Executive Summary
The record supports a more useful conclusion than the catalog's instructor-assigned
success label. ASML did execute the customer share issuances and synthetic buyback,
recognized at least some CCIP contribution income, and enter Intel NRE agreements that
preserved ASML's development control and IP while allowing either the 450mm or EUV
project to be abandoned and remaining funding redirected under stated conditions.
ASML later paused 450mm when demand timing was unclear, while EUV progressed to
multi-customer shipments and material annual system volumes. That sequence is favorable
to a workstream-separation thesis, but it does not prove that co-investment caused EUV's
progress or created an attractive return. The durable lesson is to contract frontier
co-investment as separable options and keep equity proceeds, funding commitments,
recognized contribution income, project spending, and commercial concessions in
separate ledgers. [claim.asml.outcome.share-issuance-completed;
claim.asml.outcome.contractual-modularity;
claim.asml.outcome.mixed-workstream-result;
judgment.asml.outcome.workstream-outcome]
Evidence and findings
Approval was prospective; the filed record establishes execution
On July 9, 2012, ASML announced separate next-generation EUV and 450mm development projects within a customer co-investment program. The maximum announced structure contemplated EUR 1.38 billion of customer R&D funding during 2013–2017 alongside a customer equity investment. [claim.asml.cutoff.program-announced; evidence.asml.cutoff.announced-program-terms; @src.asml.2012.ccip-announcement]
Intel simultaneously described agreements totaling EUR 3.3 billion and stated an objective of shortening deployment schedules by as much as two years. That was a participant's forward-looking objective, not a measured acceleration result. [claim.asml.cutoff.intel-acceleration-objective; evidence.asml.cutoff.intel-acceleration-objective; @src.intel.2012.asml-agreements]
After shareholder approval, ASML reported terms under which customers had agreed to contribute EUR 1.38 billion to R&D over five years, ASML would issue an aggregate 23% minority stake for EUR 3.85 billion, and the future share proceeds would be returned to non-participating shareholders through a synthetic buyback. That release used prospective verbs, so the approval tables remain labeled as terms rather than completed amounts. [claim.asml.outcome.final-terms; table.asml.outcome.approved-euro-terms; table.asml.outcome.approved-stake-term; @src.asml.2012.ccip-approval]
ASML's 2013 Form 20-F separately establishes execution. It reports issuing 62,977,877 shares for Intel and 12,595,575 for Samsung on September 12, 2012, then 20,992,625 for TSMC on October 31, and receiving EUR 3,853.9 million in relation to the CCIP shares. It also reports that the synthetic buyback effectively repurchased 93,411,216 shares at an average EUR 39.91 for EUR 3,728.3 million. The filing attributes the EUR 125.6 million difference from share proceeds to repayment on ASML treasury shares. These are executed equity and distribution steps, not R&D funding or program spend. [claim.asml.outcome.share-issuance-completed; claim.asml.outcome.synthetic-buyback-completed; table.asml.outcome.executed-share-counts; table.asml.outcome.executed-equity-cash-steps; @src.asml.2013.20f-html]
Commitment, contribution income, and R&D cost are different measures
The filing reports EUR 64.4 million of 2013 other income from participant NRE contributions. In the same passage it reports EUR 882.0 million of consolidated R&D costs net of credits and excluding those contributions. Contribution income is not a cash-receipt schedule, and consolidated R&D cost is not workstream-level CCIP spend; the two values therefore are presented on separate named accounting bases, not netted into a project return. [claim.asml.outcome.fy2013-contribution-accounting; table.asml.outcome.fy2013-accounting-bases; @src.asml.2013.20f-html]
The filed agreements identify a EUR 553 million Intel 450mm commitment and EUR 276 million EUV commitments for each of Intel, TSMC, and Samsung over 2013–2017. ASML said the three EUV commitments were being used for EUV projects. A commitment and a statement of use still do not establish the dates or amounts of customer cash received, recognized income, or actual workstream spend. [claim.asml.outcome.contractual-modularity; claim.asml.outcome.funding-disposition; table.asml.outcome.nre-commitments-by-workstream; @src.asml.2013.20f-html]
The transaction also carried commercial terms. ASML disclosed specified Intel tool purchase commitments, pricing and delivery milestones, credits for early and volume commitments, additional credits if schedules were missed, and a conditional capacity commitment. The filing establishes that these concessions existed but the cited record does not quantify their net value. [claim.asml.outcome.intel-commercial-concessions; evidence.asml.outcome.intel-commercial-terms; @src.asml.2013.20f-html]
