Why Is AI Finally Sticking Inside the Fab?

By MAHIMA SAMBRE, Acumen Research and Consulting

Fabs never bought the pitch of a single, fab-wide AI platform, writes Mahima Sambre of Acumen Research and Consulting. What they are funding instead are narrow, deterministic tools, such as a defect classifier on one inspection tool or a virtual metrology model on one deposition step, deployed wherever yield loss is most expensive.

The market numbers support the shift. Acumen values AI in semiconductor manufacturing at $6.5 billion in 2025 and projects $26.8 billion by 2035, a 15.2% CAGR, growing roughly twice as fast as the process control hardware it runs on. Asia Pacific holds 55% of the inspection systems market, and one of the most valuable uses of AI is capturing the tacit knowledge of senior process and yield engineers before they retire.

Sambre also looks at India, which is writing its process control playbook into greenfield fabs such as the Tata Electronics project at Dholera, and at emerging issues including export controls on where process data can be processed, the power cost of running AI inference on the tool, and acquisitions of metrology and inspection software teams. None of it, she concludes, adds up to AI running a fab on its own anytime soon.

Click here to read the full article in Semiconductor Digest magazine.

The post Why Is AI Finally Sticking Inside the Fab? appeared first on Semiconductor Digest.