The State on the receiving end — the exposure, the twenty-year tail, section 25A's double edge, and where a creditor would look
Research note · 14 October 2026 · Verify pinpoints before filing
One registered ICSID claim, three undisclosed notices — and a State that had never been sued under an investment treaty before
Until 2026 Ireland had never faced an investment-treaty claim. It now has one registered ECT arbitration under the ICSID Convention and, on its own Freedom of Information disclosure, three further Article 26 notices of dispute that have not reached registration. The particulars are recorded on the ICSID and the ECT note. This note is about what they mean — and why Ireland's position is not Spain's.
Lansdowne Oil & Gas PLC and Lansdowne Celtic Sea Ltd v Ireland, registered by ICSID on 8 May 2026; status pending, no tribunal yet constituted. Both claimants are British. The claim arises from the Barryroe field in the Celtic Sea: in May 2023 the Department refused the lease undertaking needed to develop it, following the 2021 ban on new exploration. Lansdowne's own disclosures put the claim at a minimum of US$100m plus interest, funded on a non-recourse basis. Full entry · ICSID record
In response to an FOI request, Ireland disclosed three further ECT Art 26 notices of dispute, none previously public (reported by IAReporter, 1 May 2026). A notice starts the three-month amicable-settlement period; it is not a registered case, so none appears in the ICSID, ECT Secretariat or UNCTAD databases. Claimants, sectors and amounts are not established from any primary source and are deliberately left blank. Placeholder entry
Article 47(3) ECT — why withdrawal does not switch anything off
Ireland notified its withdrawal from the Energy Charter Treaty on 27 April 2026, taking effect on 28 April 2027. The instinctive reading is that this ends the exposure. It does the opposite: it fixes the date from which a twenty-year tail begins to run.
Art 47(3) provides that the Treaty continues to apply to investments made in a withdrawing Party's territory by investors of other Contracting Parties — and vice versa — as at the date withdrawal takes effect, for twenty years from that date. For Ireland that is to April 2047. Any investment existing on 28 April 2027 keeps its ECT protection, and its access to ECT arbitration, for the whole of that period. Lansdowne's Barryroe interest long predates it. Withdrawal changes nothing for the claim already registered and nothing for any investment already in the ground; it only closes the door to investments made after 2027.
The UK withdrew from the ECT with effect from 27 April 2025. The same Art 47(3) preserves the protection of British investors' existing investments in other Contracting Parties for twenty years from that date — to April 2045. So a British investor in an Irish energy asset held on 27 April 2025 keeps an ECT claim against Ireland until 2045 notwithstanding that both States have left the Treaty. That is exactly what Lansdowne is. Neither withdrawal is an obstacle; each merely starts a different clock.
The EU and its Member States have pursued an inter se instrument declaring that the ECT — the sunset clause included — does not apply between EU States, to shut down the intra-EU tail. Three limits need stating plainly. It binds only its parties: it cannot reach a British, Swiss, Norwegian or Japanese investor, whose home State never disapplied anything, nor an ICSID tribunal's jurisdiction under a treaty that investor's State still relies on. It cannot bind a court in Washington, London, Canberra or Singapore asked to recognise an award — the point the third-country forums have now made against Spain in every major jurisdiction. And at most it strengthens the intra-EU objection intra-EU, which for Ireland's registered claim is beside the point.
The sunset clause is usually discussed as a problem for departing States in the abstract. For Ireland it is concrete, dated, and already engaged.
The Costello fix — a shield that faces the wrong way
In Costello v Government of Ireland [2022] IESC 44 the Supreme Court held that CETA's investment-court awards could not be made automatically enforceable without a statutory safeguard for the constitutional order. The Arbitration (Amendment) Act 2026 (No. 13 of 2026, enacted 17 June 2026) supplied it, inserting section 25A into the Arbitration Act 2010. Section 25A(3) declares that an award made under an agreement to which the section applies is “not, and never was” enforceable where enforcing it would compromise the constitutional order of the State or the autonomy of the legal order of the European Union; s 25A(2) makes enforcement of such awards a matter of High Court leave, subject to that gateway, and exclusive — such an award “shall not be enforced in the State other than under this section”. Its scope is the critical point, and it is settled on the face of the Act. By s 25A(1) the section applies to CETA, the EU–Chile Advanced Framework Agreement, and any international agreement the Minister for Foreign Affairs and Trade prescribes by order under s 25A(5) — an agreement to which the State is or intends to become party, wholly or partly concerned with the protection of investment, and providing for awards under its investment dispute provisions. The Energy Charter Treaty is not listed. Nor is the ICSID Convention. Four consequences follow, and two of them are uncomfortable. (See the Costello entry for the judgment.)
