ASML Raised Guidance Again Because EUV Capacity Is Still the Binding Constraint in the AI Chip Cycle

ASML reported Q2 2026 total net sales of €9.3 billion, gross margin of 54.0%, and net income of €2.9 billion. The company then lifted 2026 sales guidance to €43-45 billion and gross-margin guidance to 54%-56%. That is the kind of raise that tells you demand is not the problem; capacity, installation cadence, and tool output are.

The most important line in the release is not the beat itself. It is the decision to expand low-NA EUV capacity by 30% in 2027 and potentially another 30% in 2028. That is a concrete admission that customers are still pulling harder than the system can ship. When the sole supplier of EUV says it needs more capacity, the entire AI chip stack has to re-price around a longer bottleneck.

The stock reaction was positive but not euphoric because the market is already used to AI-led capex stories. The deeper takeaway is that the investment cycle is no longer only about chip demand growth. It is about whether the tool chain can physically keep up.

Installed-base management revenue rose to €2.8 billion in the quarter, which matters because it turns the installed fleet into an annuity-like revenue stream. Net system sales were also strong, and the company disclosed 86 new lithography systems sold in Q2 versus 79 in Q1. That implies the customer build-out is still active, not rolling over.

ASML also said its order intake remained extremely strong in the first half of the year. That matters because order strength is the leading indicator for the chip equipment group. By the time equipment revenue shows up, the demand decision was already made months earlier by the foundry, memory, or logic customer.

The underlying pattern is consistent with a market that needs more advanced logic, more HBM, and more memory capacity. The AI boom is not just consuming chips. It is forcing process nodes, wafer starts, and lithography exposures to move higher than pre-AI planning assumed.

The obvious beneficiaries are TSMC, Nvidia, AMD, Apple, and Broadcom, because every higher wafer-start plan eventually flows into their supply chain. But the less obvious beneficiaries are the equipment and process players that sit behind ASML: Applied Materials, Lam Research, KLA, Tokyo Electron, and the various subsystem and parts vendors that get paid when factory spending extends another year.

The read-through to Korea matters too. Samsung Electronics and SK Hynix both need advanced capacity and more memory output, but each extra round of capex also tightens the competition for tools, engineers, and fab-ready capacity. That does not mean the market is structurally bearish on Korea. It means the marginal dollar of expansion is getting more expensive.

The strategic point is simple: if ASML needs to raise supply in 2027 and 2028, then the AI chip buildout is still in the long-capex phase. That is bullish for the equipment stack and supportive for the leading foundry, but it also means investors should not assume the cycle has peaked just because the stocks have already moved.

Watch the next quarter's system-sold count, the commentary on backlog conversion, and any update on EUV capacity ramp timing. Those are the variables that tell you whether the present raise was conservative or simply the next step in a still-accelerating cycle.

Watch TSMC and Samsung Electronics capex commentary. If their spending stays firm, ASML's outlook has room to move again.

Watch the broader semi tape. If the market still sells chips on any sign of AI volatility, ASML's own guidance becomes the best argument that the cycle is not being driven by headlines, but by actual fab requirements.