After Antigua.news reported the Supreme Administrative Court’s ruling, we went back through the record: a 158-page appeal, a 247-page first-instance judgment, a 120-page final decision and a decade-old €2bn bondholder fight. What emerges is not merely a defeat for investors, but a deeper question for Portugal’s courts: how can a case of this scale be disposed of in just 23 days?

From the news to the autopsy

On 10 May, Antigua.news reported that Portugal’s Supreme Administrative Court had rejected a key pilot appeal by Novo Banco/BES bondholders and refused to refer the case to the Court of Justice of the European Union. The ruling confirmed the legality, in Portuguese administrative litigation, of Banco de Portugal’s 29 December 2015 decision to move five series of senior Novo Banco bonds back to the failed Banco Espírito Santo.

The news was important. The judgment is more troubling on a second reading. It does not merely affirm a disputed bank-resolution measure. It reveals how quickly, and how comfortably, a final-instance court can dispose of a vast financial dispute by converting a fresh 2015 administrative act into a continuation of a 2014 emergency decision.

The case concerns roughly EUR2bn of senior bonds. It has generated litigation since 2016. The appeal ran to 158 pages. The first-instance judgment ran to 247 pages. The Supreme Administrative Court’s own judgment runs to about 120 pages. Before even considering the administrative record, exhibits and parallel files, the paper universe was already counted in hundreds of pages. Yet the appeal was admitted on 14 April 2026 and the judgment was issued on 7 May 2026: 23 days later.

Speed is not proof of impropriety. Nor does the text provide evidence that the judgment was generated by artificial intelligence. That allegation would require proof that is not visible from the document. The more serious and more defensible concern is different: in a case of this scale, the judgment reads in significant parts less like a searching appellate review than a compressed institutional defence of the resolution architecture.

That distinction matters. The question is not whether technology wrote the judgment. It is whether judicial reasoning itself became mechanised.

A decade of litigation, 23 days of appellate deliberation

Novo Banco was created in August 2014 as the bridge bank for the viable business of BES. Seventeen months later, Banco de Portugal ordered the retransfer of five series of non-subordinated bonds from Novo Banco to BES. The official retransfer decision described the measure as urgent, waived prior hearing of interested parties, and justified the selected bonds by reference to their EUR100,000 denomination and original placement with qualified investors.

Antigua.news first framed the broader dispute in July 2025 as a story of justice delayed: a decade without a merits judgment in one of Europe’s most contentious post-crisis banking cases. In January 2026, it reported that the Lisbon Administrative Court had upheld the 2015 retransfer in the pilot cases. The Supreme Administrative Court has now affirmed that outcome.

But the appellate timetable is startling. The court records that the appeal was admitted by order of 14 April 2026, that the formal “vistos” were dispensed with, and that a draft judgment was sent to the adjoint judges before the case went to conference. By 7 May, the court had produced a final judgment rejecting the appeal and the request for a preliminary reference.

The procedural detail deserves more than a passing mention. Under Portuguese appellate procedure, dispensing with the ordinary vistos is not meant to be a casual shortcut. Article 657(4) of the Portuguese Code of Civil Procedure allows the rapporteur judge to dispense with them only where the nature of the issues to be decided, or the need for celerity in the appeal, so advises — and with the agreement of the adjunct judges. What the judgment records is that the vistos were dispensed with and that the draft ruling was sent to the adjunct judges in advance. What it does not explain is why the legal conditions for that dispensation were considered satisfied in a case of such scale and complexity.

That omission matters. In a routine appeal, the absence of such an explanation might pass unnoticed. In a pilot case involving a 158-page appeal, a 247-page first-instance judgment, a 120-page final ruling, EU-law questions and almost EUR2bn in disputed senior bonds, it becomes part of the story. The court did not say whether urgency, the nature of the questions, the pilot-case status or some other reason justified bypassing the ordinary vistos period. Nor did it make transparent how that procedural acceleration was reconciled with the need for visible collegial deliberation.

This procedural compression may help explain the speed. It does not remove the concern. A 120-page judgment is not necessarily 120 pages of independent reasoning. In this case, much of the document is descriptive or reproductive: procedural history, parties’ conclusions, counter-arguments, a long factual record and extensive reliance on existing case law. The genuinely contested issues receive answers that are often categorical, but not always sufficiently interrogative.

For a top administrative court deciding a pilot case in litigation said to affect other investors, the optics matter. Justice must not only be done; in a market-sensitive case, judicial scrutiny must be visible. Here, velocity became part of the story.

The decisive move: treating 2015 as 2014

The central legal question is deceptively simple: should the December 2015 retransfer be judged under the Portuguese bank-resolution regime as it stood in August 2014, when BES was resolved, or under the amended framework in force in December 2015, after Law 23-A/2015 had transposed much of the Bank Recovery and Resolution Directive into Portuguese law?

The bondholders argued for the latter. Their case was that the retransfer was an autonomous administrative act. It was adopted 17 months after the original resolution, selected a narrower group of creditors, changed the effective debtor of the bonds from Novo Banco to BES, and produced fresh external legal effects. Under ordinary administrative-law logic, the law in force when an act is adopted should govern that act.

The Supreme Administrative Court rejected that view. It treated the retransfer as part of the same resolution continuum that began on 3 August 2014. The original resolution had stated that Banco de Portugal could “at any time” transfer or retransfer assets and liabilities between BES and Novo Banco. On that basis, the court concluded that the applicable framework remained the 2014 RGICSF.

This is the foundation stone of the judgment. Once the court pulls the 2015 act back into the legal world of 2014, many later arguments become easier to dismiss. The stricter 2015 framework becomes peripheral. The BRRD-transposition rules lose force. The retransfer becomes corrective rather than autonomous. The concentrated loss imposed on five senior bond series is then framed as the completion of an earlier resolution, not as a new interference requiring new justification.

