My Inquiry
My Account
TSMC Chairman and CEO C.C. Wei used the company's latest shareholder meeting to send a clear message to the semiconductor market: AI chip demand remains stronger than available supply, and TSMC is still working to expand capacity fast enough to satisfy customers.
Wei said AI demand is still in a state of supply constraint, while TSMC is trying to increase capacity so customers can receive more reliable support. He also said the company's business remains strong and that both revenue and earnings are expected to reach new highs this year. Reuters reported that Wei described AI demand as showing no sign of easing, while also saying TSMC would like to raise prices but avoid abrupt moves like some memory suppliers (Reuters, TSMC boss upbeat as AI boom shows no sign of easing).
The comments matter because they touch three of the most important questions in the semiconductor market today: whether AI demand is real, whether advanced foundry capacity remains tight, and whether TSMC has enough pricing power to pass higher costs back to customers. Wei's answer was not aggressive, but it was direct. AI demand remains strong, capacity is still not enough, and TSMC is working on pricing while keeping long-term customer partnerships intact.
C.C. Wei's Key Messages at the Shareholder Meeting
Wei's remarks were not framed as a formal price increase announcement. Instead, they showed how TSMC is balancing strong demand, customer relationships, capacity investment, and pricing discipline. That makes the comments more important than a simple price notice.
| Topic | Wei's Message | Market Meaning |
|---|---|---|
| AI demand | AI demand remains supply-constrained | Advanced foundry and packaging demand remains strong |
| Capacity | TSMC is working to increase capacity | Customers still need more AI chip supply |
| Growth outlook | Revenue and earnings are expected to reach new highs | AI demand is supporting growth visibility |
| Pricing | TSMC is working on pricing | Foundry pricing power is improving |
| Operating style | TSMC will manage pricing in an orderly way | Price adjustments may be gradual rather than abrupt |
| Main risk | Global economic downturn remains the key concern | AI is strong, but macro demand remains a risk factor |
AI Demand Keeps TSMC in a Strong Position
TSMC sits at the center of the AI semiconductor supply chain. Leading AI GPUs, custom AI accelerators, networking chips, and high-performance processors depend on advanced foundry manufacturing. As AI servers become more complex, demand does not stop at wafer production. It also extends to advanced packaging, high-density interconnect, power delivery, substrates, test capacity, and memory integration.
This explains why Wei emphasized capacity expansion rather than demand uncertainty. For TSMC, the main issue is not whether AI customers want more chips. The issue is whether the company and its upstream suppliers can expand enough capacity to support the next wave of AI infrastructure.
AI demand also has a multiplier effect across the supply chain. More AI accelerators require more advanced wafers, but they also require high-bandwidth memory, server DRAM, power semiconductors, advanced substrates, and cooling systems. This is closely connected to the wider trend discussed in our analysis of memory chip shortages driven by AI servers, HBM, DDR5, and CXL demand.
TSMC's Pricing Power Is Returning, But Not Like a Memory Cycle
One of the most interesting parts of Wei's remarks was his response to the suggestion that TSMC should raise prices more aggressively because its customers are making strong profits. Wei said he would like to do so and is working on it. But he also emphasized that TSMC should not suddenly raise prices in the same way some memory companies have done.
That distinction is important. Memory pricing is often more cyclical because parts of the DRAM and NAND markets have commodity-like characteristics. When supply tightens, prices can move quickly. When inventory builds, prices can fall just as quickly. TSMC's business is different. It is built around long-term process roadmaps, customer-specific product planning, and deep manufacturing partnerships with fabless chip designers.
A foundry price increase therefore carries different implications from a memory price increase. TSMC is not simply selling capacity into a spot market. It is supporting customers across multiple generations of chip design, process migration, yield improvement, advanced packaging, and production ramp. Abrupt pricing could damage customer relationships even when demand is strong.
| Category | Memory Suppliers | TSMC |
|---|---|---|
| Pricing behavior | More cyclical and sometimes faster repricing | More disciplined and relationship-based |
| Product nature | More commodity-like in some segments | Customer-specific technology and process partnership |
| Customer relationship | Often volume and cycle driven | Long-term roadmap and platform driven |
| Capacity constraint | HBM, DRAM, NAND, packaging and test | Advanced nodes, CoWoS and leading-edge packaging |
| Pricing risk | Sharp price swings can happen | Gradual adjustment is more likely |
Why Foundry Pricing Power Matters for the AI Chip Market
Foundry pricing power is becoming more important because AI chips are among the most demanding products in semiconductor manufacturing. Advanced AI accelerators require leading process nodes, large die sizes, complex packaging, high yield control, and close coordination between foundry, packaging, memory, substrate, and system customers.
