Samsung aims to cut chip-material development from five years to two with AI

Samsung Electronics said on October 26 it aims to use artificial intelligence (AI) to sharply compress the time needed to develop new semiconductor materials: from five years today to two years. The goal was set out by Choi Young-sang of the Samsung Advanced Institute of Technology (SAIT) at a Korea Industry Association Forum on AI in industrial research and development.

The bottleneck: too many combinations, too few experiments

Choi said semiconductor materials consist of multiple thin-film layers, creating more than 100 million possible combinations for a seven-layer structure, while researchers can run only about 50 experiments a month. Under the conventional approach, which relies on researchers' experience and intuition, identifying candidate materials can take months or even years, and data from failed experiments has not been fully used.

The targets in numbers

Samsung's targets are threefold: cut the development period from five years to two; increase the number of promising material candidates identified by more than fivefold; and reduce the number of experiments needed to find optimal conditions by more than 80 percent. In other words, AI is meant less to speed up a single experiment than to narrow the search first, so fewer trials are wasted.

Beyond chip materials

Samsung said the approach could also lower development costs and be extended to other industries, including battery materials. For Samsung, it is a move to take AI from the product side into the research side: with demand for advanced processes and AI chips running high, the speed of materials research itself has become part of competitiveness.