The University of Florida’s Center for Heterogeneous Integration and Photonics (CHIP Lab), led by Hamed Dalir, Ph.D., has been awarded up to $435,000 by the Naval Surface Warfare Center, Crane Division, to pioneer a transformative approach to analog-to-digital (ADC) and digital-to-analog (DAC) conversion for next-generation AI accelerator processors. The project is a joint effort with Mahdi Nikdast, Ph.D., of Colorado State University, a recognized leader in photonic interconnects and circuit-level optimization.
The initiative aims to eliminate longstanding bottlenecks in signal conversion by developing a novel integrated architecture that directly bridges optical, photonic, analog, and digital domains. Unlike conventional systems that rely on multiple discrete components—each contributing to latency and energy inefficiency—this new class of converters uses engineered phase-change materials (PCMs) and tailored optical loss elements to perform domain conversion in a single compact unit.
“Traditional ADCs and DACs require several intermediary stages that slow down data transfer and consume excessive power. Our integrated system collapses those stages into one, using phase-change materials and binary-weighted optical losses to achieve efficient, high-speed, and low-power domain transformation.”
– Hamed Dalir, Ph.D., Associate Professor, ECE
Architecture enables direct conversion of optical signals to digital data—and vice versa—without the need for analog intermediaries. This innovation is expected to dramatically reduce power consumption and latency, offering a major advantage in edge computing, defense systems, and autonomous platforms.
“Transitioning from conventional electrical ADCs to PCM-based ADCs, combined with photonic links, marks a fundamental shift in how we design energy-efficient AI hardware,” Dalir added. “This could pave the way for fully photonic AI accelerators where data remains in the optical domain until absolutely necessary.”
The significance of this work has already been recognized at the national level. The University of Florida Innovation Office has filed a U.S. patent application for the technology, citing its potential to drive a paradigm shift in AI hardware, neuromorphic computing, and brain-inspired devices.
This new project builds on Dr. Dalir’s earlier work, including a Semiconductor Research Corporation award to design the digital core of the chip. In collaboration with TSMC, Dr. Dalir is also advancing the integration of Indium Tin Oxide (ITO) technology on the silicon platform—pushing the limits of compact, low-power photonic systems.
With this award, University of Florida researchers continues to lead national efforts in pushing the boundaries of integrated photonics, packaging, and AI acceleration technology.