A seven-judge Court held, 4–3, that the Constitution precluded ratification of CETA as Irish law then stood, and 6–1 that amendments to the Arbitration Act 2010 would cure it. The objection was not that a treaty tribunal would sit outside the Irish courts; it was what happens at enforcement. An award enforceable under the 2010 Act is enforced “almost automatically” — refusable only on narrow public-policy or Greendale-type grounds, never on the merits — so an Irish court could be compelled to enforce an award that ran directly contrary to a finding of the Supreme Court itself. Baker J's hypothetical (at [742]) was an investor who loses in the Irish courts, as in Cromane Seafoods, and then wins the same damages before a treaty tribunal. That collision with the finality of Irish judgments (Art 34) and the democratic nature of the State (Art 5) is what s 25A exists to cure.
Three things follow for this note. First, Costello is the Irish Supreme Court's own statement of the Art 54 point: enforcement of a treaty award under the 2010 Act is near-automatic and unreviewable — which is exactly why a gateway applied to an ICSID award (point 2 below) would be a breach rather than a refinement. Second, the logic is CETA-specific only in subject, not in reasoning. The objection to near-automatic enforcement without merits review applies to any treaty award, and the ICSID Convention is the purest instance of it. The Court did not have to confront that, because the Convention was not before it and Ireland had ratified it forty years earlier. But the reasoning leaves a question hanging over s 25 itself — and unlike CETA, Ireland cannot answer it by declining to ratify. It is already bound. Third, Baker J's hypothetical has the same shape as Lansdowne: a treaty tribunal asked to award damages for an Irish regulatory decision — the refusal of the Barryroe lease undertaking — which an Irish court might or might not have condemned. Costello was written about a threat that has not materialised; its reasoning describes the one that has.
On Europcar factor 3, a foreign proceeding conducted under a less deferential standard of review is a reason for a US court to stay enforcement and defer to that forum. Section 25A is, on any view, a less deferential standard than the near-automatic enforcement the New York or ICSID Conventions contemplate. A vigorous s 25A challenge in Dublin therefore does not make US enforcement harder for the State; it gives the US court a factor-3 reason to adjourn. The full analysis is in the Europcar entry.
Article 54(1) of the ICSID Convention obliges each Contracting State to recognise an award as binding and to enforce its pecuniary obligations as if it were a final judgment of a court in that State. There is no gateway, no public-policy exception, no Art V menu. A domestic statute cannot excuse non-performance of that obligation: Vienna Convention on the Law of Treaties, Art 27. If s 25A were ever applied to refuse an ICSID award, Ireland would be in exactly the position Luxembourg put itself in with Micula — a refusal that is coherent in EU law and a breach in international law (Cour de cassation, 14 July 2022). The statute does not resolve the conflict between the two legal orders; it writes it into Irish law.
Micula v Romania [2020] UKSC 5 turned on Art 351 TFEU: the UK's ICSID obligations were undertaken in 1966, before its 1973 accession, so they were prior international obligations that EU law could not displace. Ireland's chronology runs the other way. It joined the EEC on 1 January 1973; it signed the ICSID Convention in 1966 but ratified it only on 7 April 1981 (in force 7 May 1981) — after accession (ICSID member-state database; verify the dates against the ICSID/3 list). Whether signature or ratification is the moment an agreement is “concluded” for Art 351 is arguable; what is not arguable is that Ireland's position is materially weaker than the one that carried Micula in London. The consequence is stark: EU law may require Ireland to refuse an intra-EU ICSID award (sincere cooperation, Komstroy) while Art 54 requires it to enforce — the Luxembourg dilemma, with no Art 351 exit. Section 25A is the statutory expression of that dilemma, not a solution to it.
