Bridging the gap between academia and industry in semiconductor innovation
gf.com, Jul. 22, 2025 –
Behind every technological breakthrough is cutting-edge research and development happening on the grounds of universities and research institutions. Semiconductors are no exception as the relentless demand to optimize performance, maximize power efficiency and reduce costs requires constant innovation at the intersection of science and engineering.
In fact, these academic-industry collaborations have deep roots, dating back to the 1940s. At Purdue University, physicist Karl Lark-Horovitz led pioneering work on germanium crystals, advancing rectifier technology for radar technology and laying critical groundwork for the invention of the transistor.
Through its University Partnership Program, GlobalFoundries is bridging this gap between academia and innovation to power the next wave of chip innovation. GF’s university network is comprised of collaborations with over 80+ universities, 110+ professors and 600+ students at leading institutions across the globe, driving innovation to push the boundaries of what is possible in semiconductor R&D.
Visionary research born in Princeton’s lab
A shining example of this comes from Kaushik Sengupta, Professor of Electrical and Computer Engineering at Princeton University. Within the four walls of the laboratory, Professor Sengupta and his intellectually diverse team of PhD and post-doctoral students are all in on next-generation, state-of-the-art wireless sensing communication. Their groundbreaking efforts in enabling the first AI-enabled radio and millimeter-wave frequency chips earned them the prestigious IEEE IMS Advanced Practice Paper Award back in 2022 [1] and the Best Paper Award in 2023 in IEEE Journal of Solid-State Circuits [2].
With this technology at the heart of critical applications including automotive radars, autonomous systems and robotics, Sengupta and his team are dedicated to researching the future of intelligent environments and the wireless interfaces behind the scenes that enable these advancements. Before the rise of artificial intelligence tools like ChatGPT to the mainstream, this team was working on AI-enabled radio frequency integrated circuits (RFICs) poised to revolutionize the industry.
The leap of AI in redefining RF circuit design
Traditionally, designing these circuits has been an art, requiring extensive experience and iterative design processes that can take several months. “RFIC design sits at the intersection of circuits and electromagnetics. As you have to jump between these multiple dimensions, it becomes a really complex design process,” explains Sengupta. On top of that, these circuits operate at very high frequencies, making even the smallest parasitic elements extremely significant.
Inspired by the advances in AI as seen in other scientific fields like protein folding, Professor Sengupta and his team set out to accomplish how they could change this paradigm by leveraging AI to algorithmize the design process. By training custom AI algorithms on curated data sets, they hoped to identify new, undiscovered RF circuits and electromagnetic passives and allow rapid end-to-end design of RFICs, reducing time of design from months to days.
The GF chips turning research into reality
This is when Professor Sengupta’s research team approached GF with their visionary idea. GF recognized the novelty of this approach and supported the research by supplying its best-performing silicon germanium technologies, especially as Professor Sengupta’s AI-driven methodology for RFIC design complemented the efforts of GF’s Reference Design team, which is deeply focused on applying AI and ML to accelerate design productivity and enhance the quality of next-generation RF solutions.
Across the world through the GlobalShuttle Multi-Project Wafer Program, GF is empowering startups, researchers and system innovators to bring differentiated chip designs to life more efficiently and affordably. By aggregating multiple projects onto a single wafer, GlobalShuttle lowers the barriers to custom silicon with scalability and flexibility, allowing partners to bring their design visions to life while avoiding the cost-limitations of test silicon. This program enabled the researchers at Princeton to demonstrate the feasibility of their design concepts and secure grant funding for their continued efforts.
The AI-created circuit designs break new ground
The results were nothing short of revolutionary, as the team successfully fabricated the first deep learning-enabled high-frequency transmitter system using GF’s silicon germanium 9HP platform. The AI-designed circuits feature extremely complex structures that defy conventional understanding of the field. “The electromagnetic structures that came from these AI algorithms looked like very complicated QR codes,” said Professor Sengupta. By looking at it, nobody can tell what it does. However, once you add those circuit elements, the entire circuit works exceptionally well. “What this does is universalize RF circuits and RF passives. All we now have to figure out is how these active devices are connected with passives.” Since then, the group has demonstrated several advancements demonstrating AI-enabled synthesis of multi-port structures, passives, antennas [3] and even end-to-end power amplifiers with both circuit and passive co-design [4].
As a result of these efforts, the viability of these AI-designed circuits has drawn significant attention from academic, industrial and government counterparts. The project has led to numerous publications, and more notably, was selected as one of three “AIDRIFC” awardees to receive $30 million in funding from the National Semiconductor Technology Center (NSTC). What started as an intellectual curiosity has now become a research field of its own, with several leading research groups demonstrating state of the art AI-enabled RFICs.
Recognizing the people powering the future of the industry
The success of Professor Sengupta and his team is just one example of how meaningful collaboration between academia and industry is accelerating innovation in semiconductors. This isn’t just a story about providing researchers with the tools and support they need to innovate. It’s a spotlight on the people co-innovating to push the boundaries of what is possible.
Today, circuit design isn’t an isolated discipline. It intersects with packaging, mechanical engineering, chemical engineering, algorithm signal processing and more. As with many other fields in the semiconductor industry, circuit design has grown to be a diverse and multi-disciplinary field. Continued advancement requires the cultivation of a strong pipeline of talented individuals who will drive the industry forward.
As GF’s University Partnership Program continues to grow, it’s not only powering groundbreaking innovations – it’s shaping the next generation of engineers to tackle the challenges of tomorrow.
References:
[1] Z. Liu, E. A. Karahan and K. Sengupta, “Deep Learning-Enabled Inverse Design of 30–94 GHz Psat,3dB SiGe PA Supporting Concurrent Multiband Operation at Multi-Gb/s,” in IEEE Microwave and Wireless Components Letters, vol. 32, no. 6, pp. 724-727, June 2022, doi: 10.1109/LMWC.2022.3161979
[2] E. A. Karahan, Z. Liu and K. Sengupta, “Deep-Learning-Based Inverse-Designed Millimeter-Wave Passives and Power Amplifiers,” in IEEE Journal of Solid-State Circuits, vol. 58, no. 11, pp. 3074-3088, Nov. 2023, doi: 10.1109/JSSC.2023.3276315.
[3] Karahan, E.A., Liu, Z., Gupta, A. et al. Deep-learning enabled generalized inverse design of multi-port radio-frequency and sub-terahertz passives and integrated circuits. Nat Commun 15, 10734 (2024). https://doi.org/10.1038/s41467-024-54178-1.
[4] J. Zhou, E. A. Karahan, S. Ghozzy, Z. Liu, H. Jalili and K. Sengupta, “25.3 AI-Enabled Design Space Discovery and End-to-End Synthesis for RFICs with Reinforcement Learning and Inverse Methods Demonstrating mm-Wave/sub-THz PAs Between 30 and 120GHz,” 2025 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 2025, pp. 1-3, doi: 10.1109/ISSCC49661.2025.10904600.