Electronics swings from 'cannot get parts' to 'cannot move stock' faster than almost any sector, and because components go obsolete, the glut does not just sit. It expires on the books.
Sources: Micron, Q2 FY2023 (SEC filing); McKinsey (2022).
During the 2021-22 shortage, buyers double-ordered and padded safety stock to secure allocation. McKinsey estimated 10 to 20 percent more chips were ordered than actually needed, pure bullwhip. When demand cooled, the excess landed hard. One major chipmaker took a $1.43 billion inventory write-down in a single quarter. That is what happens when a whipsaw in demand meets a product that does not age gracefully.
In most sectors, excess stock is cash sitting idle that you can eventually sell. In electronics, components are superseded by faster, cheaper, newer parts on a relentless cadence. A chip overbought this year may be worth a fraction next year, not because it spoiled, but because the market moved on. So excess and obsolete, E&O, is not a slow leak here; it is a cliff. The longer the wrong stock sits, the less it is worth, which is exactly why catching it early matters more in electronics than almost anywhere.
Semiconductors and many components carry long, variable lead times, so orders are committed well before the demand is known. That gap is where the bullwhip is born: uncertainty makes everyone over-order to be safe, and the sum of all that caution upstream becomes the glut. The only real defence is a demand signal good enough that you do not have to pad it heavily to feel safe.
A single component often feeds many finished products, which is good for pooling risk and bad when the forecast for one end-product is wrong and ripples across all of them. Planning at the component level, against reproducible end-product demand, is what keeps commonality an advantage rather than a shared point of failure.
The firms that avoided the worst of the write-downs did the unglamorous work: reproducible forecasting so orders are not padded out of fear, per-SKU and per-component buffers sized on real variability and lead time, and an early pre-obsolescence view that flags stock losing value while it can still be used or sold. Long-lead components are planned on the real horizon rather than a defensive guess. It is the deterministic, auditable approach DemandIntel brings to electronics and hardware makers and distributors, on your own data, beside your ERP.
Long lead times plus fear of missing allocation make buyers over-order during shortages, the bullwhip effect, estimated at 10 to 20 percent excess ordering. When demand cools, that padding reverses into a glut, as the $1.43 billion single-quarter write-down at one chipmaker shows.
Components are superseded quickly, so excess stock loses value fast whether or not it is ever used. That turns a slow leak into a cliff and makes catching pre-obsolescent stock early far more valuable than in slow-moving sectors.
Forecast reproducibly so orders are not padded out of fear, size buffers per SKU and component on real variability and lead time, and flag stock losing value early so it is used or sold before it is written down.
A shared component pools risk but spreads a bad end-product forecast across everything it feeds. Planning at the component level against reproducible end-product demand keeps commonality an advantage instead of a shared failure point.
In electronics, excess stock does not wait for you to sell it. It loses value while it sits, which is why catching it late is the same as losing it.
Picture your next S&OP meeting: one number on screen that nobody argues with. See it free, on your own data, in 45 minutes; and if it does not hold up, we part as friends.