EntryUpdated 2026/08/09

Why Did Cheap Old Chips Become a Security Problem?

Because the pain of losing supply has nothing to do with how advanced a chip is and everything to do with whether anyone else makes it. A mature-node chip may cost a few dollars, and without it a car plant stops — and that capacity is concentrating in a few places.

Read this first: Money Builds Fabs — So Why Can It Not Build a Chip Supply Chain?

The two are not points on one axis — so the same instruments do not work on both

Node	Where it goes	Who can make it	What losing it does	Do export controls bite?
Advanced nodesChasing speed and efficiency	Phone and server processors, compute accelerators. A few high-value products, concentrated in volume.	A handful of production lines, built with equipment from an even smaller set of suppliers.	Flagship products from a few firms slip or downgrade. Painful, but confined to particular industries.	Yes. The barrier is equipment, and equipment is physical, crosses customs, and comes from few enough sources to track shipment by shipment — every condition traditional control needs.
Mature nodesChasing stability, durability, long life	Vehicles, industrial motor control, grid inverters, medical equipment, appliances, drone flight controllers and power modules — spread across the physical economy.	Technically anyone: proven equipment, published processes, skills that are not scarce. In practice the number of makers is falling — and the cause is price, not technology.	The line stops. One missing microcontroller worth a few dollars keeps a car off the lot — the pain scales with substitutability, not with unit price.	Barely. There is no equipment barrier to hold. The competitor here is not technology but capital willing to absorb losses — and tariffs and controls do nothing against a price war.

Advanced and mature nodes relate like a truck and a sports car, not a new model and an old one. Ranking them on one axis is the standard misreading — and the cost of it shows up in the last column: export controls bite on the top row because the barrier is equipment, and barely touch the bottom row because there is no equipment barrier to hold.

The “what losing it does” column explains why the cheap parts deserve more scrutiny. The pain scales with substitutability rather than unit price — and cheapness is exactly what discourages anyone from tracing the source. Trace it and you often find two or three firms worldwide making that part.

Source: Impactful Creative, compiled from the division of labour described on this page and the glossary entries on mature nodes, advanced nodes, and foundries

Is a mature node just an obsolete node?

No. A mature node is a process generation that has been in stable volume production for years, and it is still in use mostly because the application does not need anything newer.

A chip driving a car window motor would, on the latest node, simply be more expensive, more power-hungry, and less tolerant of heat. Vehicles, industrial equipment, power management, and sensors want stability, durability, and long product lifetimes — exactly what mature nodes are good at. “Advanced” is not an advantage in these applications; it is a mismatch.

Advanced and mature nodes relate to each other more like a truck and a sports car than like a new model and an old one. Ranking them on one axis is the most common misreading of this subject, and where the rest of the errors begin.

Where do they actually go?

In almost everything that moves. A car contains hundreds to thousands of chips, the overwhelming majority on mature nodes; so do factory motor controls, grid inverters, medical equipment, appliances, and a drone’s flight controller and power modules.

That creates an asymmetry: the value of advanced nodes is concentrated in a few high-priced products, while the value of mature nodes is spread across the physical economy. Losing the first affects a handful of flagship products; losing the second affects production lines themselves.

The vehicle chip shortage after 2021 demonstrated exactly this. What ran out was not the most advanced compute silicon but low-cost microcontrollers and power management chips — and one missing part worth a few dollars keeps a whole car off the lot.

These chips have existed for decades — why is it a problem now?

Because the distribution of capacity is changing, and not through a technical breakthrough — through price.

Mature nodes have a low technical barrier: the equipment is proven, the processes are published, the skills are not scarce. That means anyone can build, and also that nobody can stop anyone else from building. When one side backs large-scale expansion with policy, added supply pushes prices down, and once prices fall below other fabs’ operating costs those fabs stop investing and drift out.

This is what overcapacity means geopolitically: on the surface a market phenomenon, in effect a mechanism that concentrates capacity in whoever can sustain low prices longest. It requires no unlawful act — only capacity and patience, the same mechanism as the price war in the critical minerals piece.

So the accurate version of “why now” is: the technology never changed. What changed is how many firms still make it.

How is this different from an ordinary shortage?

A shortage ends; concentration does not. A shortage is supply and demand temporarily out of step — prices rise, capacity follows, the problem closes. Concentration is a structural change that remains after prices normalise.

The remedies are opposite too. A shortage calls for short-term allocation: priority rules, extra shifts, temporarily relaxed specifications. Concentration calls for long-term demand commitments — nobody builds into a price war without a committed buyer, and that is not the same thing as a construction subsidy.