The contracts contained real modularity, and the workstreams diverged
The share governance addressed customer-control risk: program shares were to be held by a foundation, ordinarily non-voting, and subject to a lock-up intended to protect ASML's strategic and operational independence. The later filing adds a stronger fact. Under the Intel NRE agreements, ASML retained sole development control and related IP, and could in reasonable discretion abandon either 450mm or EUV for technical infeasibility, insufficient industry demand, or excess remaining funding versus the project estimate. The parties could redirect remaining funding to an agreed alternative; if they did not agree, ASML could invoice remaining committed funding in years when the specified gross-R&D threshold was met. [claim.asml.cutoff.proposed-governance; claim.asml.outcome.contractual-modularity; evidence.asml.outcome.contractual-modularity; @src.asml.2013.20f-html]
Those were actual contractual abandonment and redirection rights, not merely inferred modularity. They were also not identical to Part A's proposed internal technical, binding-demand, and tranche-release gates: conditional invoicing could preserve funding after abandonment rather than operate as a pure stop-loss. [judgment.asml.outcome.workstream-outcome; claim.asml.outcome.mixed-workstream-result]
In November 2013, ASML paused 450mm lithography development until customer demand and timing became clear, reassigned relevant staff, and said it was discussing with Intel which projects should receive the 450mm funding. A December 2013 industry report independently quoted an ASML spokesperson describing the same demand-contingent pause. The pause is consistent with use of the contracted economic option; it does not measure sunk cost, reuse value, funding redirected, or cost avoided. [claim.asml.outcome.450mm-paused; claim.asml.outcome.funding-disposition; evidence.asml.outcome.450mm-pause; evidence.asml.outcome.450mm-funding-disposition; @src.asml.2013.20f-html; @src.semiconductor-engineering.2013.450mm-pause]
EUV later progressed. For 2017 ASML reported ten EUV system shipments to multiple customers and demonstrations of 125 wafers-per-hour throughput, a full-size defect-free pellicle, and two-nanometer EUV-to-DUV immersion overlay. These issuer-reported milestones do not independently establish high-volume customer economics or isolate the 2012 program's contribution. [claim.asml.outcome.fy2017-euv-progress; evidence.asml.outcome.fy2017-euv; table.asml.outcome.euv-volume; @src.asml.2018.fy2017-results]
Consolidated scale increased, but it is not a project return
ASML reported FY2017 net sales of EUR 9.053 billion and net income of EUR 2.119 billion. For FY2025 it reported total net sales of EUR 32.667 billion and net income of EUR 9.609 billion. These are nominal consolidated endpoints, not co-investment revenue, EUV profit, or a controlled value bridge. [claim.asml.outcome.fy2017-results; claim.asml.outcome.fy2025-results; table.asml.outcome.consolidated-scale; @src.asml.2018.fy2017-results; @src.asml.2026.fy2025-results]
ASML's 2025 annual-report highlights also reported rounded R&D of EUR 4.7 billion and 48 EUV lithography systems among 535 total system sales. The 2017 measure is systems shipped and the 2025 measure systems sold, so the table preserves those labels rather than asserting perfect comparability. Consolidated R&D is not allocated to CCIP or EUV. [claim.asml.outcome.fy2025-innovation-scale; table.asml.outcome.euv-volume; table.asml.outcome.rd-scale; @src.asml.2025.annual-report-web; @src.asml.2025.20f]
Causal assessment
The primary bounded hypothesis is that customer co-investment provided at least some recognized contribution income, coordination, and demand information while contractually separate workstreams and abandonment rights preserved option value. The completed transaction, 2013 accounting, stated EUV funding use, 450mm pause, and later EUV progress are favorable to and consistent with that mechanism. They do not identify its causal magnitude. [hypothesis.asml.customer-risk-sharing-contribution; claim.asml.outcome.attribution-bounded]
The rival hypothesis assigns more weight to ASML's own R&D, suppliers, source and optics progress, customer node demand, product generations, acquisitions, and later execution. It has real support: field tools and eleven next-generation EUV orders existed before the decision, source power remained a production constraint, and 450mm encountered weak demand visibility. [claim.asml.cutoff.euv-learning; claim.asml.cutoff.euv-production-risk; hypothesis.asml.technology-demand-and-execution-dominated]
