TSMC CEO Warns AI Capacity Crunch Will Last a ‘Very Long Time’ as Price Hikes Loom

Benzinga reported Thursday that TSMC CEO C.C. Wei delivered a sobering outlook on the global semiconductor industry. Speaking at the company’s annual shareholders’ meeting in Hsinchu, Taiwan, Wei cautioned that meeting U.S. customer demand through domestic production alone would take a “very long time.” TSMC AI capacity constraints, he suggested, are not a near-term problem to be solved quickly.

Supply Crunch Clouds the AI Boom

Wei acknowledged that component costs continue to climb across the board. Despite that pressure, he said customers remain broadly optimistic about the long-term trajectory of artificial intelligence demand. The message was blunt, however. The chipmaker’s existing manufacturing footprint simply cannot keep pace with where orders are headed.

TSMC shares fell more than 2% in Thursday pre-market trading following Wei’s remarks. The stock had already slipped 2.24% the prior session, closing at $436.69. On a year-to-date basis, shares are still up roughly 36%, reflecting the enormous market enthusiasm that preceded the latest warnings.

Price Hikes Already in Sight

The CEO’s comments arrive against a backdrop of widely reported plans to raise prices. According to earlier coverage, TSMC is weighing a 15% increase on its advanced 3nm chips in the second half of this year. A further 10% increase is reportedly being considered for 2027. Neither figure has been officially confirmed, but Wei’s tone at the shareholder meeting did little to dismiss the prospect.

A Pattern Forming Across the Industry

TSMC is not alone in flagging these constraints. Dell raised similar concerns about AI infrastructure supply earlier this week, underscoring that demand-side pressure is hitting the full technology supply chain simultaneously. Analysts have noted that the bottleneck extends well beyond any single manufacturer or geography.

The energy dimension is also growing harder to ignore. TSMC has set a target to reduce chip power consumption by 30% as data center electricity demand accelerates with each new generation of AI workloads. The company is treating energy efficiency as a core engineering priority, not merely a sustainability talking point.

Background

TSMC has long dominated advanced chip fabrication, producing processors for clients including Nvidia, Apple, and AMD. Its Arizona expansion, backed partly by the CHIPS and Science Act, was intended to ease geopolitical risk and serve U.S. customers more directly. Wei’s remarks suggest that buildout, however significant, will not resolve supply pressures on any short timeline.

For markets tracking semiconductor exposure, the combination of constrained capacity, rising prices, and surging AI demand sets up a complex second half of 2026.

Read Next: Nvidia Earnings Preview: Wall Street Eyes AI Spending Signals