Technical notes on fab planning
Written by the founder while building aPlanner. One article is published; the next topics are listed below without dates.
Yield-aware scheduling: why Excel breaks at 100 process steps
Reentrant flow, a tool × recipe × chamber capacity matrix, batch tools, cycle time as a distribution, yield that arrives six weeks late, a moving bottleneck, overlapping node transitions and hot lots — eight reasons a spreadsheet cannot hold a fab, and the three design decisions that follow when yield lives inside the optimizer.
In preparation
Topics we are writing. They are published after the December 2026 OASIS-6 presentation; no dates are promised before then.
Node-migration planning: 176L → 236L → 300L+ without demand gaps
Why generic tools treat a new node as a new product, and how learning curves, yield ramp and shared tools must be planned as one system.
Explainable re-planning: frozen, flex and far zones in practice
How anchored stability zones keep a re-plan from rewriting the whole fab, and what a planner needs to see before approving a change.
What a Monday morning looks like with a living plan
A walkthrough of the MVP use cases on synthetic data: furnace down, hot lot, yield excursion, process tuning.
Why Korea first: data residency, the Korean systems landscape, Korean support
The three questions a Korean fab asks an overseas planning vendor, and why we are building the company in Korea.
Want the next article when it is out?
There is no newsletter yet. Send a one-line e-mail to the founder and you will get the next article and the prototype walkthrough recording when they are published.
danniel@amoza.aiSee the MVP behind the articles
The prototype preview shows the screens on synthetic data; the design-partner program explains how a Korean fab can run them on its own line in 2027.