Samsung aims to cut semiconductor material development from 5 years to 2 using AI

Samsung is applying AI to one of the slowest stages of chip progress: discovering and qualifying new materials. Yonhap reports the company aims to cut development time from five years to two, a timeline compression that could matter as much for future nodes as lithography advances.

The targets are specific. Samsung wants a fivefold increase in the number of promising material candidates surfaced and an 80% reduction in physical experiments. The mechanism is the familiar AI-for-materials loop: models predict properties of candidate compounds, rank them, and send only the most promising to the lab, with results fed back to improve the model. Samsung says the approach could extend to battery materials.

The effort extends Samsung's broader AI push, including on-device reasoning research such as its CG-TTRL framework and the physical-AI roadmap it outlined at Samsung AI Forum 2026. It also mirrors this week's AI-for-science surge, from Anthropic's drug-discovery partnerships to OpenAI's math claims.

Competitors including Google DeepMind (with its GNoME materials work) and Microsoft have published materials-discovery models, while TSMC and Intel invest heavily in materials R&D. Caveat: these are targets, not results, and materials qualification for production involves reliability testing that AI cannot shortcut. Watch for any disclosed materials entering Samsung's process roadmap.