Lansdowne is an extra-EU, ICSID claim under the ECT. The autonomy of the EU legal order is not engaged by a British investor's award, and the ECT is not among the agreements to which s 25A applies. So, as the Act stands, an award would be enforced under s 25 of the 2010 Act in the ordinary way — with the near-automatic effect Costello described. That makes Dublin itself the creditor's easiest forum, a possibility the rest of this site never has to consider for Spain: no sovereign immunity in the State's own courts, no Achmea objection, no obvious EU-law reason not to pay.
But s 25A(5) is a switch, and the Minister holds it. The ECT satisfies every condition for prescription: Ireland is a party (and remains bound under the sunset), the Treaty is concerned with the protection of investment, and Art 26 provides for awards. A single order — laid before the Oireachtas and annullable only by resolution of either House within 21 sitting days (s 25A(6)) — would bring ECT awards within the section, and s 25A(2)(ii) would then route them exclusively through the leave-and-gateway procedure, displacing s 25's automatic effect and arming the “not, and never was” clause against them. Applied to an ICSID award, that is the Art 54 breach described at point 2. The position is therefore not a fork but a timing question: today a Lansdowne award walks through s 25; the day an order is made, it meets the gateway — and Ireland meets the Convention.
Two further points on the face of the Act. It comes into operation only on a day appointed by Ministerial order (s 5(2)), so confirm commencement before treating s 25A as in force. And s 25A(4) excludes any appeal from the High Court's determination other than directly to the Supreme Court under Art 34.5.4° — a leapfrog that keeps the gateway in the hands of the court that wrote Costello.
Ireland abroad — the map a Lansdowne-type creditor would draw
Ireland is a member of Eurocontrol. The en-route charges for Irish-controlled airspace — heavily used North Atlantic transit — are collected by Eurocontrol's Central Route Charges Office in Brussels and remitted to Ireland's air-navigation service provider, AirNav Ireland. That is precisely the mechanism Spain's creditors have turned on ENAIRE, with a reported ~€482m frozen (execution note). The three hurdles identified there apply — with one decisive difference. The first hurdle, that the Belgian court must eventually confront the Achmea objection when validating the award, falls away for a British investor's award: there is no EU-law objection to a Belgian court giving effect to an extra-EU ICSID award. What remains is Eurocontrol's own privileges and immunities as an international organisation, and the Stati-type “attachable obligation” question — are the charges owed to the State, or to AirNav Ireland as a separate entity? Those are real hurdles; they are not the EU-law wall. Against Ireland the route is cleaner than it has ever been against Spain.
The Ireland Strategic Investment Fund, managed by the National Treasury Management Agency, holds commercial investments, some in the United States. Under 28 U.S.C. § 1610(a) the question is whether that property is “used for a commercial activity in the United States”; on the Weltover nature-not-purpose test, fund investments look commercial. The State's answer runs the other way: ISIF is a sovereign reserve deployed under statutory ministerial direction (NTMA (Amendment) Act 2014), closer in character to a central-bank or reserve-management holding — and § 1611(b)(1) shields central-bank property held for its own account. Both arguments are respectable. The point for Ireland is that the asset most visible to a US creditor is also the one whose characterisation is least settled.
ESB, EirGrid, Coillte, Bord na Móna, the port companies, AirNav Ireland: all separate juridical entities. The Bancec presumption in the United States, s 14 of the State Immunity Act 1978 in England, and the same alter-ego analysis everywhere protect their assets from a claim against the State unless control is shown. The wall that has stopped Spain's creditors at SEPI and the Instituto Cervantes would stop Ireland's at ESB. Which is why the Eurocontrol route matters: attaching a debt owed to such an entity, in a third party's hands, is the one way the asset hunt has found round the wall.
Diplomatic and consular premises and accounts (VCDR Art 22; VCCR); Central Bank of Ireland reserves (§ 1611(b)(1); s 14(4) SIA); military property (§ 1611(b)(2)). The same perimeter as for every State.
The map is Spain's map, with one difference that changes everything: an extra-EU creditor meets none of the EU-law walls. Ireland's exposure is to the ordinary law of sovereign immunity, undiluted.
Six points for anyone advising on Irish ECT exposure
The practical conclusion is the inverse of the one the rest of this site reaches for Spain's creditors. There, recognition is easy and payment nearly impossible. For Ireland facing a third-State claimant, recognition would be easy and the usual reasons payment fails are, one by one, absent.
Primary materials cited on this page, and the entries on the companion notes this analysis builds on
Treaties and legislation
Cases and records
On the companion notes