CluenexWhat is the memory chip cycle?
The memory chip cycle is the repeating pattern of shortage, expansion, glut and cutbacks in DRAM and NAND, the two main kinds of memory chip. DRAM holds data while a device is running. NAND holds it when the power is off.
One maker's standard DDR5 module does nearly the same job as another's. Because buyers choose mainly on price and availability, no producer can hold its price above the market level for long.
That lack of pricing power, combined with factories that cost billions of dollars, produces earnings swings that are rare in software, banking or consumer staples. Micron Technology's annual report for fiscal 2023 showed a net loss of about $5.8 billion on revenue that fell roughly half. Check the current filings for the latest figures, since the numbers move every quarter.
What are the four phases of the memory cycle?
The cycle has four phases, and each follows from the one before it.
- Shortage. Demand outruns installed capacity and contract prices rise. Because the cost of making a chip barely changes, almost every extra dollar of price becomes profit.
- Capacity response. Record profits fund record spending on new fabrication plants (fabs). Every producer sees the same price signal, so they all expand at once.
- Glut. The new capacity arrives together, often after the demand surge has cooled. Prices fall, sometimes below the cash cost of making the chip.
- Discipline. Producers cut output, delay equipment orders, write down inventory and post losses. Supply growth slows while demand keeps growing, and the shortage returns.
A wheat market works the same way. A good year prompts everyone to plant more, and the next harvest floods the market. A wheat farmer can respond in one season. A memory maker needs about two to three years to bring a new fab online, so the overshoot is bigger and the correction is deeper.
Why are memory profits so volatile? Operating leverage
Memory profits are volatile because most costs are fixed. Operating leverage is the share of a company's costs that do not change with sales. The higher it is, the more profit moves for each percent change in revenue, in both directions.
A fab's biggest costs are depreciation on the buildings and equipment, plus the engineers who run it. Those costs do not shrink when prices halve or grow when prices double.
A simple worked example shows the effect. Suppose a maker sells chips for $100 million with $80 million of mostly fixed costs, leaving $20 million of profit. If prices rise 25% with volume unchanged, revenue becomes $125 million and costs stay near $80 million. Profit rises to $45 million, up 125% on a 25% price rise. Run it backwards: a 25% price fall gives $75 million of revenue and a $5 million loss.
Micron's own filings show the real version. Its 8-K earnings releases for fiscal 2023 reported a negative gross margin and inventory write-downs, meaning chips made at a cost above their selling price. Its filings for fiscal 2026 show the opposite extreme. Same company, same product, a few years apart.
Why does a low P/E look cheap on a cyclical stock?
A low P/E on a cyclical stock often means earnings are at a peak, which is when risk is highest. The P/E ratio divides price by trailing earnings. At the top of the cycle, trailing earnings are at their highest just before they fall.
| Cycle position | Reported earnings | Trailing P/E looks | What is really happening |
|---|---|---|---|
| Peak | Record profit | Very low | Earnings are about to fall. Highest risk. |
| Downturn | Falling profit | Rising | Price already reflects some of the decline. |
| Trough | Losses | Negative or meaningless | Earnings are about to recover. |
| Recovery | Rising profit | Falling | Multiple shrinks as earnings catch up. |
For a steady business, a low P/E can signal value. For a cyclical, it mostly signals where the company sits in the cycle. Two common fixes are mid-cycle earnings (average earnings over a full boom and bust, such as five to seven years) and price-to-book, which moves far less than earnings. For more on the ratio itself, see P/E ratio explained and forward vs trailing P/E.
What should you check on a memory stock?
Check industry capital spending, contract pricing, customer inventory, product mix and the balance sheet, rather than one quarter's profit.
- Capital spending (capex). Capex is the leading indicator of future supply. Rising capex across the industry is how earlier gluts were funded. Read it in each company's cash flow statement in its 10-Q and 10-K.
- Contract pricing. Most volume is sold on quarterly or annual contracts. The direction of contract prices says more about the cycle than a single spot quote.
- Customer inventory. If customers hold large stockpiles, current orders may be restocking, not real use.
- Product mix. High-bandwidth memory (HBM), used with AI accelerators, is qualified per customer and supply is concentrated among a few producers. Standard DRAM and NAND are more commodity-like.
- Balance sheet. A cyclical has to survive the trough. Cash versus debt at the peak tells you how much of a downturn it can absorb. See how to read a balance sheet.
The semiconductor stock guide covers wafer starts and lead times in more depth.
Has AI ended the memory cycle?
No one knows yet, and the evidence points both ways. Arguments that the cycle is softer: HBM is hard to make, supply is concentrated, and some volume now sits under multi-year customer agreements that management teams say make results more predictable.
Arguments that it is not: earlier waves, such as PCs, smartphones and cloud, each produced the same claim, and each time capacity caught up. Producers spending more on new capacity is the standard precondition for the next glut. The reasonable position is that the cycle may be longer or shallower, not that it has been repealed.
See AI capex and revenue payoff for how to judge whether spending is paying off.
Where does this idea fail?
The cycle framework is a lens, not a timing tool. Common mistakes:
- Calling a single-digit P/E cheap. It usually marks peak earnings, not a bargain.
- Reading a large drop as failure. Drawdowns of 40% or more are routine for memory stocks. The better question is whether the balance sheet can fund losses through a trough.
- Buying because profits are booming. For a growth business that can make sense. For a cyclical, peak profit coincides with peak prices, peak capacity plans and peak analyst optimism.
- Assuming the cycle is on a schedule. Historical peak-to-peak spans of a few years are a pattern, not a calendar. Demand shocks and policy changes can shift timing.
- Treating one company as the whole industry. A diversified company can hide the cycle inside other divisions, or be dominated by it. Read segment results in the filings.
The framework also says nothing about which stock to own or when to act. Valuation, position size and your time horizon still matter. See value trap vs bargain.
What this means for you
If you own or are considering a memory or other commodity-driven stock, a few habits help.
- Look at five to seven years of earnings and free cash flow, not the latest quarter.
- Treat a very low P/E on a cyclical as a prompt to ask where earnings sit in the cycle.
- Follow industry capex and contract pricing in company filings and earnings calls.
- Check cash and debt in the latest 10-Q before assuming the company can ride out a bad stretch.
- Decide the size of the position in advance, because 40% swings are normal for this group. See position sizing and concentrated stock positions.
If you use a research tool, make sure it shows what a valuation rests on. Cluenex gives each covered stock a plain-English verdict with cautious, base and optimistic fair value cases, and some stocks get no number when no method values them reliably. The method is at how we value.