Chokepoints, Tollbooths, and the Complexity of the Sovereign Semiconductor Industry

The Semiconductor Stack Meets the Realities of Globalization 3.0

A few weeks ago, I was sitting in an interview with a veteran executive who has spent forty years in the semiconductor industry. As we got to talking about the current state of the global semiconductor supply chain, he said something that stuck with me: “The semiconductor industry is the true global village today, where everyone still has to talk to each other.”

He’s exactly right in describing today’s world.

In the current political landscape, there is a mad scramble to “onshore” and “verticalize” microchip production. Leaders in Washington, Brussels, and Beijing are tossing around massive subsidies — like the US CHIPS Act and the European Chips Act — under the impression that a country or region needs to build a self-contained ecosystem from scratch.

But when you unpack the mechanics of how a silicon chip is physically created, designed, and packaged, you quickly realize that the semiconductor industry isn’t a collection of independent national fortresses. Recreating it would be a non-trivial task for any country.

What Is the Semiconductor Stack?

As discussed previously, technologists frequently talk about a “tech stack” to describe the layers of software and infrastructure that deliver an app to your phone or computer. At the national level, I’ve introduced the concept of the New Stack to map the contemporary determinants of a country’s competitive advantage.

In the New Stack, semiconductors fall into the Innovation section of the “4 I’s.”

But the semiconductor industry has its own stack. It is a highly specialized ladder where no single geographic region holds all the rungs.

To move from an idea to a functioning slice of silicon, a chip must travel through every single one of these specialized layers. If a nation is dominant in layer four but completely empty in layer one or layer six, its industrial independence is an illusion.

(NB: Sometimes there is overlap in parts of this stack; for example, some Nvidia modules include memory and other core IP components. Nonetheless, almost all chip solutions require each of these layers.)

Where the Heavyweights Sit

When you look at the top tier of this industry, the distribution of market power and scale is staggering. The world is divided into distinct domains of specialization.

The United States: The Superpower Architects

The US excels towards the top of the stack—the asset-light, ultra-high-margin realms of intellectual property and architecture design. The fabless design stars, driven by the generative AI explosion, are huge. NVIDIA’s revenue surged to an unprecedented $215.9 billion in 2025. To put that in perspective, NVIDIA’s explosive growth alone over the last two years is larger than the entire European semiconductor industry combined. Companies such as AMD, Intel (despite its recent troubles), Broadcom, and Marvell are semiconductor powerhouses.

Blueprints from giants like Synopsys ($7.1 billion in 2025 revenue) and Cadence Design Systems ($5.3 billion in 2025) are the tollbooths of the ecosystem. You literally cannot draw a cutting-edge chip without their products.

Asia: The Physical Foundries and Computer Memory

If the US owns the brains, Asia owns the brawn. Taiwan’s TSMC operates as the absolute bottleneck of physical manufacturing, pulling in an astronomical $121.9 billion in 2025 and controlling over 72% of the contract foundry market. Even with severe Western sanctions, China’s national champion foundry, SMIC expanded its monthly production capacity to clear $9.3 billion in 2025 revenue, feeding a seemingly insatiable local market for smartphones and automotive tech.

Meanwhile, South Korea’s SK Hynix ($68.7 billion annual revenues) and Samsung Electronics ($59.9 billion semiconductor annual revenues) rule the vital memory systems required for AI accelerators. Samsung Foundry is also the world’s second largest semiconductor contract manufacturer behind TSMC, with an additional $12.6 billion in revenues in 2025.

Europe: The Enablers and Operators

Europe’s footprint is entirely unique. The continent doesn’t host the giants making consumer smartphone or cutting-edge AI processors. Instead, it dominates the ultra-complex machinery required to print chips in the first place, alongside the rugged, industrial microchips that run our cars and power grids.

ASML (Netherlands — $37.7 billion annual revenues) in extreme ultraviolet (EUV) lithography, Infineon (Germany — $16.2 billion annual revenues) in automotive & power management, NXP (Netherlands — $12.3 billion annual revenues) in connected mobility & infotainment, STMicroelectronics (Switzerland — $11.8 billion annual revenues) in microcontrollers, and Arm Holdings (United Kingdom — $4.0 billion annual revenues) in developing and licensing semiconductor IP are the continent’s largest players.

