ASU researcher honored with Google award for AI-driven chip design

Vidya Chhabria receives funding and mentorship to push the boundaries of electronic design automation

Vidya Chhabria has been named a recipient of the inaugural Google ML and Systems Junior Faculty Award, which recognizes early-career faculty whose research is advancing the frontiers of machine learning and systems.

An assistant professor of electrical engineering in the School of Electrical, Computer and Energy Engineering, part of the Ira A. Fulton Schools of Engineering at Arizona State University, Chhabria’s research interests lie in computer-aided design, or CAD, for very large-scale integration, or VLSI, systems. Her work primarily focuses on physical design, optimization and analysis algorithms.

She is one of more than 50 assistant professors across 27 U.S. universities selected for the award by a distinguished group of Google engineers and researchers.

Chhabria highlights Google’s commitment to collaborating with academia and its recognition of research in leveraging machine learning for hardware design.

“Being recognized by Google via this junior faculty award is extremely rewarding, not just to me but to our entire group,” Chhabria says. “The research our group does focuses on developing electronic design automation tools, specialized software that aid in the design of computer chips behind everything from smartphones to data centers.”

Her research group develops electronic design automation tools, which are specialized software that aid in the design of computer chips used in everything from smartphones to data centers.

“Designing chips is complex, time-consuming and resource-intensive, and AI has shown enormous potential in addressing challenges of scale, automation and optimization in this area,” she says.

Stephen Phillips, professor of electrical engineering and director of the School of Electrical, Computer and Energy Engineering, notes that awards like this have an impact that extends to students as well.

“Our students have an amazing opportunity learning from faculty like Dr. Chhabria, whose research is not only impactful but also receives recognition from industry leaders like Google,” Phillips says. “It’s incredibly motivating for students to be guided by someone shaping the future of the field.”

In addition to the award recognition, Chhabria will receive $100,000 in unrestricted funding.

She underscores the importance of this resource in supporting research projects and enabling them to move faster and increase scale when using AI to automate the design of computer chips.

“Training AI for chip design is especially challenging because of the lack of open-source, industrial-scale chip designs makes it difficult to create and test AI models,” she says. “This award helps bridge that gap by creating collaboration opportunities with Google. Mentorship and access to industry perspectives — and possibly computational resources — help keep our work industry relevant and advance AI in chip design.”

The focus moving forward

As chips become more complex, traditional electronic design automation, or EDA, tools struggle with scalability and optimization. By advancing AI-driven EDA tools, her group at ASU is working on the next generation of chip design technology — tools that can use AI to build the hardware needed for future AI applications.

“This work positions ASU not only as a leader in microelectronics broadly, but also as a leader in EDA specifically, which is essential for shaping the future of the semiconductor industry,” she says.

Chhabria is excited about using large language models, or LLMs, as intelligent “agents” for chip design.

Agents can learn to perform design tasks automatically, improving efficiency and optimization while lowering the barrier to entry for non-experts. For example, in physical design, the stage of chip design where abstract circuits are turned into a real chip layout on silicon, the process is extremely complex, requiring careful balancing of speed, power, area and manufacturability.

“Developing autonomous AI agents for different challenges in physical design could be transformative,” she says.

Chhabria highlights that building such agents requires large amounts of high-quality, well-labeled training data, which is currently scarce in chip design. So, another major thrust of her work is using generative AI to create synthetic data.

Synthetic data generation fills critical gaps where open-source designs are in short supply, ensuring that academic research remains relevant and competitive.

“By creating realistic, large-scale datasets where real ones don’t exist, we can train AI models more effectively and make progress even in areas where access to industrial data is limited,” she adds. “Together, these two directions, AI agents for design and generative AI for data, are helping redefine what’s possible in the future of EDA.”

For society, the work to generate faster and more efficient chip design translates to technology that is cheaper, greener and available more quickly — enabling advances in fields as diverse as health care, transportation and energy.