to serve as academic role models in research and education.