The Hard-Sci SF Zoom Proudly Presents Our Favorite Engineering Professor, Tosi Adegbija

On February 7, 2026, at 9:00 a.m.
The Hard-Sci SF Zoom proudly presents Our Favorite Engineering Professor, Tosi Adegbija, who returns by popular demand.

Live From Tucson!

TITLE: “Right-Sizing Intelligence: Hardware, Software, and the Sustainable Future of AI ”

DESCRIPTION:
The rapid deployment of AI systems has created a physical infrastructure problem that the industry’s breathless marketing tends to gloss over. Data centers now rival mid-sized countries in power consumption, largely because the dominant paradigm treats trillion-parameter general-purpose models as the hammer for every nail. Meanwhile, the silicon running these models still primarily relies on underlying computer architectures that burn more energy shuttling data than performing useful computing. The gap between capability claims and efficiency reality is widening. In this talk, I make a pragmatic case for efficiency at two levels: context-specific AI that right-sizes models to actual task requirements and neuromorphic hardware that borrows the brain’s sparse, event-driven efficiency.

Sustainable AI won’t come from building more power plants—it requires rethinking what we’re building and why.

PROFILE:
University of Arizona Electrical and Computer Engineering Director of Graduate Studies,Thomas R. Brown Endowed Fellow, Associate Professor of Electrical and Computer Engineering, Member of the Graduate Faculty. In 2019, Adegbija received the National Science Foundation Career Award for his research on runtime adaptable spin-transfer torque RAM (STTRAM) cache memories for energy-efficient microarchitectures and low-power embedded systems design and optimization methodologies, and microprocessor optimizations for the Internet of Things, or IoT.

Register in advance for this meeting: https://us02web.zoom.us/meeting/register/3MW8AMnSSjWWK5lzg27tXQ
No password