ASML Expands Production, TSMC Doubles Down: Why Is the Market Still Unsatisfied with the 'Second Wave' of AI Chips?

The expansion cycle for AI semiconductors is far from over, but under high expectations, "strong demand" no longer automatically constitutes a new catalyst for price increases.

Written by: Jim, MSX Maitong

Edited by: Frank, MSX Maitong

In the past week, the pressured AI hardware sector has been waiting for a strong stimulus to reverse the sentiment.

What ASML and TSMC subsequently delivered were financial reports with sufficiently strong fundamentals, yet they failed to fully meet high expectations: the former significantly raised its full-year revenue and gross margin guidance and began to increase lithography machine production capacity for 2027-2028; the latter maintained historical highs in revenue, gross margin, and operating profit margin while raising its full-year capital expenditure to $60 billion to $64 billion.

Logically, this should be the ideal combination for AI semiconductors—equipment companies proving that customers are still placing orders, and the leading foundry proving that orders are being converted into revenue and willing to continue investing heavily in expansion.

However, the market's feedback did not fully match the strength of the performance.

The reason is not that the fundamentals of the two companies have deteriorated, but that the expectations for the AI industry chain have been pushed to an unusually high level. The market is no longer satisfied with "demand remains strong"; instead, it hopes that every financial report continues to be revised upward, every profit margin breaks limits, and all massive capital expenditures can be immediately converted into higher profits.

This has led to ASML and TSMC's financial reports conveying two seemingly contradictory yet simultaneously valid signals: the expansion cycle of AI semiconductors is still ongoing, with some key links even accelerating; however, the capital market's pricing for this cycle has shifted from verifying demand to verifying returns.

1. ASML and TSMC Both Double Down: The Expansion Cycle Is Far from Over

ASML was the first to unveil the answers for this earnings season.

The company's net sales in the second quarter reached €9.326 billion, exceeding the previous guidance of €8.4 billion to €9 billion; the gross margin reached 54%, with a net profit of €2.918 billion. The company subsequently raised its third-quarter sales guidance to €11 billion to €12 billion and significantly raised its full-year sales forecast for 2026 from €36 billion to €40 billion to €43 billion to €45 billion.

More importantly than the quarterly data is that ASML has begun to adjust its equipment production capacity for the next two years. The company plans to increase its low numerical aperture EUV capacity by 30% from about 65 units in 2026, and the DUV immersion equipment will also increase by 30% from about 130 units; at the same time, ASML is also researching the possibility of further expansion in 2028.

The lithography machine supply chain is complex, and the delivery cycle is long. ASML will not rashly increase production capacity two years later based on fluctuations in orders from one or two quarters. Such an expansion plan indicates that wafer fab customers are locking in advanced processes and high-end storage capacity for 2027-2028 in advance.

A day later, TSMC provided corresponding validation from the wafer manufacturing end.

The company achieved revenue of $40.2 billion in the second quarter, a 12% quarter-on-quarter increase, at the upper limit of the previous guidance of $39 billion to $40.2 billion; the gross margin reached 67.7%, slightly above the upper limit of the guidance, and the operating profit margin reached 60.3% for the first time. The net profit was NT$706.56 billion, a year-on-year increase of 77.4%, with earnings per share of NT$27.25.

The revenue structure continues to tilt towards AI and advanced processes. In the second quarter, revenue from high-performance computing business grew by 20% quarter-on-quarter, accounting for 66% of the company's revenue; advanced processes of 7nm and below accounted for 77% of wafer revenue, with 3nm and 5nm contributing 30% and 33%, respectively, while the 2nm process, which has just entered the ramp-up phase, contributed 3% of wafer revenue for the first time.

More significantly, capital expenditure is still on the rise. TSMC raised its capital expenditure plan for 2026 from the original $52 billion to $56 billion to $60 billion to $64 billion. About 70% to 80% of this will be used for advanced processes, and about 10% to 20% will be allocated to advanced packaging, testing, photomask manufacturing, and other areas.

