Artificial intelligence, or AI, adoption is accelerating demand for faster computing and data communication. As these systems grow larger and more capable, they demand enormous amounts of computing power, and building the hardware required to support these systems poses a major engineering challenge.
Jiaqi Gu is working on a cross-disciplinary solution that converges physics, AI and semiconductor engineering to address the problem. His goal is to make it faster and easier to design electronic-photonic chips that combine electronics and light-based technologies to support more efficient AI computing. He also wants to make this research area more accessible and collaborative.
Gu, an assistant professor in the School of Electrical, Computer and Energy Engineering, part of the Ira A. Fulton Schools of Engineering at Arizona State University, has been awarded a distinguished National Science Foundation Faculty Early Career Development Program (CAREER) Award to develop the foundational infrastructure and toolchains needed to advance these systems.
“Receiving the CAREER Award is deeply meaningful and encouraging,” Gu says. “It is an important acknowledgment of the long-term research and education vision my group is building at the intersection of electronic design automation, photonics, AI and semiconductor systems.”
Bringing light to the future of computing
Modern computing systems rely on electrical signals. As AI workloads continue to expand, the energy required to move and process data has become a growing challenge.
Gu sees light as a new direction for AI computing, communication and sensing systems.
His research focuses on integrated photonics, which uses light to transport information on a chip. Photonics offers a promising alternative because light can transmit more information at higher speeds while consuming less energy.
“Integrated photonics attracted me because it offers a fundamentally different way to move and process information,” Gu says. “Light can carry information with very high bandwidth, low latency and low energy consumption, which makes it exciting for future computing and communication systems.”
Electronic chip design has benefited from decades of mature design automation tools, while photonic chip design automation is still in its infancy.
Light behaves like a wave, so small changes in the shape of a device, the chip layout, fabrication processes, or even the operating environment, can change how the system works. That makes photonic chip design very different from writing software or even designing digital logics.
Engineers also often need expertise across physics, optics, electronics and system architecture design. Creating a working prototype remains costly and can require extensive simulations and numerous rounds of trial and error.
These challenges have slowed the adoption of photonic technologies at a time when demand for more efficient computing continues to accelerate. Addressing those barriers requires a new generation of design tools.
Building the tools to accelerate innovation
Through his work, Gu aims to make photonic chip design faster, more systematic and more accessible.
His research will develop an open-source electronic-photonic design automation, or EPDA, framework that helps engineers design and evaluate complex electronic-photonic systems. The goal is to provide a more streamlined process that enables designers to consider physical behavior, manufacturing constraints and system-level performance earlier in the development process.
The project will also create new tools to help researchers simulate photonic devices more efficiently, design systems that are more tolerant of manufacturing variations and automate the layout of large-scale photonic circuits. Together, these capabilities will help reduce development time, lower costs and improve reliability.
Gu hopes the work will help transform photonic design from a specialized, expert-driven process into a more predictable engineering workflow.
“Better design automation does not replace domain experts,” he says. “It amplifies them by moving repetitive design iterations into software and giving experts more time to focus on architecture, products and foundational innovation.”
Connecting research and education
A hallmark of the CAREER Award is an emphasis on integrating research and education. Gu built that guiding principle directly into his project.
“The research and education are integrated through open-source tools, hands-on learning and new curriculum development,” Gu says. “The same infrastructure we build for research can help students learn how physics, AI, algorithms and semiconductor design come together in electronic-photonic systems.”
The award funding will support graduate students working on highly interdisciplinary EPDA research projects. Students will gain experience in photonics, AI, optimization, chip design and system modeling — a combination of skills critical for the semiconductor and AI workforce.
Gu says these experiences will help prepare students for careers in industry where expertise across multiple disciplines is increasingly valuable.
“The project also provides opportunities for us to collaborate with experts in key disciplines,” he says. “These working relationships help us stay connected to real-world challenges while developing forward-looking solutions for both near-term needs and future systems.”
A foundation for the future
The CAREER Award also provides resources for Gu’s research group to pursue long-term goals that extend beyond this project. He says the research could help address some of the most significant challenges facing next-generation computing systems, including rising energy consumption, increasing bandwidth demands and growing design complexity.
“Over time, I hope this work will help establish a foundation for the EPDA community and inspire new ideas,” Gu says. “This means not only open-source tools, but also common design representations, benchmarks, and workflows that enable researchers to define new problems, compare ideas, and build on each other’s work more easily.”
