Materials research has always been a marathon run in years: developing a new semiconductor material from concept to usability traditionally takes about five years. On October 26, Samsung Electronics announced an aggressive target — using artificial intelligence to compress that cycle to two years.
How AI Rewrites the Screening Process
According to Samsung's presentation at an industry forum, AI in the R&D workflow is expected not only to shorten the timeline but also to increase the number of viable candidate materials identified by more than fivefold.
These are two sides of the same logic. The most time-consuming part of materials research is trial and error: synthesize, test, reject, repeat. Machine learning models can learn structure-property relationships from existing experimental data and push the most promising candidates to the front of the queue, effectively replacing unproductive experiments with compute. Fewer wasted experiments shorten the cycle; the same budget then covers a much wider candidate space.
What It Means for the Chip Supply Chain
Materials sit at the very top of the semiconductor chain — the hardest part to substitute, with high research barriers, long validation cycles, and concentrated suppliers. If AI can turn five years into two, the payoff is not just cost savings but a knock-on effect on process evolution: faster material iteration shifts the upgrade windows for downstream process nodes.
For the broader industry, Samsung's statement is a clear signal that AI's impact on semiconductors is spreading from design (EDA tools) and manufacturing (yield optimization) upstream into materials. Once leading firms attach quantified AI targets to materials research, the method will quickly become an industry default — and competition in materials will pivot from lab capacity toward data accumulation and model capability.