TSMC

Taiwan Semiconductor Manufacturing Company (TSMC) is the world's principal pure-play semiconductor foundry, founded in 1987 by Morris Chang and headquartered in Hsinchu, Taiwan. The company manufactures leading-edge semiconductor chips on contract for fabless customers including NVIDIA, Apple, AMD, Qualcomm, Broadcom, and other principal AI and consumer-electronics chip designers. TSMC operates the most advanced process technology globally, with N3 and N2 process nodes in volume production and continued investment in the next-generation A16 (1.6-nanometer-class) and A14 nodes. The company has substantial in-house process-development ML research, machine-learning-augmented yield-management software, and applied-AI integration across the fabrication-and-test workflow. As of April 2026, TSMC is one of the structurally consequential companies in the global AI-compute supply chain, with the manufacturing capacity for substantially all leading-edge AI accelerators globally.

At a glance

• Founded: 1987 in Hsinchu, Taiwan, by Morris Chang. Listed on Taiwan Stock Exchange (2330.TW) and NYSE (TSM).
• Status: Public. Multi-trillion-Taiwan-dollars market capitalization (over $700 billion USD as of April 2026). Among the largest semiconductor companies globally.
• Funding: Public-market financing. Annual revenue exceeds $90 billion as of fiscal year 2025 to 2026.
• CEO: C.C. Wei, Chief Executive Officer (since 2018). Long-tenured TSMC executive.
• Other notable leadership: Morris Chang, Founder and former Chairman (retired 2018; remains an advisor). Mark Liu, former Chairman (retired June 2024). Wendell Huang, Chief Financial Officer.
• Open weights: N/A. TSMC is a foundry, not a model producer.
• Flagship products: N3 (3-nanometer-class) and N2 (2-nanometer-class) process nodes in volume production; A16 (1.6-nanometer-class) under development; CoWoS (Chip-on-Wafer-on-Substrate) advanced packaging; ML-augmented yield-management and process-control systems.

Origins

TSMC was founded in 1987 by Morris Chang, a former senior executive at Texas Instruments and General Instrument, with backing from the Taiwanese government and Philips. The founding business model of pure-play foundry (manufacturing chips for external designers without competing in chip design) was structurally novel at the time and proved to be one of the most consequential business-model innovations in semiconductor history. The model enabled the rise of fabless chip companies (NVIDIA, Qualcomm, Broadcom, MediaTek, AMD post-spinout of GlobalFoundries) that designed chips without operating fabs.

The 2000s and 2010s saw TSMC's structurally consequential process-leadership investment that took the company from a peer fabricator alongside Samsung, Intel, and GlobalFoundries to the global leader at leading-edge process nodes. The 2020 transition to N5 (5-nanometer) volume production established TSMC's clear leadership over Intel (whose 10-nanometer-class process struggled with scale-up) and Samsung (whose 5-nanometer process produced lower yield than TSMC's). The 2022 transition to N3 (3-nanometer) further extended the lead.

The 2022 to 2026 period has been structurally consequential for TSMC's role in the AI-compute supply chain. The substantial NVIDIA AI accelerator demand (H100, H200, B200, GB200, and successor SKUs) is fabricated entirely at TSMC, with CoWoS advanced packaging the principal capacity bottleneck through 2024 to 2025. The company has invested substantially in CoWoS capacity expansion, advanced-packaging research, and the next-generation A16 and A14 process-node development.

The geopolitical environment around Taiwan has anchored substantive supply-chain risk discussions across the global AI-compute ecosystem, with US export-control restrictions on TSMC chip sales to certain China-based AI-and-semiconductor customers, US CHIPS Act incentives for TSMC's Arizona-based fab construction, and continued political attention to Taiwan's role as the principal supplier of leading-edge AI compute.

Mission and strategy

TSMC's stated mission is to be the technology and capacity provider of choice for the global semiconductor industry. The strategy combines three threads. First, leading-edge process technology investment, with N3, N2, A16, and A14 process-node development providing the principal capability moat. Second, advanced-packaging investment (CoWoS, SoIC, InFO) that enables high-bandwidth-memory integration and chiplet-based AI accelerator architectures. Third, applied ML and process-control research that enables the yield, cycle-time, and operational-efficiency improvements that the leading-edge process-node economics require.

