Which Semiconductor Stocks Survive the Cycle, and Which Ones Get Crushed

Why the cycle persists, decade after decade.

The  industry has been cyclical since it began in the 1950s, and that pattern has not gone away.

Based on Morris Moore’s analysis, the gap between major downcycles is usually 3 to 4 years. Demand surges, capacity expands with a lag, and the mismatch creates the swing.

That cycle has repeated for decades. The industry has already weathered more than 15 downcycles while still growing towards an estimated US$1 trillion in revenue by 2030.

So the long-term growth story is real. But the path is anything but linear.

What triggers an upswing cycle

Each cycle has its own trigger, but the mechanics are usually the same. Demand runs ahead of supply, companies add capacity, and by the time that capacity arrives, the market often looks very different.

That is what makes chips so volatile. Fabs take years to build, the market is dominated by a small number of major players, and the capital intensity is enormous.

A new fabrication facility can take years to bring online, while the equipment inside it can depreciate quickly as chip technology advances. That matters for investors because what looks cutting edge today can become less competitive within a few years.

This is also why the current AI capex boom needs to be watched carefully. The Mag 7 are spending hundreds of billions of dollars on data centres, chips and infrastructure, but those assets will not stay fresh forever. As new architectures and more efficient technologies emerge, today’s infrastructure will eventually need to be upgraded, replaced or written down.

There are two key drivers of semiconductor downcycles

The first is the capacity cycle. Fabs make investment decisions years before mass production begins. We are seeing that now with TSMC’s expansion into Arizona and the broader push to bring more chip manufacturing capacity into the US.

The second is the inventory cycle, often called the bullwhip effect. Small changes in end demand can create much larger swings up the supply chain. When customers over-order during a boom, suppliers ramp production. When demand slows, that inventory has to be worked through, and the downturn can become sharper than expected.

Five resets in three decades

We have laid out the major peaks and troughs of the modern semiconductor era below. Each cycle had its own story, including which segment cracked first, which companies survived, and which ones were caught too far out on the risk curve.

Internet infrastructure overbuild meets reality

During the dot-com bubble, internet and telecom infrastructure buildouts drove a major surge in chip demand through the late 1990s and early 2000s.
Telecom equipment companies invested more than US$500 billion laying fibre optic cables, building wireless networks and provisioning switches. Chipmakers raced to expand capacity into what looked like a structural demand boom.

The unwind was brutal.

The SOX fell 82% from peak to trough. Intel issued a severe revenue warning in September 2000 and dropped 40% in a single day, triggering contagion across the semiconductor supply chain.
Valuations told the same story. The SOX traded at a price-to-book ratio of 8.8x at the June 2000 peak. By September 2002, the multiple had collapsed 80% to 1.5x.

When capacity is built for perfect demand, even a small reset can cause serious damage.

The shortage that became the glut

In 2021, the pandemic created one of the most severe supply-demand imbalances the industry had seen in decades.

Remote work drove a spike in PC and laptop demand, while auto OEMs cancelled orders early in the cycle and later scrambled to secure supply. Pricing power swung sharply towards suppliers, and the market began treating the shortage as a permanent condition.
That was the danger.

The SOX price-to-book ratio reached 8.2x in 2021, its highest reading since the 8.8x peak before the dot-com bust. According to Synovus, both readings came before major drawdowns.
By that point, the next downturn was already being built into the system.

Memory write-downs and capex cuts

The reset arrived in 2022.

By Q3, the industry recorded its first full quarter of negative year-on-year contraction, and memory took the worst of the hit.

Memory revenues fell 30% across 2023, with prices only stabilising after inventories had been written down and producers cut capex.

The pressure did not stop there. PC and mobile processor revenues fell by more than 10% in 2023 as consumer electronics demand normalised from pandemic highs.
Equipment makers initially held up better, because their order books lag the broader cycle. But by Q1 and Q2 2023, the slowdown had started to reach them as well.

Survivors, casualties, and opportunists.

Micron Technology

Micron is one of the clearest winners during a semiconductor boom.

When DRAM pricing surges, the operating leverage is enormous. The stock has risen more than 700% over the past 5 years, which shows how aggressively the market rewards memory exposure when the cycle is moving in the right direction.

But the downside can be just as sharp.

Micron is highly sensitive to peaks and troughs because memory is one of the most cyclical parts of the semiconductor supply chain. When demand is strong, margins expand quickly. When demand softens, pricing can fall hard, inventories build, and earnings can reset faster than investors expect.

That makes timing critical. The best opportunities in Micron usually come after the bust has already been priced in, when the market has punished the stock and the next demand catalyst is starting to form.

Intel

Intel is an international stock we cover more frequently, and the investment case is more complicated.

The foundry business remains deeply negative, while the PC market is stable but still offering limited growth. At this stage, Intel’s pricing power is being driven mainly by its data centre segment, which is helping offset the losses from the fab buildout.

That support matters. But it also creates a clear risk.

If data centre demand softens, Intel could face pressure on both sides of the business. The growth engine would slow, while the foundry losses would still need to be absorbed.

This is why the recovery could be slower than the market expects. Intel has upside if execution improves, but the downside risk is still meaningful if the cycle turns against data centre demand.

ASML

ASML is a different kind of semiconductor cycle stock.

It is the only company in the world that manufactures EUV lithography systems, which are required for the most advanced semiconductors being produced today. That monopoly position makes ASML far more resilient through industry peaks and troughs.

Its downturn behaviour is different.

Memory and logic companies can face sharp pricing pressure, inventory corrections and customer cancellations. ASML is more exposed to order timing shifts. Customers still need the tools, but deliveries can move around depending on fab readiness, capex budgets and the pace of customer expansion.

That distinction matters for investors.

The recent warning around 2026 shipments being heavily skewed towards the second half of the year is a useful example. It does not suggest demand has disappeared, but it does show that timing risk is already emerging as customers manage fab schedules and readiness.

For us, ASML remains one of the highest-quality ways to play the long-term semiconductor buildout. But even the best-positioned company in the supply chain is not immune to cycle timing.