St. Thomas faculty are developing a graduate micro-credential program that will study AI innovation for semiconductor manufacturing, targeted to begin in pilot form this fall.
The three semiconductor manufacturers signed on to provide data and industry expertise for the program — Polar Semiconductor, Seagate Semiconductor and Skywater Technologies — met with program organizer Kundan Nepal Monday for a kickoff meeting.
“There are people who know how to do semiconductor manufacturing, there are people who are now trained in AI (machine learning), but there’s a disconnect between those two,” Nepal said.
Over $10 million will be allocated to the 5-year program, officially titled the Minnesota AI Semiconductor Hub. According to Nepal, the hub will serve a dual purpose: workforce development for students, and manufacturing advancements for the companies.
The Crest did not receive a response to requests for comment from any of the three companies before this article’s publication.
Nepal said that the program, which will be made up of two courses, will be offered to employees of the three manufacturers first in order to expand their versatility in the workplace. After that, he said that graduate and upper-level undergraduate students would be able to enroll and apply for associated scholarships.
In either case, Nepal, who is also an electrical and computer engineering professor, said that the courses will be taught by industry professionals whom the university will seek out.
“We want to make sure that the people who are actually experts on the manufacturing floor with AI embedded in it are the ones who are also teaching it, so that they can bring in very much first-hand information to the students taking it,” Nepal said. “I think a lot of our graduate coursework does this very well in the school of engineering; we want to keep that tradition alive.”
The Minnesota Forward Fund provided $5 million for the hub’s scheduled five-year timeline, which Nepal said will mainly go toward cloud infrastructure and AI training. The leftover cost will be covered by the companies, who will be committing almost $6 million in the form of time spent using expensive metrology machines that can interpret data from all three companies’ past manufacturing.
Nepal said the high cost of procuring data on computer chip manufacturing means the program will spend its first year solely planning for what kind of data to collect and how to do so. For context, Nepal said that using one of the needed data collection machines for an hour can run up costs “close to seven figures.”
Coordinating the rights to that data between all parties was a lengthy process. Nepal said that contract negotiations with lawyers have taken the last six months, and even now he said, “I’ll believe it when I see the actual data flowing.”
Nepal said that while industry scientists will be studying the generated data, the hub also provides a key training opportunity for students.
“This would be a perfect example: a graduate student using the data and working on coming up with a new algorithm for their master’s project thesis,” Nepal said. “So it’ll have lots of opportunities for our students to engage directly with the real work, real-world data.”
Kevin Lynch can be reached at lync1832@stthomas.edu.
