Revisiting Our ASML Thesis After the Q1 2026 Backlog Print

The prior thesis argued three points. First, the High-NA EUV roadmap was decoupled from the customer adoption curve; Intel's High-NA fab plans had been pushed out, TSMC's pilot deployment timeline had slipped, and Samsung's logic node migration had stalled. Second, the FY2026 EUV system shipment guide of 90-100 units embedded a customer demand profile that was not yet visible in the order pipeline. Third, the multiple at 48x trailing earnings was paying for High-NA acceleration that the data did not support.

The call-out at the time was that the consensus 2027 EPS estimate of $42 required an EUV unit shipment cadence and a High-NA mix that had no precedent in the deployment data. The fair value range was set at $1,250-1,400, against a then-current price of approximately $1,420. The implication was that the multiple needed to compress 8-12% before the risk-reward inverted.

The thesis was not a sell call. It was a recommendation to reduce position size and wait for either a multiple compression entry or operational evidence that the High-NA timeline had stabilised. The Q1 print partially answered the second condition; deep-UV strength and Korean DRAM order acceleration are operational positives that the prior thesis did not fully weight.

Three pieces of the Q1 print update the model. First, deep-UV system orders were stronger than expected, supporting a $4.5-5.0 billion FY2026 deep-UV revenue contribution, roughly 8% above the prior model. Deep-UV is a lower-margin business than EUV but the volume and the geographic mix give the segment a defensive characteristic that supports the consolidated cash flow line.

Second, the EUV ASP step-up was confirmed for the next-generation NXE:3800 platform shipping into TSMC's N2 ramp. The ASP increase of approximately 12% on like-for-like specifications adds roughly €0.6-0.8 billion to the FY2026 EUV revenue line. This is a margin-accretive development; the next-gen platform carries a higher gross margin than the legacy NXE:3600 series.

Third, the China revenue contribution stayed at the lower end of the guided range. Export controls on the most advanced lithography systems have constrained the Chinese deep-UV business, but the legacy immersion business has held up better than the bear case modelled. China revenue contributed €5.2 billion in 2025, against the €4.5 billion bear-case scenario. The runway for legacy immersion sales into China is finite but the cash flow contribution while it lasts is meaningful.

The one piece that did not improve was High-NA. Cumulative High-NA shipments to customers remain in the high single digits. Customer commentary on the deployment timelines has been carefully ambiguous. The capital cost per High-NA system, at approximately €380 million, is roughly twice the cost of the latest standard-NA EUV system. The customer ROI calculation requires a node migration that produces meaningfully better wafer economics, and the early data points from customer pilots have been mixed.

Consolidated 2025 revenue printed at $32.7 billion, operating income at $11.3 billion, and net income at $9.6 billion. The operating margin held at 36% despite the High-NA development costs running through the cost base. Free cash flow grew to $10.6 billion, a 12% expansion year-on-year. The forward earnings multiple at 41.2x implies the market is paying for low-twenties EPS growth over the next 24 months. The Q1 print supports that earnings trajectory more than the prior thesis assumed.

The updated 2026 EPS estimate is $35.50, up from $34.00 in the prior thesis. The 2027 EPS estimate is $40.50, down from $42.00. The reshape of the trajectory captures the deep-UV strength in the near term and the slower High-NA contribution in the back end. Apply a 38x forward multiple, which is below the 5-year median of 42x but above the post-2018 trough of 28x, and fair value sits in the $1,350-1,500 range.

The trailing twelve months has seen the share price compress from $1,547 to roughly $1,400. The compression has restored some of the multiple cushion. The 50-day moving average sits at $1,403, the 200-day at $1,101. The technical setup is firmer than the underlying customer concentration concern would suggest. Beta of 1.38 is elevated relative to the broader equipment peer group but reflects the recent operational volatility rather than a structural change in the business.

Three pieces re-weight the model. First, the customer concentration is worse than the prior thesis assumed. With three customers driving over 70% of new bookings, the order book is structurally more volatile and the visibility into the FY2027 trajectory is dimmer. Second, the deep-UV business is more resilient than the prior model gave credit for, supporting a higher cash flow floor even if the High-NA timeline continues to slip. Third, the absolute multiple has compressed enough to bring the entry zone into the constructive range; below $1,300 the asymmetry is favourable, between $1,300 and $1,500 it is balanced.

The historical pattern in semiconductor equipment is well-defined. When the leading-edge cycle slips, the deep-UV and legacy systems business carries the cash flow line through the gap. ASML executed exactly this sequence in 2018-2019 when the EUV ramp slipped relative to the original timeline. The cash flow held up, the multiple compressed temporarily, and the next leg of the EUV deployment cycle delivered the multiple expansion. The current setup rhymes with that earlier cycle.

