Memory chips are the parts of a computer, phone or server that store data, either while a device is running or for the long term. The industry that makes them is famous for its boom and bust pattern, often called the memory chip cycle. Prices can climb sharply when supply is tight and then slump when new capacity floods the market. With chip shares swinging hard this week, understanding the cycle helps make sense of the headlines.
The two main types of memory
There are two broad families of memory chip:
- DRAM, or dynamic random access memory, holds data that a device is actively using. It is fast but loses its contents when the power is switched off.
- NAND flash stores data for the long term, as in phone storage and solid state drives, and keeps it without power.
A newer and closely watched product is high bandwidth memory, often called HBM. It stacks DRAM chips together to move huge amounts of data quickly, and it has become an important part of the processors used for artificial intelligence.
Why the industry runs in cycles
Memory chips are, to a large extent, standardised products. One maker’s chip can often replace another’s, which means price is a big part of competition. At the same time, building new factories costs a great deal and takes years.
That combination tends to create a pattern:
- Demand rises, perhaps because of a new wave of phones, PCs or data centres.
- Supply cannot keep up, so prices rise and profits jump.
- Manufacturers invest heavily in new capacity.
- That capacity arrives, often just as demand cools, and prices fall.
- Companies cut spending and output until the market tightens again.
The result is a cycle where profits can swing from very strong to very weak within a short period.
Who makes memory
The market is concentrated. Samsung Electronics and SK Hynix in South Korea and Micron Technology in the US are the main DRAM makers, and they are also large players in NAND alongside a few other firms. Because so few companies dominate, decisions by any one of them about investment or output can move the whole market.
This is different from the model described in our guide to a semiconductor foundry, where a contract manufacturer builds chips designed by other companies.
What AI changes, and what it does not
Demand for AI servers has boosted interest in high bandwidth memory and other high-end products, and many investors argue this makes the cycle less severe than in the past. That may prove true, but it is a debate rather than a settled fact. Traditional markets such as phones and PCs still matter, and heavy investment in new capacity could, in time, recreate the familiar pattern of oversupply.
Why markets react so sharply
Memory company shares often move ahead of the actual change in profits, because investors try to anticipate where the cycle is heading. Signals they watch include:
- Contract and spot prices for DRAM and NAND.
- Inventory levels at chipmakers and their customers.
- Spending plans for new factories and equipment.
- Comments from major buyers such as cloud computing firms.
Higher interest rates can add pressure too, since much of the value of fast-growing technology companies sits in expected future profits. Our explainer on how discount rates shape semiconductor valuations covers that link.
Common mistakes to avoid
- Assuming record profits mean the good times will last. In a cycle, peak profits often arrive close to the turn.
- Treating all chip companies as the same. Memory, processors and equipment makers follow different drivers.
- Reading one month of price data as a new trend.
Bringing it together
The memory chip cycle is the repeated swing between shortage and glut in DRAM and NAND, driven by changing demand and slow, expensive supply. AI has added a powerful new source of demand, but the basic economics of the industry have not disappeared. Watching prices, inventories and investment plans gives a clearer view than headlines alone.
If you want to test how well you read sector stories like this, take our free trader assessment.
