Taiwan Semiconductor Manufacturing Co. (TSMC) just reported a staggering $122.4 billion in 2025 revenue, a 35.9% year-over-year surge driven overwhelmingly by artificial intelligence demand. But behind these eye-watering numbers lies a complex story of geopolitical maneuvering, supply chain tensions, and a fundamental shift in how AI is reshaping both global manufacturing and everyday work. As TSMC accelerates its Arizona fab construction to meet “very strong AI-related demand,” the ripple effects are being felt from corporate boardrooms to individual developers building their own apps with AI tools.
The AI Chip Gold Rush
TSMC’s fourth-quarter revenue hit $33.7 billion, slightly exceeding guidance, with high-performance computing accounting for 58% of total revenue – up from 51% in 2024. “This is driving need for more and more computation, which supports the robust demand for leading-edge silicon,” said Chairman and CEO C.C. Wei on the earnings call. The company expects 2026 to be another strong growth year, anticipating nearly 30% revenue increase.
But here’s what makes this more than just another corporate earnings story: TSMC is simultaneously navigating a high-stakes geopolitical chess game. The company announced plans to invest an additional $100 billion in the United States, bringing total U.S. investment to $165 billion. This includes six advanced wafer manufacturing fabs in Arizona, with the second fab’s production schedule being accelerated due to “strong customer demand.” Construction on the third Arizona fab has begun, and TSMC is applying for permits for its fourth fab and first advanced packaging facility.
The Geopolitical Backdrop
This expansion isn’t happening in a vacuum. According to a TechCrunch report, the Trump administration recently signed a multi-billion-dollar trade deal with Taiwan where Taiwanese semiconductor and tech companies will invest $250 billion into U.S. semiconductor, energy, and AI production. Taiwan will provide an additional $250 billion in credit guarantees for further investments. In return, the U.S. will invest in Taiwan’s semiconductor, defense, AI, telecommunications, and biotech industries.
“Our plan will enable TSMC to scale up an independent giga-fab cluster in Arizona to support the need of our leading-edge customers in smartphone, AI and [high performance computing] applications,” Wei said. The Wall Street Journal reported that TSMC was planning to expand once again in the U.S. as part of a $250-billion trade deal with Taiwan aimed at reshoring the domestic semiconductor sector.
The Productivity Paradox
While TSMC builds the hardware enabling AI’s expansion, companies are discovering that implementing AI isn’t as straightforward as promised. A Workday survey of 3,200 practitioners reveals what might be called “the AI productivity paradox”: 37% of time saved through AI is lost to fixing low-quality output. At least 85% of employees report saving one to seven hours per week using AI, but the follow-up rework washes out those savings – to the tune of an average 1.5 weeks a year spent fixing AI outputs.
Only 14% of employees consistently achieve net-positive outcomes from AI use. “AI has been layered onto roles that were never updated to accommodate it,” the survey authors state. In most organizations (89%), fewer than half of roles have been updated to reflect AI capabilities. Companies are more likely to put AI savings back into technology (39%) than into employee development (30%), and only 37% of employees experiencing the highest amount of rework say they’re getting access to AI training.
The Micro-App Revolution
Meanwhile, at the individual level, something remarkable is happening. People with no technical backgrounds are building their own apps using AI tools like Claude and ChatGPT. Rebecca Yu spent seven days building a dining app called Where2Eat to help her and her friends decide where to eat. “Once vibe-coding apps emerged, I started hearing about people with no tech backgrounds successfully building their own apps,” she told TechCrunch.
This trend of “micro apps” or “fleeting apps” represents a new era of app creation where individuals build context-specific applications for personal use, then shut them down when the need disappears. Software engineer James Waugh built an app for a friend who had heart palpitations – a logger to record when she was having heart issues to show her doctor. “Great example of a one-off personal software that helps you keep track of something important,” he said.
Supply Chain Realities
Back at the manufacturing level, TSMC faces significant challenges. According to The Financial Times, supply chain constraints for components like memory chips and power transformers present hurdles for AI data centers. An executive with a Nvidia supplier expressed concern: “I put a big question mark on whether we could still grow this year. We probably will, but it may be limited by how smooth the supply chain is.”
TSMC plans $56 billion in capital expenditures for 2026, expecting 30% revenue growth this year. But Wei admitted he was “very nervous” about whether AI demand was “real,” though the decision to expand was made after months of discussions with TSMC’s customers and its customers’ customers.
The Global Divide
Anthropic’s analysis of its Claude AI chatbot usage reveals another dimension: higher adoption in rich countries risks deepening global economic inequality. The research shows richer countries are more likely to adopt AI for work tasks, while lower-income countries use it primarily for education, with no evidence of catching up. This could lead to divergence in living standards as productivity gains concentrate in early-adopting nations.
“If the productivity gains materialise in places that have early adoption, you could see a divergence in living standards,” said Peter McCrory, Anthropic’s head of economics. The report estimates AI could add 1-2 percentage points to annual US labor productivity growth over the next decade.
Looking Ahead
As TSMC accelerates its U.S. expansion, the company faces both enormous opportunity and significant challenges. Capacity is “very tight,” according to Wei, and the company is working on narrowing the gap. All of TSMC’s U.S. customers are asking for “a lot of support” from the Arizona fab, driving the acceleration of expansion in the state.
The story of TSMC’s AI-driven growth is really three interconnected stories: the geopolitical reshoring of critical manufacturing, the corporate struggle to realize AI’s productivity promises, and the democratization of app development at the individual level. As AI continues to reshape everything from global supply chains to personal productivity tools, TSMC finds itself at the center of it all – building the chips that power this transformation while navigating the complex realities of making it work in practice.