Look, the High-NA delay is real and the customer mix is uncomfortable. Both are pricing in. The cash flow line is producing $10+ billion annually, the buyback authorisation has runway, and the dividend is covered with room to grow. The setup does not need to be a hero call; it needs to be a discipline call. We are sized smaller than the prior thesis and waiting for either a print below $1,300 or a TSMC High-NA capacity announcement to add.

Three-customer concentration at 71% of new orders is the kind of structural feature that does not reverse in a single quarter. TSMC, Samsung, and Intel together drive the vast majority of leading-edge logic and DRAM EUV capacity decisions. If any single one of those three pauses or paces capacity additions differently than expected, the impact on the ASML order book is immediate and outsized. The historical analogue is Applied Materials in 2008, where customer concentration in the memory tier produced a single-quarter bookings drop of 38% as Hynix and Samsung paused capacity additions in lockstep. The recovery took 18 months and included a multiple compression of approximately 30%.

The defensive feature in ASML's case is the deep-UV revenue line, which has a broader customer base across the 200+ active fab sites globally. Deep-UV does not generate the leading-edge ASP economics, but the sheer breadth of the customer base diversifies the order book volatility. Stripping the leading-edge concentration risk and the deep-UV defensive base apart in the model produces a more nuanced view; the leading-edge component deserves a discount for concentration, the deep-UV component deserves a premium for resilience. The blended multiple should sit modestly below the 5-year median rather than at it. That is consistent with our $1,350-1,500 updated fair value range.

A piece of the volatility we are watching is the China legacy lithography demand profile. The trade restrictions have constrained the leading-edge sales pipeline but the legacy immersion business has continued to grow as Chinese fabs build out mature node capacity. China contributed approximately €5.2 billion of revenue in 2025, materially above the bear-case modelling. The runway depends on policy continuity. A meaningful tightening of the export control envelope could remove €2-3 billion of revenue at short notice. We are not modelling that as the base case but it remains a measurable risk.

Capital allocation tells you what management actually believes. ASML's buyback execution rate accelerated through Q1 2026 to roughly €1.4 billion, the highest single-quarter pace in the company's history. The implied annualised pace of €5-6 billion would retire approximately 1.0-1.2% of shares outstanding per quarter at current prices. That is meaningful capital return given the buyback authorisation history of the company. The acceleration is itself a signal; management is treating the current price as accretive on a forward earnings basis. The Capital Desk view is that this signal carries real informational weight when paired with the dividend policy continuity.

The dividend policy has been stable through the share price volatility. The 2025 dividend per share grew 5% year-on-year and the 2026 interim dividend was set at the upper end of the policy band. Combined dividend and buyback yield at current prices runs in the 3.5-4.0% range, which is competitive against the broader semiconductor equipment peer group and well above the European technology sector average. Capital return is doing real work in the total return profile.

The other piece of capital allocation that matters is the High-NA investment cycle. ASML has been investing approximately €1.5-1.7 billion annually in High-NA development and the new manufacturing capacity at Veldhoven. The investment cycle has another two years to run before it normalises. The argument for the investment is durable; even if the High-NA adoption curve runs slower than the consensus model assumes, the eventual deployment is essentially certain given the physics of next-generation logic node lithography. The question is timing rather than terminal value. The investment cycle is correctly continuing despite the near-term slippage.

The position sizing implication of the update is the most actionable part. The prior thesis recommended reducing position size to about half of the long-run target weight while waiting for either a multiple compression entry or High-NA stabilisation evidence. The Q1 print gives partial credit on operational stabilisation, with deep-UV strength offsetting the High-NA delay, while the multiple has compressed roughly 8% from the prior thesis publication. Together, the two-part adjustment moves position sizing from half-weight to about 70% of the long-run target. We are not back to full conviction. We are incrementally adding rather than waiting flat.

The execution mechanics matter. Adding into the $1,300-1,400 zone over multiple tranches is the disciplined approach. Single-tranche adding at the current price exposes the position to the customer concentration risk we just flagged. The catalyst events over the next 90 days, the Q2 booking print, the TSMC capex commentary, and the Intel 18A ramp update, each provide an opportunity to either confirm or reverse the constructive update. Sizing into those catalysts rather than ahead of them is the better risk-adjusted approach.

Across two complete EUV deployment cycles, the pattern is consistent; the multiple compresses through deployment slippage windows, the cash flow line carries the value, and the next-generation customer commitments produce the re-rating. We are at the compression phase of the third cycle. The data does not yet support full conviction but it supports more conviction than the prior thesis encoded.