A Few Dozen Firms Hold the Key to Europe's Faster Arms Transfers

Analysis as of 6 August 2026.

Buried in the June 2026 compromise on the EU's defence omnibus is a bet on speed that depends on a remarkably small population of companies. The negotiators' text, agreed provisionally on 10 June and cleared in committee on 24 June, obliges every Member State to publish two new general transfer licences: one covering shipments where both sender and receiver hold certification under the transfers directive, and one covering movements inside a formally structured cross-border industrial partnership — a lasting arrangement through a shared group, consortium or joint venture, now extended to joint development and through-life support, software and technology included.

The prize is time. The Commission's own November 2025 staff paper puts a typical general-licence transfer at one to three days, against an average of six to seven weeks for a simple individual authorisation. The gap between those two clocks is the entire commercial substance of the reform. Yet the gate to the fast lane is certification, and the recorded uptake of that status is thin to the point of paradox. The Commission counted fifty-five certified undertakings across the whole Union in its 2016 evaluation — fourteen German, twelve French, with half the Member States hosting none at all. Parliament's 2021 report, drawing on the public register, gave sixty-six, describing this as a rise of merely twenty-seven in five years; the arithmetic of those two official figures cannot be reconciled, and no later count has been published anywhere. Meanwhile roughly 89 per cent of transfers were still running on individual licences, the very instrument the 2009 directive set out to make exceptional, and a recent survey found only about a third of responding firms had ever used a general licence at all.

What the compromise does not do matters as much as what it does. National authority over the downstream life of a transferred product is left standing at every point: originating states keep their leverage over re-export consent, end-use conditions and customs action, retain a thirty-working-day power to block onward export of incorporated products, and may still suspend their general licences toward a certified recipient elsewhere. The reform's method is therefore not federalisation but sorting — moving the decisive discretion upstream, from approving each transaction to deciding which firms qualify for the faster category in the first place. Certification itself thickens accordingly: it becomes bilateral, covering suppliers as well as recipients, with a compliance apparatus specified down to audit trails, staff training and transaction screening, and a state review at least every three years.

Nor is any of it law yet. At the mid-July cut-off the file still awaited Parliament's first reading, a legal-basis opinion had been opened on 1 July, and the statute in force remained the 2009 directive as previously amended, without the omnibus changes. Even after adoption, states have a year to transpose and fifteen months before application. The political stakes are set by the surrounding programme: 2024 defence spending of €326 billion, an investment ambition of at least €800 billion over four years, and a readiness horizon of 2030 that the White Paper tied explicitly to faster cross-certification.

Three tests will decide whether the two new licences reorganise intra-EU supply chains or merely re-label a bottleneck, and none of them can be answered from the text alone:

— Will national authorities certify component makers and mid-caps at a pace that widens the fast lane beyond the low-tens population — historically skewed to integrators in two countries — that the old regime produced?

— Will the public register finally publish a count, a time series and a breakdown by industrial function, so that anyone outside the ministries can even measure who has qualified, and where?

— And when the first certified-to-certified shipment meets an originating state's residual powers — a withheld re-export consent, a suspended licence, a safeguard invoked — will the one-to-three-day promise hold, or will the six-week world reassert itself one derogation at a time?