Technological and economical power go hand in hand. This is best reflected in the geographical distribution of semiconductor fabrication plants.
We have charted selected plants, or "fabs" on a map, using data that is publicly available on Wikipedia.
If a plant is capable of producing multiple node sizes, the smallest, also the most advanced size is picked. Plants that cannot make smaller than 60 nm nodes do not qualify for our list.
Fabs with the capability of manufacturing <20nm nodes are shown in green dots.
Very few companies possess the technology to accurately and reliably print MOSFET nodes on a silicon wafer. Node size determines the density of the logical circuitry on a microchip. Smaller the size, faster and more energy efficient the processor. Before the 10th Gen, Intel's iCore series were 14nm. Some of the 10th Gen are designed in 10nm. Intel is also getting close to the capability of 7nm.
The flagship smartphones from highly competitive vendors use even smaller nodes. Although we cannot compare an Intel processor to an ARM chip, the sheer capability of producing small nodes is highly sought-after. In fact, 5nm fabs are the crown jewel and chip making.
Building a fabricating plant requires serious capital. We are talking about a starting price of 2 billion USD. TSMC, the world's largest semiconductor foundry, spent 17.08 billion for their 2020 5nm plant.
To make the investment heftier, the production line is not reusable. When 5nm fabs are phased out, they cannot be repurposed to make 3nm processors.
The above two points combined, a financially feasible plant has no choice but to make the smallest possible nodes, in large quantities. The economy of scale can only be reached through the decoupling of chip design and chip manufacturing. Therefore, all the high-caliber fabs are pure-play foundries.