Your Fab's AI Goes Blind on the 2nm Node — Here's Why
The AI That Finds Every Defect — Except on the Node That Cost You Billions
This one has not been published to the channel yet. Subscribe on YouTube and it will turn up there.
GARTNER 2026
- Your fab's new AI spots defects better than any engineer — and on the brand-new node it goes almost blind.
- Not because the model is bad, but because every one of these systems learns from history: thousands of wafers, labelled defects, known-good outcomes.
- This video breaks down the cold-start problem in yield ramp — why AI inspection and virtual metrology deliver on mature lines while struggling exactly where the economics are worst, what the first six months of a ramp actually feel like from the yield engineer's seat, and the one approach that's genuinely closing the gap: transfer learning, which carries what a model learned on your existing tools across to the new one and cuts the modelling cycle from weeks to days.
- Plus the numbers — Samsung's 2nm wins set to more than double this year, and Gartner's 2026 finding that only one in three MES vendors can show a live production AI use case.
Why this matters
A mature line has years of that. A node you just spent billions building has none.
Starting anything new? Fix the cold-start problem on day one. Paste into Claude or Copilot: "I'm starting X next week with zero historical data. List the 8 signals I should log from day one so that in 90 days I can show a baseline, spot drift, and attribute a change to a cause. Then name the two people usually forget." Run on a new support queue it flagged the two nobody logs — freeze the category taxonomy on day one, and record arrival time separately from created time.
- 0:00The AI that goes blind
Six months into a ramp — what did you actually trust first, the model or the engineer who'd seen it before?
Argue with me on LinkedIn- Gartner 2026 MES Market Guide (1 in 3 vendors with a live production AI use case)
- SemiEngineering 2026 (cold-start — insufficient historical data at new process / equipment ramp; transfer + few-shot learning cuts the modelling cycle weeks→days)
- DIGITIMES 2026-08-03 (Samsung 2nm project wins expected to more than double in 2026).
Full transcript, 288 spoken words
Keep reading
All pieces
imec Says Fabs Can't Measure Every Wafer. So 95% Are Predicted.
In a fab running virtual metrology, only one to five percent of wafers are physically measured. The rest get a number anyway — predicted by a model that reads the process tools rather than the wafer: gas flows, RF power, chamber pressure, roughly a thousand sensor variables per run, turned into a thickness, a line…
AI Now Lets Chip Fabs "Measure" Wafers They Never Touch (Virtual Metrology)
Chip factories physically measure only about 3 wafers in 100. For the other 97, an AI called virtual metrology predicts the exact dimension straight from the process tool's own sensor "fingerprint" — pressure, power, gas flow, hundreds of signals per wafer — in milliseconds, with no measurement step and no throughput hit.
OpenAI Just Published the Number Chip Fabs Stopped Trusting in 1986
In this Fabspeak episode: • The figure OpenAI published on 6 September 2026: 3.1 agent-workdays per human workday, an agent-workday being 8 hours of machine runtime; the median researcher using over $600 of inference a day, the top 10% over $7,000. • OpenAI's own caveat: a runtime ratio, not a productivity multiplier.