The True Chokepoints: Zooming Past the Logos

When looking closely at these dynamics, we see certain patterns. Much economic policy focus is drawn to the high-profile foundries and intellectual designs. But many of the structural chokepoints — the places where a system-wide failure can occur — lie in the unglamorous minutiae.

Consider ASML. They hold a global monopoly on extreme ultraviolet lithography systems. If you are using a modern smartphone or an accessing AI in a remote data center, their brains were printed using an ASML machine. There is no alternative.

But let’s go a layer deeper into materials science. An advanced fab is essentially a hyper-controlled chemical laboratory operating at massive industrial scale. It requires:

• Advanced Photoresists: The light-sensitive liquid polymer spun onto a wafer before laser printing. This sector is a literal geopolitical chokepoint dominated almost entirely by Japan via companies like Tokyo Ohka Kogyo (TOK) and JSR Corporation.

• Industrial Gases: Fabs consume millions of gallons of water and massive quantities of ultra-pure specialty gases like silane and nitrogen trifluoride to alter electrical conductivity. The piping and infrastructure are dominated by two large players: Linde (UK/Switzerland) and Air Liquide (France).

• Back-End Advanced Packaging: Once a wafer is printed, it must be sliced, tested, and stacked. This market—known as OSAT (Outsourced Semiconductor Assembly and Test) — is heavily centralized in Asia. Taiwan’s ASE Technology controls over 30% of the market ($20.8 billion in 2025), while the premier US player, Amkor Technology ($6.7 billion), still operates the majority of its square footage overseas.

When you look at this level of granularity, the concept of a single country safely verticalizing its own semiconductor supply chain becomes an engineering and economic challenge that would take years and years to develop.

Is This the True Global Village? Or Is That Naive?

In my book The Systems Leader, I argued that Globalization 1.0 was built on simple labor arbitrage, while Globalization 2.0 operated like a communication mesh network. Today, we are deep into Globalization 3.0, where geopolitical clashes are being fought directly within the business arena through economic ideology rather than pure political ideology.

The semiconductor stack is the ultimate implementation of Globalization 1.0 — a global village that is a completely interdependent, multinational marvel where a single chip requires the cooperation of American software, Dutch physics, Japanese chemistry, Taiwanese fabrication, and Malaysian packaging.

But here is the paradox: while it is a global village, it is anything but peaceful.

It is a village defined by weaponized interdependencies. It is a world where the United States uses its software moats to cut off Chinese tech firms, and where Western tech giants face immense pressure to decouple from Taiwan out of fear of a cross-strait conflict. European semiconductor leaders also worry about becoming too dependent on either the United States or China.

This has led to massive, counter-activities. For example, the US is backing Amkor to build a $2 billion advanced packaging site in Peoria, Arizona, right down the road from TSMC’s new multi-billion-dollar fabs — attempting to forge a full-stack domestic corridor on US soil.

But working the system requires understanding its real constraints. You can spend billions on the concrete shell of a foundry, but if you are still single-threaded through a handful of chemical plants in Japan for your photoresists, or reliant on a single industrial gas pipeline from a European multinational, you haven’t bought sovereignty. You might have more agency, or potentially resiliency, but you don’t have full control.

Key Takeaways

What does this mean for today’s business and political leaders?

1. Size Creates Competitive Advantage — As mentioned in the Substack on the Culture of Creativity, size creates competitive advantage in today’s world. The technological moats held by companies like ASML, TSMC, and Nvidia are so deep precisely because the global aggregation of capital and talent has concentrated into their hands. This part of the New Stack is tied deeply to capital and labor force development.

2. National Interests and Company Interests Are Getting Harder to Separate — The large drivers of today’s economic growth and profits for corporations are increasingly tied to key technologies that impact consumers, enterprises, and militaries. Leaders in government need to become conversant in technology and business fundamentals, and corporate executives need to understand the geopolitical conflicts facing national leaders, and how government and citizen’s interests are increasingly interwoven into business objectives.

We cannot design our way out of physical dependencies, and we cannot build our way out of digital dependencies. For a country or a company to gain agency and control in parts of the New Stack requires seeing dependencies and realizations of where political and business interests come together and where they conflict.

There are stacks within stacks. And they are all intertwined.