On the 16th at the company's headquarters in Hwaseong, Gyeonggi, Jeon Ho-yeon, head of Itta Semiconductor, stressed that the way chips are designed must change fundamentally. Jeon, who worked for 16 years on system-on-chip (SoC) design at Samsung Electronics, said, "Engineers spend more time on documentation and meetings than on actual design," adding, "We need to innovate the design process itself to raise productivity."
Itta Semiconductor is a chip design automation startup founded in 2022. It is developing a design platform that automatically generates code, design documents, and various deliverables when a chip designer draws a block diagram. The goal is to realize "no-code chip design," in which the diagram itself becomes the starting point of design instead of the traditional code-centric approach. Last year, it raised a pre-Series A investment of 3 billion won from We Ventures, Bluepointpartners, and L&S Venture Capital, and is currently a team of about 20 employees advancing power and clock design automation solutions.
Jeon said the biggest inefficiency in current chip design is the structure in which code and documents are managed separately. "Whenever a design changes, the process of rewriting documents leads to repeated mismatches between the actual design and the documents," Jeon said. "If design and deliverables are generated simultaneously, you can reduce errors and manage development schedules more reliably." Jeon added that at an actual client, a data verification task that used to take months was completed within a few days.
Jeon said the importance of design automation will grow in the AI Semiconductor era. "Generative AI is being adopted quickly into design workflows, but AI also has to be grounded in accurate design information to be used properly," Jeon said. "People will define the requirements, AI will optimize the design, and the platform will provide accurate design information."
Itta Semiconductor currently supplies power and clock design automation solutions and plans to expand its lineup to bus, DFT, and more. "Ultimately, our goal is to become part of the chip design infrastructure," Jeon said. "We will automate repetitive system design and create an environment where engineers can focus on chip architecture and algorithm development." The following is a Q&A with Jeon.
― What is the biggest difference from the conventional chip design approach?
"Chip designers do not think of code first; they understand the chip through block diagrams. But traditionally, people wrote the code first and then created the documentation, which often led to discrepancies between the actual design and the documents. In our approach, when you draw the diagram, the design, documentation, and various deliverables are generated at the same time."
― What benefits can clients expect?
"You can reduce design errors and manage development schedules more reliably. At an actual client, a data review that used to take months before handing off to a design house was completed within three days."
― How do you think Generative AI will change chip design?
"AI use will keep expanding, but it will be hard to fully replace people in the short term. People will define the requirements, AI will coordinate the design, and our platform will provide accurate design information that AI can use."
― What do you see as the biggest challenge for Korea's chip industry?
"It's the market, not the technology. The design capabilities of domestic engineers are sufficient, but there are few opportunities to build market experience. The market has to grow for the technology to advance with it."
― What are your future goals?
"Our goal is to become essential chip design infrastructure corporations. We want to automate repetitive system design and help engineers focus on more creative design and algorithm development."