As generative artificial intelligence (AI) technologies are beginning to be used in the semiconductor industry, the proportion of AI agents (assistants) is increasing across various areas including design, manufacturing, and verification. Although it is still in the early stages, industry experts analyze that the evolution from Level 1 (L1) to Level 5 (L5) is progressing rapidly. As it evolves to L5, AI is expected to replace more of the engineer's role, which could reduce the corporations' dependence on the capabilities of engineers.
According to industry sources on the 26th, Samsung Electronics, SK hynix, Qualcomm, and Broadcom are actively utilizing AI in their mobile chip research and development (R&D) processes and are currently reported to be using L1 or L2-level AI agents. The AI assistants, also known as agentic AI, have advanced beyond mere repetitive tasks and are now actively analyzing, coordinating, and assisting in decision-making for the entire job.
A source familiar with Samsung Electronics noted, "L1 is comparable to a junior engineer level in semiconductor design, and it can provide significant assistance in their work." Recently, higher-level AI assistant functions are also being utilized in design and manufacturing, such as generating scripts (a collection of codes written for specific functions) and manufacturing recipes (specific methods).
The increased utilization of AI in the semiconductor sector is due to the growing complexity of semiconductor design and manufacturing, which has expanded the demand for essential design automation (EDA) tools. While the roadmap for next-generation semiconductor development has accelerated, a shortage of skilled semiconductor design personnel is also a contributing factor. As the development cycle for advanced semiconductors like AI shortens, there is a need for faster design environments, but the number of engineers with high semiconductor design capabilities remains severely lacking.
A representative from the world's largest EDA company stated, "When I meet with clients, they always complain about the shortage of skilled semiconductor design personnel," adding, "As corporations like Synopsys and Cadence expand their AI-based EDA solution portfolios, AI is now actively utilized not only in system semiconductor design but also in memory design and manufacturing fields."
In the case of Samsung Electronics, it first began using Synopsys' AI design tools in mobile application processor (AP) design in 2021, and SK hynix is also expanding its AI design adoption. The two companies are known to be implementing AI design tools in high bandwidth memory (HBM), which is considered the battleground of the memory semiconductor market, for use in development and verification processes.
Anand Thiruvengadam, a senior director at Synopsys, commented, "Synopsys has its sights set on reaching the final form of AI design tools, L5, and if we achieve this, it raises questions about the necessity of junior-level engineers in the design process due to the high involvement of AI." He added, "Ultimately, the competitive edge in design technology will hinge on how well AI can be effectively utilized."