What's happening

Speaking at TSMC's annual shareholders' meeting on June 4, 2026, in Hsinchu, Taiwan, Chairman and CEO C.C. Wei delivered a candid assessment of the semiconductor industry's capacity limitations relative to accelerating AI infrastructure demand. 'Customer demand is so high, and we can only support so much. We are already working very hard,' Wei told shareholders, adding that 'It will be a long time before we can meet customer demand.' Wei identified Nvidia, Broadcom, and American hyperscalers as among the customers driving that demand, while pointing to autonomous vehicles and robotics as additional long-term growth engines that will further strain supply.

Despite the supply-side constraints, TSMC presented a robust financial outlook at the meeting. The company is forecasting a 30% sales increase for 2026, and announced plans to raise cash dividends from NT$18 to at least NT$24 per share. Employee profit sharing is also set to grow by another 30%. Wei also indicated he would like to raise prices, while signaling an intent to avoid abrupt increases — a posture that reflects both the company's pricing leverage and its awareness of customer dependencies. TSMC shares declined 1% in Taipei following the meeting.

Why it matters for markets

TSMC's capacity constraints carry significant financial implications for its largest customers. Nvidia, which relies on TSMC's leading-edge process nodes to manufacture its H100 and Blackwell GPU families, reported revenue of $253.49 billion and carries a market capitalization of $5.30 trillion — a valuation that is heavily predicated on continued AI data center buildout. Any sustained gap between wafer supply and customer orders introduces execution risk into Nvidia's ability to fulfill demand from hyperscalers and enterprise customers. AMD, which similarly depends on TSMC fabrication for its Instinct AI accelerators and EPYC server processors, reported revenue of $37.45 billion, and faces comparable supply-side exposure as it attempts to expand its share of the AI accelerator market.

TSMC's own 30% sales growth forecast for 2026 underscores the scale of demand it is attempting to serve, but Wei's explicit acknowledgment that supply will remain insufficient for years suggests that allocation decisions — which customers receive wafer capacity and on what timeline — will be a defining competitive variable across the semiconductor landscape. The company's stated desire to raise prices, even if implemented gradually, would affect the cost structures of fabless chip designers across the industry. TSMC's current price-to-earnings ratio of 38.0 and a market capitalization of $2.31 trillion reflect investor expectations of sustained revenue growth, but the 1% share decline in Taipei following Wei's remarks indicates the market is also weighing the operational and reputational risks of prolonged supply shortfalls.

The supply-demand imbalance Wei described is not confined to a single product cycle. By naming autonomous vehicles and robotics alongside AI data centers as long-term demand drivers, Wei signaled that capacity pressure is structural rather than cyclical. This framing has implications for capital expenditure planning across the semiconductor supply chain and for the revenue visibility of any company whose product roadmap depends on access to TSMC's most advanced process nodes.

Sectors and assets to watch

The primary tickers directly implicated by Wei's comments are TSM, NVDA, and AMD. TSMC (TSM), as the world's largest dedicated semiconductor foundry with 76,907 employees and a 52-week price range of $202.28 to $450.16, sits at the center of the supply constraint narrative — its ability to expand capacity at advanced nodes will determine how quickly its customers can scale AI product shipments. Nvidia (NVDA), with a 52-week range of $138.83 to $236.54 and a P/E of 33.4, is explicitly named as one of the customers whose demand TSMC is struggling to meet, making wafer allocation a direct variable in Nvidia's near-term revenue execution. AMD (AMD), trading with a P/E of 175.6 and a 52-week range of $114.71 to $546.44, faces analogous foundry dependency for its Instinct accelerator line as it competes for AI workload market share.

Broadcom, also named by Wei as a key customer, operates in the same foundry-dependent segment and faces comparable allocation dynamics. Beyond individual companies, the broader hyperscaler ecosystem — the American cloud infrastructure operators Wei referenced — will be watching TSMC's capacity expansion timeline closely, as their own AI capital expenditure programs are contingent on a reliable supply of advanced chips. The foundry sector more broadly, including domestic Chinese alternatives positioned outside TSMC's ecosystem, may also draw attention as customers and governments assess supply chain diversification strategies.

What to watch next

Investors and analysts will be monitoring TSMC's quarterly earnings calls and capital expenditure announcements for concrete timelines on new fab capacity coming online, particularly at advanced nodes relevant to AI accelerator production. Any formal price adjustment announcements from TSMC — given Wei's stated desire to raise prices without abrupt increases — will be closely tracked for their downstream impact on the cost structures of Nvidia, AMD, and Broadcom. Updates from Nvidia and AMD on their own order backlogs, lead times, and product allocation guidance will serve as a secondary indicator of how the supply gap is affecting near-term revenue recognition. Progress on TSMC's overseas fab projects, including facilities in the United States and Japan, will also be relevant to assessing when meaningful incremental capacity may enter the supply chain.