According to a filing with the U.S. Securities and Exchange Commission on Tuesday morning, chip manufacturing giant Taiwan Semiconductor Manufacturing (TSM.US), often called the "King of Chip Foundries," has approved an additional capital expenditure of up to $20 billion for its super chip factory in Arizona, USA. This high-intensity capital spending by TSMC is not only about actively expanding production capacity for AI chips and 2.5D/3D advanced packaging but also about building long-term competitive barriers for future 3nm, 2nm, and even more advanced platforms, strategically enhancing its core competitiveness. Therefore, all capacity expansion moves by similar chip manufacturers serve as a significant catalyst for companies like ASML, which covers EUV lithography machines, and other semiconductor equipment giants focused on advanced process technologies such as etching, thin-film deposition, and CMP. TSMC Arizona, a wholly-owned subsidiary located in Phoenix, Arizona, is constructing a large-scale semiconductor manufacturing base there. TSMC had previously approved a massive $165 billion investment for this project. The base plans to include six large semiconductor wafer fabs, two advanced packaging infrastructure facilities, and a major R&D team center. TSMC management stated this represents the largest foreign direct investment in a greenfield project in U.S. history. In October 2025, the factory began mass production of Nvidia's (NVDA.US) major Blackwell architecture AI GPU using its advanced N4P process technology. The company plans to achieve mass production of N3-level (roughly equivalent to 3nm process) advanced process technology in the second half of 2027. Its N2 and A16 advanced process technologies are expected to be deployed for chip manufacturing and advanced packaging capacity in 2030. Furthermore, TSMC's board approved a capital budget of approximately $31.28 billion on Tuesday for constructing advanced chip manufacturing technology capacity, progressing fab construction, and installing fab infrastructure systems. This is to meet the AI computing infrastructure capacity expansion demands led by major clients like Nvidia, AMD, and Broadcom, aiming to maximize alignment with the exponentially expanding AI computing demand in recent years. Simultaneously, the board approved a cash dividend of $0.22 per share for the first quarter of 2026. This dividend will be paid on October 8, 2026, as per TSMC's announcement. The global "chip bottleneck" is becoming an unprecedented strategic opportunity for TSMC. Within the current global AI computing supply chain, TSMC's chip manufacturing capability has become the scarcest critical resource globally. As U.S. tech giants like Microsoft, Meta, Google under Alphabet, and Amazon collectively push annual capital expenditures toward $800 billion, primarily for AI computing infrastructure construction, TSMC, as the primary manufacturer for almost all advanced AI chips and highest-performance computing chips for data centers, sees its most advanced process capacities (including 3nm/2nm/future N2) consistently operating at full capacity or even oversubscribed. This supply-demand imbalance makes its nearly irreplaceable manufacturing capability the "bottleneck node" for the entire industry. Unlike traditional memory or general-purpose logic chips, chips handling AI inference/training workloads, such as AI accelerators, ASICs, and data center multi-core CPUs, require extremely high technical standards in wafer design, lithography, interconnects, and high-bandwidth packaging, with very few production lines and slow expansion—TSMC's long-standing global leadership position makes it virtually the "hub and power source of the advanced chip supply chain." Under the unprecedented AI infrastructure frenzy, TSMC's U.S. ADR (TSM.US) trading price has entered a long-term bull market, surging an astonishing 150% over the past year, with a market capitalization exceeding $2 trillion. In the most advanced 3nm process domain, TSMC faces almost no direct competitors. While Samsung has strength in memory and mature processes, its capacity and market share at the 3nm/2nm nodes are far inferior to TSMC's; Intel's foundry business is still in a cultivation phase, and new wafer fabs in Japan and the U.S. will need years to reach stable mass production. This supply-demand mismatch not only gives TSMC significant leverage in customer capacity allocation but has also led some customers to book capacity years in advance or even pay deposits to secure future output, which is extremely rare in the history of the global semiconductor supply chain. To address the exceptionally strong and continuously expanding AI computing demand, TSMC has been expanding capacity at a record pace in recent years—not only maintaining a record capital expenditure range of $52 billion to $56 billion for 2026 but also specifically approving an additional $20 billion capital injection into its U.S. Arizona subsidiary to accelerate advanced process capacity and related large-scale chip manufacturing factory infrastructure construction. This signifies TSMC's full commitment to a global expansion strategy, actively expanding advanced lines in Taiwan while accelerating the deployment of high-end manufacturing and advanced packaging capabilities in the U.S. market (TSMC plans to complete an advanced packaging plant by 2029) to alleviate global chip supply tightness. Strategically, this high-intensity capital spending is not merely about capacity expansion but about constructing long-term competitive barriers for future 3nm, 2nm, and even more advanced process platforms, strategically enhancing its core competitiveness. The multiplier effect brought by the "AI bull market narrative": TSMC's capacity expansion ignites semiconductor equipment demand. Current global AI computing infrastructure and data center enterprise storage chip demand continue to show exponential growth trends, with supply far lagging behind demand intensity. This is evident from the exceptionally strong performance and significantly higher-than-expected capital expenditure guidance recently announced by "global chip king" TSMC (TSM.US), as well as the substantially increased performance and outlook from global semiconductor equipment leaders Applied Materials and Lam Research Corp. TSMC's latest disclosed trend of expanding capital expenditure and capacity expansion most directly benefits semiconductor manufacturing equipment suppliers and packaging equipment manufacturers. In expanding wafer fabs and advanced packaging lines, TSMC requires a large number of extreme ultraviolet (EUV) lithography machines, thin-film deposition, etching, and other equipment, which are the core products of equipment giants like ASML, Applied Materials, Lam Research, and KLA. As TSMC's global capacity expansion directly drives equipment demand, it catalyzes substantial order growth for these equipment suppliers and will propel performance expansion and capital expenditure recycling across the entire advanced manufacturing equipment supply chain. Recently, several Wall Street financial giants have published research reports stating that the semiconductor equipment sector is one of the biggest winners under the overwhelming AI computing and storage demand. As the global hyperscale AI data center construction process led by tech giants like Microsoft, Google, and Meta intensifies, it comprehensively drives chip manufacturing giants to accelerate expansion of 3nm and below advanced process AI chips, CoWoS/3D advanced packaging capacity, and DRAM/NAND storage chip capacity. The long-term bull market logic for the semiconductor equipment sector is becoming increasingly robust. The unprecedented AI infrastructure wave and storage supercycle have pushed semiconductors into a new phase that is more "material-intensive, process control-intensive, and with packaging processes moving forward": three-dimensional structures and new materials on the logic side, HBM stacking and interconnect upgrades on the memory side, and CoWoS/hybrid bonding on the packaging side converting system performance into manufacturing difficulty—these three forces collectively increase the value density of key segments like deposition/etching/CMP/advanced packaging/core metrology and have more clearly rewritten semiconductor equipment demand from "cyclical fluctuations" to a "structural major expansion cycle."
Comments