The company also raised its full-year dollar revenue growth forecast from previously over 30% to slightly above 40%. Management stated that AI-related demand remains extremely strong, and demand signals from cloud service providers and downstream customers remain positive.

ASML is preparing to increase lithography equipment capacity, while TSMC is expanding wafer manufacturing and advanced packaging capabilities through higher capital expenditure.

So when both the equipment leader and the world's largest foundry simultaneously raise their future investments, at least one thing can be confirmed: AI semiconductor capital expenditure has not entered a contraction cycle; the industry chain is even still accelerating to prepare capacity for future demand over the next few years.

2. With Such Strong Performance, Why Does the Market Still Feel It's Not Enough?

The problem is that the market is waiting for more than just a "goal achieved" financial report.

Since TSMC publishes revenue data every month, the second-quarter revenue of $40.2 billion has already been largely digested by the market. Therefore, before the financial report was released, the real expectation gap was in the gross margin, third-quarter guidance, and how much capital expenditure could be raised.

From this perspective, TSMC's second-quarter gross margin of 67.7%, although higher than the company's previous guidance of 65.5% to 67.5%, is only roughly in line with the mainstream expectations that the market has revised upward, failing to meet some investors' aggressive judgments of nearly 69% or even higher.

For the third quarter, the company expects revenue to be between $44.6 billion and $45.8 billion, which would represent a quarter-on-quarter growth of about 12% based on the midpoint; however, the gross margin guidance has decreased to 65% to 67%, with a midpoint of about 66%.

The decline in gross margin does not mean that demand is weakening.

TSMC expects that the rapid ramp-up of the 2nm process will dilute the gross margin by about 3 to 4 percentage points in the second half of the year; overseas wafer fabs' expansion will also continue to increase depreciation and manufacturing costs. Strong demand for advanced processes, high capacity utilization, and improvements in manufacturing efficiency can only partially offset these pressures.

In other words, TSMC is facing a typical paradox of high prosperity expansion— the stronger the demand, the more the company needs to procure equipment, build wafer fabs, and introduce new processes in advance; and the higher the capital expenditure, the earlier depreciation, overseas production costs, and new node ramp-up pressures will reflect in the profit margins.

This is also the most important aspect to understand about this financial report.

From the management's statements on pricing strategy during the earnings call, TSMC does not pursue pushing short-term gross margins to the limit during the tightest supply periods. The company emphasizes that it is a long-term partner of its customers and will not suddenly raise prices significantly to squeeze customers, but rather hopes to maintain a profit level sufficient to support long-term expansion.

This means that TSMC currently prefers to maintain a balance between pricing power, customer relationships, and continuous expansion, rather than cashing in all scarcity premiums at once. From an industry perspective, this is undoubtedly a positive signal, but from a short-term trading perspective, it means that investors need to accept a reality: strong AI demand does not mean that every dollar of new revenue can immediately translate into higher profit margins.

Therefore, the market's lukewarm reaction to TSMC's financial report cannot be simply understood as a peak in AI demand; a more accurate explanation is that in an environment where expectations have already soared, strong performance is becoming a necessary condition for valuation but no longer automatically constitutes a new catalyst for price increases.

After the financial report was released, strong performance did not immediately translate into sustained sector growth, reflecting that investors are digesting profit margin pressures and overly high expectations.

3. Only by Putting ASML and TSMC Together Can We See the 'Second Wave' of AI Chips Clearly

If we look at ASML and TSMC's financial reports together, the outline of the so-called 'second wave' of AI chips is already clearer than before.

It is not a return to a comprehensive shortage of "all chips are insufficient" nor a simple replication of the past two years' market centered around NVIDIA GPUs, but rather that supply bottlenecks continue to spread to the entire AI system.