There is a simple test for which one you face: if prices rose thirty percent today, would new supply appear within six months? If yes, it is a shortage. If no, it is concentration. Mature nodes have been drifting toward the latter.

The foundry model makes it harder to see, incidentally: buyers deal in modules and finished goods, with packaging houses, module makers, and system integrators in between. By the time it is felt, it is usually the day the line stops.

Five steps, each of them an ordinary market outcome on its own

Policy-backed expansion at scaleStep 1 · still time
The technical barrier is low: proven equipment, published processes, skills that are not scarce. Anyone can build, and nobody can stop anyone else from building — export controls have nothing to grip here, because there is no equipment barrier.
Supply rises, prices fallStep 2 · still time
This step looks like an ordinary market outcome and buyers even benefit. It is also the one stage outsiders notice and widely read as good news — and demand-side instruments, long-term offtake and price floors, only work if applied here.
Other fabs stop investing and drift outStep 3 · turning irreversible
Once prices fall below operating cost, exit is rational — and it happens one firm at a time. No single exit is an event, so no single exit triggers a response. By the time it can be called a trend, most of it has already happened.
Capacity concentrates in a few placesStep 4 · only rebuilding remains
The structure has changed. Even if prices recover, the exited fabs do not return — re-entry means facing the same price war against a larger incumbent. This is where concentration parts company with a shortage: shortages end, concentration does not.
Dependence forms and becomes usable leverageStep 5 · nothing more is required
No further action is needed at this point. The mere possibility of tightening changes buyers’ behaviour — the same thing the sanctions piece describes as the secondary effect: what works is not the measure but the expectation of it.

Still time — demand-side instruments work hereTurning irreversible — exits happen one at a time, never as an eventOnly rebuilding remains — five to ten years

The whole path requires no unlawful act — only capacity and patience, the same mechanism as the price war in the critical minerals piece. What makes it hard to answer is that step three happens one firm at a time: no single exit is an event, so no single exit triggers a response.

The right-hand tags mark whether intervention is still possible. It is, for the first two steps; after the third only rebuilding remains, and rebuilding takes five to ten years. That also settles the choice of instrument: what works at step two is demand-side — offtake agreements and price floors — rather than construction subsidy, which addresses capital cost while the price war targets a decade of operating revenue.

Source: Impactful Creative, compiled from the concentration process described on this page and the glossary entries on overcapacity and mature nodes

What does this mean for Taiwan and for ordinary companies?

For Taiwan the question is quite unlike the advanced-node one. Advanced nodes are about staying ahead; mature nodes are about how many suppliers remain — and the competitor there is not technology but capital willing to absorb losses. That changes which instruments work: tariffs and export controls bite on the first and do little for the second, because the issue is not technology leaking but price.

For companies, the practical move is to split part numbers in two: those that are expensive and specialised, and those that are cheap and look available everywhere. The second group is the one to investigate — cheapness discourages anyone from tracing the source, and tracing it often reveals only two or three firms worldwide making that exact part.

This also completes the chip supply chain piece: that one maps the vertical division of labour at the advanced end, while this covers the overlooked segment of the same chain. The risks differ in kind, and so do the responses.

Terms used here

Mature Node — Older, standardized process generations with broadly available capacity, used across automotive electronics, industrial control, and appliances. Mature nodes stayed outside the control conversation until governments realized how much civilian and defense equipment depends on them — and how quickly that capacity was concentrating in a single source. Controlling them is hard precisely because they are everywhere, making end-use tracking impractical.
Advanced Node — The leading generation of chip manufacturing, dependent on equipment — extreme ultraviolet lithography above all — available from a very small number of suppliers. Advanced nodes are the primary target of export controls because they combine high technical barriers, extreme geographic concentration, and clear military relevance. Where the control line is drawn between nodes shapes the industry more than the existence of controls does.
Overcapacity — Output capacity persistently exceeding absorbable demand, usually the product of subsidy-driven investment rather than market signals. The spillover is prices sustained below replacement cost, which prevents unsubsidized competitors from funding their own investment cycles. It has recently escalated from a trade-remedy question to an economic security one, because what remains after competitors exit is a new single-source dependency.
Foundry — The model in which a manufacturer fabricates chips designed by others and sells no products of its own. Separating fabrication from design lowered entry barriers for designers, but concentrated the industry's manufacturing risk into a handful of sites. In resilience terms, diversity at the design layer does not offset concentration at the fabrication layer; they are distinct risks.