Figures from ASML’s Q2 2026 results (July 15, 2026), via ASML Investor Relations.
ASML’s above-guidance Q2 and 30% capacity expansion plan, read alongside TSMC’s record quarter and $60–64B capex, point to one thesis: the physical bottleneck on advanced silicon is expanding, not flinching.
What Happened
ASML reported second-quarter 2026 results on July 15 that came in above its own guidance on every headline metric. Net sales reached €9.3 billion, gross margin landed at 54.0%, and net income was €2.9 billion. The company sold 86 new lithography systems in the quarter, up from 67 in Q1 — a meaningful sequential acceleration. ASML then raised its Q3 outlook to €11–12 billion and lifted full-year 2026 guidance to €43–45 billion, citing AI-driven demand and customers accelerating their own capacity-expansion plans.
The more consequential signal came in the capacity announcements. ASML said it plans to expand its 2026 low-NA EUV system capacity — approximately 65 units — by 30% for 2027, and is currently investigating a further 30% addition for 2028. It flagged the same 30% expansion trajectory for its DUV immersion line (approximately 130 units). To be precise about what that language means: the 2027 expansion is a plan; the 2028 step is explicitly described as under investigation, not committed. One strong half-year is not a multi-year guarantee, and guidance remains the company’s own characterization of forward visibility.
Read ASML’s print alongside TSMC’s — reported a day later on July 16 — and the two results tell a single, mutually reinforcing story. TSMC posted record revenue, raised its full-year guidance, and lifted 2026 capex to $60–64 billion while committing an additional $100 billion to its Arizona buildout. ASML sits one step upstream, building the machines; TSMC sits downstream, running them to manufacture chips for Nvidia, Apple, and the hyperscalers. Both raised. Neither flinched. (Full TSMC analysis: TSMC Q2 2026 Capex Flip and TSMC’s $265B Arizona Commitment.)
The key insight: When the two physical chokepoints of global advanced-chip production — the tool-maker and the foundry — both raise guidance and expand capacity in the same week, that is a signal about the state of the cycle. It does not prove a multi-year demand floor; it does show that the companies with the most direct order-book visibility are not seeing a demand reversal. The market sold off the complex anyway, which is a different and important data point about how investors are repricing the AI trade.
The Structural Read
The most useful way to read ASML’s quarter is not as an earnings event but as a capacity-cycle signal. ASML does not sell commodity components. It sells EUV lithography systems — machines that no other company on earth manufactures — and its order book and delivery schedules are among the highest-resolution signals available for where advanced-node chip supply is actually heading. When ASML adds capacity, it is not responding to sentiment; it is responding to contracted demand from the handful of customers sophisticated enough to plan chip fabs multiple years in advance.
Paired with TSMC — which is simultaneously the world’s dominant advanced foundry and ASML’s most important EUV operator — the two results form a coherent picture. AI infrastructure spending has reached a scale and a multi-year commitment horizon that is now showing up in the physical production layer: more machines ordered, more fabs planned, more capex committed. The compute-demand dynamics driving this are structural, not episodic.
The market’s selloff on beat-and-raise results deserves its own reading. It is not evidence of a demand reversal. It is evidence of a repricing: investors are beginning to distinguish between companies that control scarce production capacity — grounded in order books, delivery slots, and multi-year backlogs — and companies that benefit from the AI narrative — grounded in sentiment and multiple expansion. Those are two different risk profiles, and the market is learning to price them differently.
Business Engineer — The Foundry Is the New Federal Reserve
“If the foundry is the new central bank of the AI economy — setting the real supply of compute the way a central bank sets the real supply of money — then ASML is the supplier of the printing press. You cannot run the press without the machine. And right now, the machine-maker is building more machines.”
That framing — developed in full at The Foundry Is the New Federal Reserve — helps explain why both ASML and TSMC behave differently from the rest of the AI complex in a selloff. Narrative beneficiaries get repriced on sentiment shifts. Capacity controllers get repriced on order books and delivery schedules. The moat is not a story; it is a physical constraint that takes years to replicate, if it can be replicated at all.
Three Implications
IMPLICATION 1 — THE TWO CHOKEPOINTS ARE BOTH RAISING
ASML and TSMC together represent the physical bottleneck on how much advanced silicon the world can produce. One makes the machines; the other runs them. If this were a genuine demand reversal, these would be the first companies to signal it — through order deferrals, capacity holds, or guidance cuts. Instead, both raised. That asymmetry between the market’s reaction and the chokepoints’ behavior is the most important data point in the results. Hedges apply: guidance is guidance, and one strong half does not lock in a multi-year trajectory.
IMPLICATION 2 — CONTROLLING SCARCITY IS THE STRUCTURAL MOAT
The market is undergoing a segmentation it has not fully priced. AI-narrative beneficiaries — companies whose valuations rest on the AI story holding — reprice on sentiment. Scarce-capacity controllers — companies whose revenues are grounded in multi-year order books and physical delivery constraints — reprice on fundamentals. ASML’s EUV monopoly and TSMC’s advanced-node dominance are not narrative advantages; they are structural ones. Investors learning to distinguish these two categories is a durable repricing, not a tactical rotation.
IMPLICATION 3 — THE NEXT STAGE IS PHYSICAL, AND THE WATCH ITEMS ARE SPECIFIC
Whether this expansion holds will not be decided by AI narrative or investor sentiment. It will be decided by four physical-layer variables: EUV delivery schedules against the expanded capacity plan; CoWoS advanced-packaging and HBM memory yields at TSMC and its supply partners; upstream materials capacity for the substrates and gases EUV requires; and whether financing costs stay contained enough to keep multi-year capex commitments intact. These are the seams where a phasing correction would show first — and they are the honest bracket around a thesis that otherwise looks compelling.
Where Each Layer Stands
Tool Layer — ASML
RAISING€9.3B Q2, above guidance. FY2026 lifted to €43–45B. +30% EUV and DUV capacity planned for 2027; 2028 under investigation.
Sources: asml.com · fourweekmba.com · fourweekmba.com · asml.com · globenewswire.com










