GLOBALFOUNDRIES Inc. (GFS)

GlobalFoundries manufactures computer chips for other companies — it does not design chips itself, but instead makes them to specification in vast, expensive fabrication plants. The industry calls this a foundry model. It is one of the few semiconductor manufacturers in the United States, a fact that gives the company strategic importance and makes it a beneficiary of government subsidies, but it is also a perpetually capital-hungry, cyclical business where a quarter of weak demand from smartphone makers can trigger layoffs and underutilized plants.

“We do not design chips; we make chips that others designed.” — The foundry model in one sentence.

A newer player in an ancient industry

GlobalFoundries was created in 2009 when Advanced Micro Devices (AMD) spun off its manufacturing operations into a separate company — a move that let AMD focus on chip design while exiting the capital-intensive, low-margin manufacturing business. The new company was given a portfolio of several older fabrication plants and contracted to manufacture AMD’s chips.

For the first decade, GlobalFoundries was a smaller player in a market dominated by Taiwan Semiconductor Manufacturing Company (TSMC) and South Korea’s Samsung. It focused on less advanced nodes — the 28-nanometer and 14-nanometer technologies that require billions in capex but not the hundreds of billions that leading-edge 3-nanometer manufacturing demands. It also remained heavily dependent on AMD as a customer, a relationship that created both revenue stability and strategic risk.

In 2021, the company went public, and management shifted strategy. GlobalFoundries announced plans to build multiple new manufacturing plants in the United States, supported by billions of dollars in federal subsidies. The framing was strategic: the United States should not depend entirely on Taiwan for advanced semiconductor capacity. The company would invest heavily in the latest technology nodes and reduce its dependence on any single customer.

The foundry economics

The foundry business is conceptually simple but economically brutal. A customer — call it a fabless chip designer — sends GlobalFoundries the specifications and design files for a chip it wants to manufacture. GlobalFoundries runs the production process, delivering thousands of units. The company charges per chip or per wafer, earning a margin that covers its enormous fixed costs and ideally leaves profit.

The problem is scale and cycle time. A modern semiconductor fab costs billions of dollars to build and billions more per year to operate and maintain. Utilization matters obsessively — an underutilized fab is a money-losing machine. A fab that runs 24/7 at full capacity can spread its fixed costs across millions of chips. The same fab running at 60 percent capacity is inefficient.

Demand is volatile. A fabless chip designer might place an order for fifty thousand units of a new processor. If that processor sells well, the designer returns with orders for another million units, driving GlobalFoundries’ utilization up. If the processor flops or the end market softens — as happened across the smartphone industry in 2022 and 2023 — orders evaporate. A customer might disappear or demand significant price reductions. GlobalFoundries is left with idle capacity.

Customers and concentration risk

GlobalFoundries’ customer base is dominated by a small number of large fabless designers. Advanced Micro Devices remains a significant customer, as does Qualcomm. Several large customers design chips for automotive, industrial, and telecommunications applications — less cyclical than smartphones but lower volume. The company also manufactures for some Chinese-owned chip designers. No single customer typically accounts for more than around 25 percent of revenue, but the concentration is real.

This creates a perpetual strategic tension. GlobalFoundries needs to diversify away from any single customer to reduce risk. But most of the highest-volume, most profitable chips are designed by the biggest fabless companies — AMD, Qualcomm, Broadcom — and winning share from them means competing on price, which compresses margins. The foundry market is also increasingly dominated by TSMC, which has enormous scale advantages, superior technology, and long-standing relationships with the biggest customers. GlobalFoundries is the third-largest foundry globally by capacity, but TSMC’s lead is vast.

The United States capacity play and government backing

GlobalFoundries’ most significant recent strategic move is its pivot toward U.S. manufacturing. The company has announced several new fabrication plants in the United States, financed partly by company capital and partly by subsidies from the U.S. government through programs like the CHIPS and Science Act. The rationale is that U.S. national security depends on maintaining domestic semiconductor manufacturing capacity, and that GlobalFoundries should expand to serve that need.

The investment is enormous — tens of billions of dollars — and the payoff is highly uncertain. New fabs in the United States face higher labor, energy, and real-estate costs than fabs in Taiwan or South Korea. To be competitive, GlobalFoundries must either charge premium prices (limiting customers) or accept lower margins. Government subsidies help, but they are uncertain — future administrations might not continue them, and the political support for subsidizing semiconductor manufacturing could erode.

Capital intensity, cycles, and profitability

The semiconductor manufacturing business is the most capital-intensive manufacturing business that exists. A single modern fab can cost $15 billion to $20 billion to build and requires billions annually to operate and upgrade. That capital burden means foundries must achieve massive scale to be profitable, and they are extremely sensitive to capacity cycles. A few quarters of weak demand can wipe out years of profit.

GlobalFoundries’ profitability has been inconsistent. In strong cycles — when demand outpaces supply — the company earns healthy margins and generates cash. In weak cycles — when orders decline and customers run down inventory — the company burns cash, sometimes cutting headcount and pushing back capital projects. The shift toward more advanced technology nodes requires even larger capex, raising the stakes on hitting the demand cycle correctly.

What to follow

Anyone researching GlobalFoundries should begin with the quarterly and annual filings (SEC CIK 0001709048), which break revenue by customer and by technology node (indicating what percentage of revenue comes from advanced vs. mature processes). Watch the trajectory of capital expenditure — it indicates whether management is confident enough in demand to fund new plants. The most useful operating metrics are fab utilization rates and average selling price per wafer, which reveal the pricing power and the cycle.

Follow the progress of new U.S. fab construction carefully. Are new plants coming online on schedule? Are customers committing to using them, or are they waiting to see if the investment pays off? And track government policy — any changes in subsidy programs or trade restrictions would ripple through the business. Finally, watch competitive announcements from TSMC and Samsung, which reveal whether GlobalFoundries is gaining or losing share in the markets it serves.