How to Analyze a Semiconductor Company

What it takes to analyze a semiconductor company

To analyze a semiconductor company, hold two facts together: it makes some of the most advanced products in the world, yet it behaves in many ways like a cyclical commodity business. Chips power almost everything, from cars to data centers, but demand for them rises and falls sharply, and the industry expands capacity in large, expensive steps that take years to complete. The result is a boom-and-bust rhythm that makes any single year a poor guide to the underlying business. Analyzing a chipmaker means reading it through the cycle, not at a point in it.

The second defining feature is the capital model. Some chip companies own the factories that make their products and carry enormous fixed costs; others design chips and outsource the manufacturing, staying light on capital but dependent on someone else's plants. These two models produce different margins, different risks, and different behavior in a downturn, so the first question about any semiconductor company is which kind it is.

Texas Instruments is a useful anchor because it shows both the cyclicality and the cost of owning factories. Its revenue dipped from about $17.5 billion in fiscal 2023 to $15.6 billion in fiscal 2024 before recovering to $17.7 billion in fiscal 2025, and its gross margin compressed from roughly 62.9 percent to 57.0 percent over the same span as its plants ran below capacity, according to its filings. The deeper walk through its strategy sits in how we analyze Texas Instruments; here it illustrates the metrics that apply to the sector at large.

Cyclicality: reading through the swings

The first discipline in semiconductor analysis is to treat the cycle as the backdrop to everything, and to refuse to judge the business on a single year. Chip demand swings with the products that use chips, and because new factory capacity arrives in big, delayed steps, supply and demand rarely match. When demand outruns supply, prices firm, factories run full, margins expand, and everyone invests. When that new capacity comes online just as demand softens, the same forces reverse, and profits fall hard.

This means a chipmaker's results can look their best right before a downturn and their worst right before a recovery, a pattern that mirrors the trap in energy and other cyclical sectors. A peak year of high margins and full factories is not proof of a permanently better business; it may simply be the top of a cycle. The disciplined approach is to look at results across several years, to normalize for where in the cycle you are standing, and to distrust extrapolation of either a boom or a bust. The wider framework for this sits in understanding business risks.

Gross margin is the clearest cyclical gauge, because it moves with factory utilization. In good times, high volume spreads fixed factory costs over more chips and the margin rises; in bad times, underused plants drag it down, as Texas Instruments' slide from the low 60s to 57 percent shows. Reading gross margin across the cycle, rather than at one point, tells you where in the swing a company sits, using the mechanics in gross margin.

Fab versus fabless: the capital models

The choice between owning factories and outsourcing them is the deepest structural fact about a semiconductor company, and it shapes the whole financial profile. A fab company, one that owns its fabrication plants, spends enormous sums building and equipping them, so it carries heavy fixed costs and heavy capital spending. A fabless company designs chips and pays a foundry to manufacture them, so it needs far less capital and can scale up or down more easily, at the cost of depending on the foundry's capacity, pricing, and priorities.

Each model has a distinct set of trade-offs, summarized below.

The capital intensity of the fab model shows up plainly at Texas Instruments, which owns its factories and has been building new ones. In fiscal 2025 its capital spending consumed roughly two-thirds of its operating cash flow, leaving free cash flow well below net income as it invested in future capacity, per Tenet data. That is the fab model's burden and its bet: heavy spending now in exchange for control of supply and lower unit costs later. Whether that spending pays off depends on the discipline behind it, which is why capital spending, examined in understanding capital expenditures (CapEx), deserves close attention for any fab owner. A fabless company would show the opposite profile, light on capital but exposed to a foundry it does not control.

Design-win stickiness

The quality that makes semiconductor revenue durable is the design win, and it is what separates a sticky franchise from a commodity supplier. A design win occurs when a chipmaker's product is chosen to go into a customer's device, be it a car, an industrial controller, or a piece of medical equipment. Once a chip is designed into a product, replacing it is expensive and slow, because the customer must redesign and re-test the whole system, so the original chip often keeps winning repeat orders for the life of that product, sometimes a decade or more.

This stickiness is strongest in areas like analog and embedded chips, where products have long lives and switching is painful, which is precisely where Texas Instruments concentrates. A company with tens of thousands of products designed into a broad base of customers enjoys revenue that is far more resilient than the cyclical headlines suggest, because each design win keeps paying out through cycles. A steady flow of new design wins is a sign that the franchise is renewing itself, while a slowing flow warns that future revenue may fade even if current sales look fine. This is a form of the switching-cost moat examined in identifying competitive advantages (moats).

Design-win economics also reward breadth. A chipmaker spread across many products, customers, and end markets is insulated from any single product cycle, whereas one dependent on a few large customers or one hot end market is exposed if that market cools. Breadth of design wins, like breadth of customers anywhere, lowers risk. It is worth checking how much of a company's revenue rests on its largest handful of customers, because a chipmaker that depends on one or two giant buyers can lose a big slice of its business overnight if a single customer switches suppliers or designs the chip out, however excellent the product was.

Inventory: the canary in the mine

Inventory is the semiconductor industry's earliest warning signal, and learning to read it can flag a downturn before it reaches the income statement. Because chips move through a long supply chain, from the maker to distributors to the final manufacturer, excess stock building up anywhere in that chain usually means demand is softening faster than orders reflect. That excess must be worked off before new orders resume, so an inventory build often precedes a revenue decline by a quarter or more.

The measure to watch is inventory turnover, the cost of goods sold divided by average inventory, and especially its trend. When turns slow and inventory days lengthen across the industry, it is a warning that a slowdown is brewing. Texas Instruments' inventory turnover eased during its recent soft patch, from about 1.63 times in fiscal 2023 to 1.45 in fiscal 2024, per Tenet data, consistent with a downturn in which stock built up faster than it sold, before conditions improved. Notably, TXN deliberately carries extra inventory to keep its many parts available, so its levels have to be read against that strategy rather than in the abstract, which is why context matters as much as the raw number.

The signal is most powerful when read across the supply chain rather than at one company. A single chipmaker's inventory can rise for benign reasons, a deliberate buffer, a product launch, but when inventories swell across makers, distributors, and customers at once, the industry is usually heading into a correction. Rising inventory across the supply chain is the canary; heed it well before the revenue line confirms what it is telling you.

Where to go from here

Analyzing a semiconductor company means reading it through the cycle, identifying whether it owns its factories or outsources them, judging the stickiness of its design wins, and watching inventory as the earliest warning of a turn. From here, compare another price-taking cyclical business in how to analyze an energy company, and the opposite, capital-light end of technology in how to analyze a software company. Then read how we analyze Texas Instruments for the full case, and open the Texas Instruments financials on Tenet to trace the cycle for yourself.