ASML's EUV and DUV equipment determine how quickly advanced processes can expand; TSMC's 3nm and 2nm determine how much wafer capacity GPUs, CPUs, and custom ASICs can obtain; HBM determines memory bandwidth; advanced packaging such as CoWoS determines whether computing chips, storage, and high-speed interconnects can ultimately be combined into deliverable data center products.

Any insufficient expansion in any of these links will slow down the shipment of the entire AI system.

TSMC's management even explicitly stated that the current advanced packaging capacity is already tight enough to limit customer growth, and the company is working hard to narrow the gap between demand and capacity while welcoming other packaging solutions to provide additional options for customers.

At the same time, AI demand is also spreading from single accelerators to a broader range of chip types.

TSMC believes that the development of Agentic AI is re-emphasizing the importance of CPUs in data centers. Regardless of whether customers adopt x86, Arm, or RISC-V architectures, the advanced chips behind them still mostly need to be manufactured by TSMC. This means that future AI capital expenditure will not only flow to GPUs but will also continue to drive demand for CPUs, networking chips, storage, and advanced packaging.

Management's expression of long-term demand is also positive. TSMC believes that AI-related trends will remain strong until 2029-2030, during which time there may be phase fluctuations, but the long-term direction has not changed. For the previously given judgment of a mid-50% compound growth rate in AI-related business, management did not provide new specific numbers, only stating that the demand trend is stronger than previously expected.

However, this does not mean that all semiconductor companies will benefit equally.

• ASML (ASML.M) directly benefits from the demand for lithography equipment and the expansion of advanced processes;
• Applied Materials (AMAT.M), Lam Research (LRCX.M), and KLA (KLAC.M) benefit from demand for deposition, etching, and inspection equipment, but the order fulfillment pace may vary;
• TSMC (TSM.M) controls advanced wafer manufacturing and packaging capabilities, making it the core recipient of AI chip capacity expansion;
• SK Hynix (SKHY.M), Micron (MU.M), and Samsung Electronics supply HBM and high-end storage;
• NVIDIA (NVDA.M), AMD (AMD.M), Broadcom (AVGO.M), as well as cloud vendors with self-developed chip capabilities such as Amazon (AMZN.M), Alphabet (GOOGL.M), Microsoft (MSFT.M), and Meta (META.M) ultimately determine how fast end demand can grow.

They are all in the same capital expenditure cycle but have completely different technological barriers, capacity constraints, profit structures, and valuation levels. Therefore, the 'second wave' of AI chips is more likely to be a structural market rather than a synchronized rise of the entire hardware industry chain.

In the next phase, the market will pay more attention to which companies truly possess irreplaceable scarce capacity, which companies are merely following customers in increasing capital expenditure, and which companies can continue to improve free cash flow and capital return after expansion.

After ASML and TSMC, the next key validation will also fall back on cloud service providers such as Microsoft, Amazon, Google, and Meta. Ultimately, the willingness of equipment companies to expand, the willingness of foundries to invest, still requires cloud vendors to continue increasing capital expenditure and prove that the increasingly large AI infrastructure can generate real model calls, corporate revenue, and cash flow returns.

In Conclusion

Objectively speaking, ASML answered whether wafer fabs are still willing to purchase equipment, while TSMC further proved that customer orders are sufficient to drive the company to continue increasing wafer and advanced packaging capacity.

From this perspective, the industry cycle of AI semiconductors has not peaked.

After all, equipment capacity is expanding, advanced packaging remains tight, and TSMC has even raised its full-year capital expenditure to a maximum of $64 billion, all of which are not signals released by an industry preparing to contract.

However, the reason the market still feels it's not enough is that the questions for the next phase have changed. In the past, investors needed to confirm whether AI demand truly existed; now, demand is hard to deny, and the market wants to know how much capital needs to be invested to meet these demands and how high these investments can ultimately translate into profits and cash flow.

Therefore, the 'second wave' of AI chips may have already begun, but it will not simply be a replay of the first round of the market.