
Foundry Manufacturer
A foundry manufacturer is a company that produces computer chips for other firms without designing chips itself. Because modern AI systems require enormous computing power, foundry manufacturers have become a central bottleneck of the global technology industry.
A foundry manufacturer is a factory that produces computer chips — but doesn’t design any of its own. Instead, foundry manufacturers receive finished blueprints from other companies and manufacture the chips according to their specifications. The clients are often companies that know exactly how a chip should look but don’t own a factory of their own. This model separates the design and manufacturing of chips into two distinct industries. The world’s best-known foundry manufacturer is the Taiwanese company TSMC, which produces chips for, among others, Apple, Nvidia, and AMD.
Why foundry manufacturers control the AI industry
AI models like ChatGPT or Google's Gemini require specialized chips known as GPUs or AI accelerators. These chips are so technically demanding that only a handful of factories worldwide are even capable of producing them. If you can’t secure access to one of these factories, you simply can’t build your chip — no matter how good the design is.
This makes foundry manufacturers a bottleneck for the entire field of AI development. When TSMC’s capacity is fully booked, dozens of companies are left waiting in line. Demand for AI chips has grown so sharply in recent years that even major corporations wait months for their deliveries. As a result, governments are now pouring billions in subsidies into the construction of new chip factories in the US, Europe, and Japan.
How a chip goes from design to factory
The process begins when a company — Nvidia, for example — designs a chip on the computer. This design specifies, down to the nanometer, how millions of tiny circuits should be arranged. Nvidia then sends this plan to a foundry manufacturer like TSMC.
At the factory, the chip is etched onto round silicon discs called wafers — through a highly precise process using ultraviolet light. A single wafer can contain hundreds of individual chips. Manufacturing precision is measured in just a few nanometers; for comparison, a human hair is roughly 70,000 nanometers thick. The smaller the structures, the more computing power fits onto a chip — and the less power it consumes.
It’s important to distinguish foundry manufacturers from so-called integrated device manufacturers, or IDMs. Intel, for instance, designs and produces many of its chips itself. A pure foundry manufacturer like TSMC, by contrast, never has its own chip product on the market — it doesn’t compete with its own customers.
Foundry manufacturers in business and in the news
In business news, the term tends to come up mainly in the context of supply shortages, trade conflicts, or the chip competition between the US and China. Taiwan, home to TSMC’s headquarters, produces more than half of all chips manufactured worldwide — and an even larger share of the most advanced ones. This gives the island a geopolitical significance far beyond its size.
That’s why the US government allocated around 52 billion dollars through the so-called CHIPS Act in 2022 to encourage foundry manufacturers to build factories on American soil. TSMC is now building in Arizona, Samsung in Texas. The EU has responded as well, with the European Chips Act. The goal is to reduce dependence on Asian foundry manufacturers.
In everyday life, most people encounter the term indirectly: every smartphone, every laptop, and every AI voice assistant contains chips manufactured in a foundry. The product itself is invisibly built in — yet the factory behind it shapes which technologies even make it to market, and how quickly.