When demand is stronger than supply, foundries gain more negotiating power. However, TSMC must balance that power carefully. Its largest customers rely on long-term capacity commitments, while TSMC relies on those customers to support new process nodes and packaging platforms. This mutual dependence explains why Wei emphasized both profitability and partnership.
The pricing question also affects fabless chip companies. If wafer and advanced packaging costs increase, chip designers may face margin pressure unless they can pass higher costs to cloud customers, AI server makers, or end users. In the AI market, customers may be more willing to absorb higher costs because advanced chips are still a bottleneck for compute expansion.
CoWoS and Advanced Packaging Remain Critical Bottlenecks
TSMC's AI capacity challenge is not only about advanced wafers. Advanced packaging is now a major part of AI chip supply. Technologies such as CoWoS allow high-performance logic chips to be integrated with HBM and other high-bandwidth interfaces. Without enough advanced packaging capacity, wafer output alone cannot satisfy AI accelerator demand.
This is why the market increasingly treats advanced packaging as part of the AI infrastructure supply chain. AI chip production requires a full stack: advanced logic wafers, HBM, interposers, substrates, packaging, test, power delivery, and server integration. A shortage in any one layer can slow shipments.
Wei's comments on supply constraint should therefore be read broadly. The bottleneck is not only one process node or one fab. It is the combined capacity of advanced semiconductor manufacturing and packaging ecosystems.
Global Economy Is the Main Risk to the Outlook
Wei said the main concern is not AI demand, but a possible global economic downturn. That is a realistic risk. AI infrastructure remains strong, but the semiconductor market is still exposed to broader economic conditions through consumer electronics, industrial spending, automotive demand, and enterprise technology budgets.
If the global economy weakens sharply, some non-AI semiconductor categories could slow. Consumer electronics and parts of the industrial market are more sensitive to macro conditions. Automotive semiconductors could also face pressure if vehicle demand weakens.
However, the AI cycle is currently strong enough to separate leading-edge foundry demand from weaker areas of the semiconductor market. That difference is important. It means TSMC may continue to grow even if some mature or consumer-linked semiconductor segments remain uneven.
How TSMC's Approach Differs From Recent Chip Price Hikes
Recent semiconductor price increases from IDMs and memory suppliers show that pricing pressure is spreading across multiple parts of the chip market. Renesas, for example, has also planned price adjustments starting in July 2026, reflecting higher material, transportation, and supplier costs. That trend was covered in our previous report on the Renesas price increase in July 2026.
TSMC's situation is different. The company is not only responding to cost inflation. It is also managing a structural shortage in advanced AI manufacturing capacity. That gives TSMC stronger pricing power than many semiconductor suppliers, but it also requires more careful execution.
A sudden, aggressive price move could create tension with customers that depend on TSMC for multi-year technology roadmaps. A gradual, disciplined pricing strategy may allow TSMC to improve returns while preserving customer trust. This is why Wei's comments are important: they indicate pricing power, but also restraint.
What TSMC's Comments Mean for the Semiconductor Market
For the broader semiconductor market, Wei's remarks point to a stronger foundry cycle led by AI. The industry is no longer only dealing with temporary shortages or inventory swings. It is dealing with structural demand for advanced compute, memory, packaging, and power infrastructure.
This cycle could benefit multiple parts of the supply chain. Advanced equipment suppliers, substrate makers, packaging providers, memory vendors, power semiconductor companies, and high-performance component suppliers may all see continued demand linked to AI infrastructure growth.
At the same time, the market should not assume that every chip company has the same pricing power. TSMC's position is unique because it controls a large share of the world's leading-edge foundry capacity. The pricing environment for commodity memory, mature-node chips, analog ICs, and consumer semiconductors may behave differently.
Key Takeaways for the Foundry and AI Chip Supply Chain
- TSMC says AI demand remains stronger than available supply.
- The company is expanding capacity to support customers, but the AI supply chain remains tight.
- Foundry pricing power is returning, but TSMC appears focused on disciplined and orderly pricing.
- TSMC's pricing logic differs from memory suppliers because foundry relationships are more roadmap-driven.
- Advanced packaging, especially CoWoS-type capacity, remains a key AI chip bottleneck.
- The main risk to TSMC's outlook is a broader global economic downturn rather than weak AI demand.
Wei's comments show that TSMC is operating from a position of strength, but not recklessness. AI demand remains strong, customer demand still exceeds supply, and pricing power is improving. The company's challenge is to capture more value from the AI boom while maintaining the long-term customer trust that made it the most important foundry in the global semiconductor industry.