The competitive premise is that semiconductor manufacturing at the leading edge requires multi-decade investment in process technology, supply chain, and operational expertise that produces structurally durable competitive advantages, with the principal commercial risks being end-customer demand cycles and geopolitical-supply-chain risk.

Models and products

• N3 and N2 process nodes. Leading-edge processes in volume production. Customer base includes NVIDIA, Apple, AMD, Qualcomm, and other fabless leading-edge customers.
• A16 and A14 process nodes. Under development. Targeted for late 2026 to 2028 volume production.
• CoWoS advanced packaging. Chip-on-Wafer-on-Substrate packaging that integrates GPU dies with high-bandwidth memory. Principal packaging technology for NVIDIA H100, H200, B200, GB200, and successor SKUs.
• InFO and SoIC packaging. Adjacent advanced-packaging technologies.
• ML-augmented yield-management and process-control. Internal applied-AI integration across fabrication-and-test workflows.
• Customer engagement programs. TSMC's Open Innovation Platform and Design Center alliance for fabless customers.

Distribution channels are direct enterprise relationships with global fabless semiconductor customers and integrated device manufacturers requiring leading-edge process technology.

Benchmarks and standing

TSMC is evaluated through process-leadership metrics (process-node leadership relative to Intel and Samsung), commercial metrics (revenue, gross margin, customer-concentration), and operational metrics (capacity utilization, yield, cycle time).

The company's process-leadership position has been consistently confirmed by industry analysis through 2020 to 2026, with Intel's process-leadership comeback efforts ongoing and Samsung's leading-edge yield consistently characterized in industry coverage as below TSMC's. TSMC's role as the principal manufacturing partner for substantially all leading-edge AI accelerators has been a frequently cited validating data point for the company's structural position in the AI-compute supply chain.

Leadership

As of April 2026, TSMC's senior leadership includes:

• C.C. Wei, Chief Executive Officer (since 2018) and Chairman (since June 2024 following Mark Liu's retirement).
• Wendell Huang, Chief Financial Officer.
• Senior process-development, manufacturing, and customer-engagement leadership across the global TSMC organization.

Morris Chang, Founder and former Chairman, retired in 2018 and remains an advisor. Mark Liu, former Chairman, retired in June 2024.

Funding and backers

Public-market capitalization on Taiwan Stock Exchange (2330.TW) and NYSE (TSM). Multi-trillion-Taiwan-dollars market capitalization (over $700 billion USD as of April 2026). Annual revenue exceeds $90 billion as of fiscal year 2025 to 2026.

Industry position

TSMC occupies a structurally distinctive position as the principal pure-play semiconductor foundry globally and the manufacturing partner for substantially all leading-edge AI accelerators. Industry coverage has consistently characterized TSMC as one of the structurally consequential companies in the global AI-compute supply chain, with the process-leadership position over Intel and Samsung as the principal validating data point.

Competitive landscape

• Samsung Foundry. Direct leading-edge foundry competitor. Lower customer share at leading nodes due to historically lower yield than TSMC.
• Intel Foundry Services. Re-entry into the foundry market. Process-leadership comeback efforts ongoing through 2024 to 2026.
• GlobalFoundries. Foundry competitor focused on mature process nodes rather than leading-edge.
• SMIC. Chinese state-backed foundry. Restricted from leading-edge capacity by US export controls on EUV systems.
• ASML, Applied Materials, Lam Research, Tokyo Electron. Semiconductor capital-equipment partners (not direct competitors).
• NVIDIA, Apple, AMD, Qualcomm, Broadcom. Principal fabless customers.

Outlook

• The continued N3, N2, A16, and A14 process-node cadence through 2026 to 2028.
• CoWoS advanced-packaging capacity expansion to address AI-accelerator demand.
• The Arizona-based fab construction under US CHIPS Act incentives.
• The geopolitical-supply-chain environment around Taiwan and US-China export controls.
• The competitive dynamic against Samsung Foundry and Intel Foundry Services at leading-edge nodes.

Sources

• TSMC official site. Company reference.
• TSMC NYSE listing TSM. Public-market financial reporting.
• Morris Chang Wikipedia. Founder reference.
• C.C. Wei Wikipedia. CEO reference.
• TSMC investor relations. Public-market reporting.