No reconciled public record here establishes cumulative NRE cash received, spending by workstream, the net value of commercial credits and other concessions, incremental schedule change, project cash flows, or a controlled counterfactual. The case therefore assigns no percentage of EUV deployment, consolidated profit, shareholder return, or enterprise value to co-investment and makes no target-price claim. [judgment.asml.outcome.causal-boundary; assumption.asml.outcome.partial-attribution]
Decision-process learning
Part A recommended staged participation because the cutoff evidence combined meaningful EUV learning and demand with a material production-readiness gap and incomplete transaction economics. The later record validates one structural concern more strongly than a pause inference alone: the Intel contracts actually separated workstreams and included abandonment and redirection rights. The observed 450mm pause is consistent with that structure, but does not prove the option's value, ASML's internal decision process, or use of Part A's exact gates. [judgment.asml.cutoff.stage-co-investment; claim.asml.outcome.contractual-modularity; judgment.asml.outcome.workstream-outcome]
A feasible counterfactual was a narrower self-funded roadmap with incremental releases after independent technical qualification and binding demand. Consolidated cash and R&D guidance made that alternative discussable at the cutoff, but do not quantify project affordability, delay, forgone customer funding, avoided dilution, concessions, or return. [counterfactual.asml.staged-self-funding; assumption.asml.outcome.self-funding-counterfactual]
For future frontier-technology decisions, maintain five enforceable ledgers by workstream: contractual commitments and actual cash receipts; independently qualified technical readiness; binding deployment demand net of cancellation and timing rights; recognized accounting income and project spend; and the full value of governance, pricing, credit, IP, exclusivity, capacity, roadmap, and product-access concessions. Release capital separately, specify abandonment and redirection rules, and require a reconciled bridge whenever definitions or commercial terms change. [rule.stage-customer-co-investment-by-workstream]
Bounded next steps
- Reconstruct annual NRE cash receipts, recognized income, and ASML spend by workstream, then reconcile each to the EUR 1.38 billion headline commitment.
- Value Intel's discounts, schedule credits, purchase commitments, capacity obligation, and any other participant concessions on a common risk-adjusted basis.
- Trace the 450mm abandonment/redirection process, funding invoices, residual assets, staff reassignment, and later reuse without treating sunk work as zero or proven value.
- Build a dated EUV milestone ledger from source power through customer acceptance and production use, preserving issuer claims versus independent tests.
- Test the candidate rule against other customer-funded deep-technology programs and counterexamples before promoting it beyond candidate status.
Further questions
- How much of each NRE commitment became customer cash, recognized contribution income, and incremental workstream spend after all credits and concessions?
- What project received the remaining 450mm commitment, and what amounts were redirected or invoiced under the gross-R&D threshold provision?
- Which independent qualification and customer-acceptance evidence governed EUV releases, and did definitions change as source power and throughput improved?
- Can a controlled schedule reconstruction substantiate or reject Intel's stated as-much-as-two-year acceleration objective?
Caveats
The evidence verifies approved terms; executed customer share issuances and the synthetic
buyback; selected NRE commitments and EUR 64.4 million of 2013 contribution other income;
Intel contractual control, IP, abandonment, redirection, and conditional-invoicing terms;
the 450mm pause and issuer-reported funding disposition; later EUV milestones; and
consolidated endpoints. It does not reconcile cumulative NRE cash receipts, workstream
spend, net concessions, the complete board process, project return, or causal attribution.
This is a retrospective named-case reconstruction: later source artifacts were pre-staged
before the revised Part A freeze, so phase-separated files enforce the cutoff but do not
prove researcher cognitive blinding. Later results also reflect suppliers, customer
demand, acquisitions, product generations, market cycles, and more than a decade of
execution. The success outcome class is an instructor classification, not a causal
finding. [claim.asml.outcome.mixed-workstream-result;
claim.asml.outcome.attribution-bounded]