EDIP and the 35% rule

EDIP is not another strategy document. It is a regulation with a budget line and application deadlines. €1.5bn for 2025–2027, split across industrial reinforcement, common procurement (€240m, funding up to 25% of contract value and capped at €20m per project), European Defence Projects of Common Interest, and a €300m Ukraine Support Instrument.

The important shift for a supplier is not the money. It is that eligibility now depends on facts about your supply chain that no one previously had to state under audit: where each component originates, what it cost, and whether anyone outside the Union can stop you changing it. EDIP turns provenance from a commercial detail into an access condition.

Article 10(3) provides that the cost of components originating outside the Union and associated countries shall not exceed 35% of the estimated cost of the components of the end product. Three details are routinely misread:

• 01. The denominator is component cost, not contract value. Labour, integration and engineering do not dilute the ratio.
• 02. Raw materials are excluded from “components” — relevant if you import titanium or rare earths.
• 03. It is a ceiling, not a target, and no component may come from a country acting against EU security interests regardless of share.

Article 11(7) adds a transitional allowance: a common procurement action may include one extra-EU subcontractor holding 15–35% of contract value, but only where a direct contractual link predates the Regulation's entry into force. New non-EU subcontracting does not get that relief.

Worked example — an €8.8m component basket:

Third-country total: 26.1%. Compliant — until engineering releases BOM revision H and moves the RF front-end to a second source at a higher price, or the dollar moves 8%. The ratio is not a document you file once; it is a figure that changes underneath you, and the burden is to know when it crosses the line before a reviewer does.

Article 10(5) requires recipients to be able to decide on the definition, adaptation and evolution of the design of the product, including the legal authority to substitute or remove components. This is a sovereignty test dressed as a procurement condition. The question is not whether your contract mentions design rights — it is whether a third-country licensor holds an effective veto over a change you might need to make in year seven.

In practice that means reading three artefacts together: manufacturing licence agreements that restrict modification, technical data agreements under ITAR or EAR (avionics and missile subsystems, mostly), and supply agreements carrying denial-of-supply clauses. Article 9(6) closes the loop on the output side: IP generated under EDIP must not be subject to third-country restrictions. The European Military Sales Catalogue is intended to signal whether products are free of such restrictions.

One narrow relief exists: under Article 12(4), ammunition and missiles may satisfy design authority through a legally binding commitment to reach full autonomy by 31 December 2033. Obtaining that commitment from a foreign government is not a paperwork exercise, which is why restructuring conversations are happening now rather than in 2032.

Your counsel can tell you what Article 10(3) means in an afternoon. What takes six weeks is answering it. The inputs are already in your estate and already disagree with each other: the part master in ERP holds a vendor and a price; PLM holds the BOM revision that is actually being built; purchasing holds the PO that supersedes the contract price; the origin declaration itself is a PDF in a shared drive, signed by someone who left in March.

The failure mode is rarely non-compliance. It is non-reproducibility — you had a defensible number in June and cannot rebuild it in October because three of the inputs moved and none of the changes were versioned. An auditable answer is not a percentage. It is a percentage plus the inputs, the timestamp, the attestation, and the diff against last time.

This is also why “ask suppliers to fill in a portal” keeps failing. Tier-two response rates are the constraint, and any approach that requires onboarding your supply chain before you get a first answer will not produce a number by 16 June. The workable order is inverted: compute from the data you already own, mark what is evidenced and what is assumed, then chase only the gaps that actually move the ratio.

• — Pick one end product likely to sit in an EDIP application. One, not the portfolio.
• — Freeze a BOM revision and pull component cost from the PO history, not the price list.
• — Classify each line: EU/associated, third country, raw material (excluded), unknown.
• — Compute the ratio with the unknowns counted as third country. That is your worst case.
• — Rank the unknowns by cost, not by count. Usually eight parts decide the answer.
• — For each third-country line above 5%, record whether a licensor can veto substitution.
• — Store the result as a versioned record: value, inputs, date, who attested, what changed.
• — Re-run it monthly. If the re-run is manual, it will not happen twice.

None of this requires replacing a system of record. It requires a layer above them that reads what is there, computes the figure, and keeps the audit trail — which is precisely what we built DefenceOS to do.