ASML Holding

Refresh delta (since 2026-06-18)

What changed

1. Q2 2026 actuals vs. the guidance ASML gave in April. ASML guided Q2 2026 to €8.4–9.0B net sales / 51–52% gross margin at the Q1 2026 call (2026-04-15) ``. Actual:

CEO Christophe Fouquet, on the driver: "second-quarter total net sales were €9.3 billion and gross margin came in at 54.0%, both above guidance, driven primarily by higher than expected Installed Base Management sales" [primary: asml.com, press release, 2026-07-15]. On the size of that specific beat, per the earnings-call transcript: "The installed base business came in at EUR 2.8 billion. That is EUR 300 million more than we expected as a result of the fact that customers are really looking for productivity enhancements" ``. The exact filing line is Installed Base Management sales of €2,762M [primary: asml.com]; the CEO's rounded "€2.8B / €300M more than expected" characterization implies an internal expectation near €2.46B — note the rounding, both figures are reported here rather than reconciled to the decimal.

Read on the beat's composition: this was a service/mix beat, not a new-systems beat — software-led upgrade sales that customers wanted for immediate productivity gains without machine downtime [primary: asml.com]. That matters for how much this print says about the forward EUV order book (see linkage paragraph below).

2. FY2026 guidance — the raise, in full, since the 20-F. Three data points now exist for FY2026, and this is the second raise:

Midpoint math ``: 20-F €36.5B → Apr €38.0B → Jul €44.0B. Since the original 20-F guide: +€7.5B / +20.5% at the midpoint. Since the April raise alone: +€6.0B / +15.8% at the midpoint, with the gross-margin band stepping up a full +3pp at both ends (51%→54%, 53%→56%). This is a materially larger step-up than the April raise (which moved the sales midpoint only +€1.5B / +4.1%).

3. Q3 2026 guide. €11.0–12.0B net sales (midpoint €11.5B, +23.3% QoQ vs. Q2 actual ``), gross margin 55–57%, R&D ~€1.2B, SG&A ~€0.4B [primary: asml.com].

4. Capacity — the roadmap moat reinforced. Management confirmed intent to add 30% to 2026's ~65-unit Low-NA EUV capacity for 2027, with a further 30% under investigation for 2028; same 30%-then-30% pattern for DUV immersion capacity off a ~130-unit 2026 base . FY2026's ~65 planned EUV units (up from 48 EUV shipped in FY2025 ) implies +35% EUV unit growth YoY `` — this is a capacity/output signal, not a bookings number (see next point).

5. Bookings/EUV order detail — still not disclosed quantitatively. Consistent with the June dossier's Lens 6 finding that ASML stopped publishing quarterly net bookings, no Q2 bookings figure was given. Forward-visibility color from the call: "starting with 2027, there we are pretty much already close to receive all the orders we need for 2027 … When we look at 2028, we have received already a large number of orders from our customers for EUV" ``. Qualitative and unquantified — treat as management color, not a hard order-intake number.

6. China. Management: "China. We still look at China as approximately 20% of our total net sales, so that percentage hasn't changed … primarily in the logic business, primarily catering to domestic-led demand" . Caution on comparability: this is **total net sales** share; the June dossier's most recent China figure (19%, Q1'26) was specifically **system sales** share, and FY2025's 29.1% was total net sales . Directionally consistent with the ongoing decline from 36% (FY2024) but the metrics aren't identical — don't chain them into one series without normalizing. No MATCH Act commentary appeared in the press release or the transcript excerpts reviewed here — that overhang (Lens 8/13 in the June dossier) is neither confirmed nor resolved by this print; still an open item.

What held

The structural thesis from the previous dossier is unchanged by this print, not undermined by it: AI demand is still broadening (this quarter's evidence is a service/upgrade beat plus a large FY raise, both consistent with "demand broadening" rather than "one-off lumpy order"), the capacity build-out continues on schedule (+30%/+30% Low-NA EUV and DUV additions), and the China de-emphasis trend continues without a new negative catalyst this quarter. The one thing this print does NOT do is resolve the bear case's two live threats — the MATCH Act (silent this quarter) and TSMC's High-NA delay (not addressed in the transcript excerpts pulled here) — both remain open per the June dossier and are out of scope for this compact delta to re-litigate.

Linkage note — TSMC capex and ASML's order book (mechanical, not directional)

TSMC reported Q2 2026 results the same week (2026-07-16): record quarterly revenue of NT$1.27 trillion (~$40.2B, +33.7% YoY in USD terms), CoWoSCoWoSTSMC’s method of packing a processor and its memory onto one carrier so they sit close together. Supply of it has been a hard limit on how many AI chips can be built. packaging reportedly sold out through year-end, and capital expenditureCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. guided toward the high end of its existing $52–56B 2026 range ``.

The mechanical link: TSMC is one of ASML's four >10%-concentration customers, and Taiwan (essentially TSMC) rose from 15.4% to 25.5% of ASML's total net sales in FY2025 as TSMC's AI/leading-edge capacity ramped . TSMC's capex line is, structurally, the demand signal that eventually becomes ASML's system order book: TSMC capex funds fab buildout → fab buildout requires litho tool purchases (EUV/DUV) and installed-base upgrades → those purchases flow through ASML's contract-liability / advance-payment structure (€19.4B in customer prepayments at FY2025 year-end ) before recognizing as ASML system or service revenue, typically with a multi-quarter lag given ASML's 12–18-month build cycle ``.

Two things this print does and does not establish. It does establish that the largest customer in ASML's book is spending at the top of its guided range in the same window ASML raised its own FY guide by €7.5B and flagged capacity expansion through 2027–28 — the two data points move in the same direction, which is what you'd expect if the order pipeline underneath both companies is durable rather than digesting. It does not establish that TSMC's specific incremental capex dollars caused ASML's specific Q2 beat — that beat was, per the CEO's own framing above, an Installed-Base/service-mix story this quarter, not a new-EUV-systems story. The TSMC linkage is a read on pipeline durability into 2027–28 (where ASML's own forward-order commentary points), not an explanation of what actually moved Q2's number. No directional call is made here on whether that durability holds; it is a structural observation about how the two companies' guidance moved in the same window, nothing more.

---

Not touched by this delta

Per task scope: _meta.json (last_updated stays 2026-04-22), our figures/our figures/our figures/our figures/our figures (still headers-only, unchanged), nextCatalystAt/coverage-registry state, and any MarketCall row. This file is the dossier's own last-refresh stamp by virtue of its dated filename, per the shelf convention — no other file in the research layer was modified. No commits, no DB writes, no publishing.