The global semiconductor foundry industry has reached an estimated $402 billion in 2026, reflecting both the scale and the structural complexity of modern semiconductor manufacturing, according to Yole Group’s latest report, “Status of the Semiconductor Foundry Industry 2026.” Foundries now sit at the crossroads of geopolitics and the AI capex boom, but do we really understand what they are about?
Advanced nodes or leading edge: Why it matters
It is widely cited that TSMC holds about 72% share of the global semiconductor foundry market and more than 90% of advanced chips. Such statements and figures only partially reflect TSMC’s true position. Its dominance at the leading edge is often understated due to inconsistent definitions of “advanced.” Depending on the source, this term may refer to nodes below 20 nm, aligned with U.S. export controls; below 10 nm, including non-planar technologies, which is the Yole Group definition; or below 5 nm, excluding China’s 7-nm capabilities and comparing TSMC with Intel and Samsung only.
A more meaningful approach is to focus on the leading edge, currently the 3-nm class in high-volume manufacturing, where TSMC holds an estimated ~95% market share. Samsung has not consistently matched TSMC’s performance at this node and primarily supplies 7-nm and 5-nm wafers to third parties seeking advanced nodes. Intel’s 18A node, also positioned within the 3-nm class, is only beginning to ramp and remains largely dedicated to internal products. The upcoming transition to the 2-nm class will be a key inflection point. Samsung’s early deployment for its internal smartphone division suggests potential shifts, but it remains to be seen whether TSMC’s ~95% share of the leading-edge market will materially change in the future.
Foundry market or open foundry: Defining the industry perimeter
A second misconception relates to the structure of the foundry market. The industry is roughly split 53%/47% between captive manufacturing (IDMs) and open foundry services. References to the foundry market typically apply only to the open segment, which accounts for about 47% of total wafer manufacturing revenue. Within that segment, TSMC’s ~72% share is accurate. However, across total global manufacturing, including both open and captive, TSMC’s share is about 34%. This distinction is indeed critical when framing the entire semiconductor industry.
Integrated device manufacturers (IDMs) operate hybrid models, balancing internal production with outsourcing. They may open portions of their internal capacity to third parties as well as rely on outsourcing partners for advanced nodes. Fabless companies engage both pure-play foundries, e.g., TSMC, and non-pure-play providers, including Samsung, Intel, and STMicroelectronics. As a result, terminology can inflate perceived market concentration. TSMC’s position represents clear dominance, but not a monopoly when viewed across the full manufacturing landscape.
Regional technology gaps: Is it a race between identifiable players?
The third misconception is captured in statements such as “China is 10 to 15 years behind.” Regional comparisons oversimplify a deeply interconnected global ecosystem. The semiconductor industry is now built on extensive international collaboration, with interdependence spanning design, equipment, materials, and manufacturing. No single region operates in isolation. China’s most advanced foundries, such as SMIC and Hua Hong, currently produce at the 7-nm node, roughly equivalent to TSMC capabilities circa 2019. This allowed Huawei to reclaim the #1 position in the 5G smartphone market in China in 2025. The only drawback is more silicon per phone, but without a meaningful performance gap. Specific player and technology constraints do not support a blanket “10 to 15 years behind” narrative. Taiwan, for example, leads in advanced manufacturing but relies heavily on European lithography equipment as well as Japanese photoresist companies. Each region contributes critical elements within a global value chain.
Semiconductors in 2026 differ sharply from those of 1966, when U.S. scientists had just invented them; 1986, when Japan dominated; and 2006, when South Korea had emerged as a leading player. In 2026, semiconductors are not the product of any single nation acting in isolation but the result of coordinated global expertise involving millions of engineers and specialists. Framing the industry through some hypothetical race of regional ecosystems overlooks this structural reality. It remains one of the few examples—if not the primary one—of technology globalization, in which innovation during the last twenty years has been fundamentally collective, and where global and peaceful interdependence, not war, is the foundation of the industry.
https://www.eetimes.com/three-misconceptions-about-the-402b-semiconductor-foundry-industry/
