Protein AI: next big thing after physical AI, eyes 70兆円 market

Protein AI is framed as the latest step in a progression of generative AI targets: natural language, program code, images, video, and now proteins. What makes the field commercially notable is the size of the prize it points at — a protein-drug market estimated at roughly 70兆円 — and the fact that drug discovery is already beginning to apply the technology.

Overseas startups are described as leading the field, with promising companies emerging from Big Tech and venture capital. Japan's position is more recent: two startups, MOLCURE and Repertoire, are said to be driving domestic protein AI, and major drugmakers 中外製薬 and アステラス製薬 are focusing on the area. MOLCURE is described as developing its own antibody-focused large language model trained partly on failure data, while Repertoire combines protein language models with machine learning rather than relying only on the newest AI. Electronics makers and cloud providers are also moving in, suggesting the field could become the next outlet after physical AI.