Display fabs: substrate size decides which products you can win

Substrate generation locks a fab into a product mix

Larger mother glass yields more panels per sheet for a given size, but only when the arithmetic works, and a plant sized for one panel dimension wastes glass on another. Because substrate handling, coaters, exposure tools and cleanrooms are all built around a specific sheet size, a plant cannot change generation. It can change product within the sizes its glass cuts efficiently, and that is the whole of its flexibility. Investment decisions therefore embed a bet on which panel sizes will sell across the plant's life, and a market moving elsewhere leaves it structurally disadvantaged.

Cut efficiency governs the economics of every order

Quoting a panel order begins with how many units come from one sheet after edge exclusion and process margin. A small change in requested diagonal or aspect ratio moves that count and shifts cost per panel sharply, which is why panel makers steer customers towards sizes their glass suits. Set makers who understand this specify around a supplier's efficient sizes and pay noticeably less. Those insisting on an awkward dimension for industrial design reasons pay for the wasted glass, whether or not the quotation makes that cost visible on its face.

Panel making and module assembly are separate factories

Producing the panel and turning it into a finished module are different activities. Panel work is a cleanroom process business with fab-like capital and yield behaviour. Module assembly attaches driver electronics, backlight units where needed, optical films, frames and connectors, behaving like precision electronics assembly with far lower capital and more manual content. Many companies place module work near their customers while keeping panel production where the fab investment sits. Buyers should establish which stage a supplier actually performs, since a module assembler and a panel maker face entirely different cost pressures.

Oversupply cycles here are structural, not accidental

Capacity arrives in very large blocks, often supported by regional industrial policy, and years pass between decision and output. Several such decisions taken during one strong market produce a wave of supply landing after demand has cooled. Because the plants carry high fixed costs, nobody idles them, so prices fall until the weakest producers stop investing. That pattern has repeated across successive display technologies. Buyers entering should expect price to fall across a product's life, and producers entering should expect at least one downturn inside their payback period. Supply agreements written during a shortage age badly, and both parties know it while signing.

The defect classes customers actually reject on

Customers reject on faults that a specification sheet handles poorly: bright or dark sub-pixels in visible positions, non-uniformity across the panel, colour shift with viewing angle, backlight leakage at edges, cosmetic marks beneath the cover glass. Acceptance is therefore governed by negotiated criteria covering how many defects of which type in which zone are tolerated. Those criteria matter more than headline specifications, because they set effective yield and therefore price. Experienced set makers negotiate them explicitly and audit incoming panels against them rather than assuming a datasheet settles anything. Grading rules deserve at least as much negotiating attention as the price per unit.

Frequently asked questions

Why can a display fab not simply make whichever panel size is selling?

Because the plant is built around one mother glass size, and panel dimensions that do not tile efficiently onto that sheet waste glass. The waste appears directly in cost, so a fab making an inefficient size is uncompetitive against one whose glass suits it. Changing substrate size means building a new plant rather than retooling. This is why capacity for particular panel sizes can be tight while other capacity sits underloaded during the very same period.

What should a buyer settle before agreeing panel acceptance criteria?

How defects are classified, how many are permitted in each zone of the screen, how uniformity and colour are measured and under what viewing conditions, and what happens to panels falling between grades. Also worth fixing: who inspects, on which equipment, and how disputes get resolved. These terms drive real cost more than headline price, since a loose specification with tight enforcement produces arguments and a tight specification with loose enforcement produces field returns.

Is module assembly worth bringing in house?

It can be for set makers with sufficient volume, since it shortens the chain, allows configuration close to demand and captures a stage with lower capital intensity than panel making. The requirements are real: clean handling, optical bonding capability where used, and disciplined cosmetic inspection. The panel itself remains the scarce input, so bringing module work in house does not improve bargaining position on panels, which is usually what buyers actually hoped it would achieve.