The court’s answer is not absurd. Bank resolution is a process, not a single clerical moment. A bridge-bank structure may require later technical adjustments. But the court’s conclusion was not inevitable. The retransfer did not merely correct a spreadsheet. It moved specific liabilities, at scale, from a going bridge bank to an insolvent estate. It altered market expectations after more than a year of public messaging around Novo Banco’s stabilisation. Those features demanded more than an assertion of continuity.

The harder question is whether a public authority can, by inserting a general reservation in a 2014 emergency measure, preserve for itself a power to reallocate private rights in 2015 under the older legal regime, even after the legislature has enacted a new resolution framework. The judgment answers yes. Investors will ask whether that answer gives administrative power a degree of temporal immunity that the rule of law should resist.

The Luxembourg door stays shut

The most consequential procedural choice was the refusal to refer questions to the Court of Justice of the European Union. Under Article 267 TFEU, national courts may ask the CJEU to interpret EU law; courts of last instance are under a stronger duty to refer where a necessary question of EU law is genuinely unresolved.

The Supreme Administrative Court said no reference was needed. It relied heavily on existing EU case law, particularly BPC Lux 2, concerning the BES resolution, and Novo Banco and Others, the 2024 joined cases arising from Spanish litigation over the recognition of Portuguese reorganisation measures. In the Portuguese court’s reading, those judgments left no material uncertainty: the BRRD did not govern the retransfer, Directive 2001/24/EC was the relevant EU framework, and proportionality was primarily for national courts.

That conclusion is contestable. The CJEU has addressed important aspects of the BES resolution. But the bondholders’ appeal raised more specific questions: the legal character of the 2015 retransfer, the meaning of Article 40 of the BRRD, the use of denomination size as a proxy for investor profile, the treatment of creditors within the same class, indirect nationality discrimination, and free movement of capital.

By treating these issues as already answered or irrelevant, the Supreme Administrative Court kept one of Europe’s most sensitive post-crisis bondholder disputes within the Portuguese judicial system. Private investors cannot themselves force Luxembourg to hear the case. If the national final-instance court refuses to refer, the EU-law route may close before it opens.

In a routine case, that may be tolerable. In a dispute involving cross-border institutional investors, a retroactive-looking allocation of losses, a bank-resolution authority and alleged indirect discrimination, the refusal to refer looks less like economy and more like institutional self-containment.

“Equitable” is not a blank cheque

The court’s second major move was to distinguish equality from equity. On its reasoning, bank-resolution law does not require identical treatment of all creditors in the same insolvency class. It requires equitable treatment. In principle, that is correct. Resolution law is not ordinary insolvency law. Authorities must preserve critical functions, protect depositors, prevent contagion and avoid transferring bank losses to taxpayers.

But the danger lies in turning “equitable” into a blank cheque. Banco de Portugal chose five series of senior bonds. Its official decision relied on the fact that they had EUR100,000 denominations, had been originally issued to qualified investors and were not typically directed at retail investors. It also argued that burden-sharing by retail investors, depositors, trade creditors, derivative counterparties or interbank liabilities could damage Novo Banco or the Portuguese banking system.

Those are policy arguments. A court reviewing a EUR2bn loss-allocation measure should test them with rigour. Why was denomination size a legally sufficient proxy for investor sophistication at the time of retransfer? What if bonds had moved in the secondary market? Why were some creditors of the same legal ranking spared? Were alternative distributions of loss assessed? Was concentration on five series necessary, or merely administratively convenient?

The judgment accepts the central bank’s line that retail confidence and systemic stability justified the boundary. It does not convincingly demonstrate that the boundary was coherent, consistently applied and supported by granular evidence. In resolution, discretion is inevitable. But discretion without searching judicial scrutiny risks becoming ex post allocation of losses to the least politically protected creditors.

Indirect discrimination: the unresolved question

The treatment of indirect discrimination is one of the most vulnerable parts of the reasoning. The bondholders did not argue, or did not need to argue, that Banco de Portugal wrote “foreign investors” into the retransfer decision. Their argument was functional: a facially neutral criterion — high-denomination bonds originally placed with institutional investors — predictably imposed a foreign-heavy burden.

The record contained material pointing in that direction. On the day of the retransfer, Novo Banco informed Banco de Portugal that the relevant issues had been syndicated by international investment banks and that, as far as could be inferred, they had been placed with institutional investors. The investors argued that this made the distributional effect foreseeable.

The court was entitled to demand proof. But indirect discrimination is not visible on the face of a measure. A court cannot dispose of it merely by observing that nationality is not expressly mentioned. The relevant inquiry is whether a neutral criterion placed a protected or cross-border group at a particular disadvantage, and whether that disadvantage was objectively justified and proportionate.

Instead, the judgment appears to set a high evidentiary and conceptual threshold. It reasons that a correlation between the criterion and affected foreign investors is not enough, and that the criterion must be by its nature capable of systematically disadvantaging persons of a particular nationality. That may be doctrinally defensible in abstract EU-law terms. But in this case, the market design of the instruments and the central bank’s own knowledge of international placement made the issue more than a speculative correlation.

The wider sensitivity is obvious. Banking union rests on the promise that cross-border capital will not be disadvantaged by national crisis management. A rule that is neutral in wording but predictably burdens international institutional capital requires careful judicial handling. The judgment gives it a relatively swift doctrinal dismissal.

Proportionality by assertion

The Supreme Administrative Court also upheld the proportionality of the retransfer. It accepted the central-bank narrative that Novo Banco had recorded large impairments and that those losses related to assets inherited from BES. The official retransfer decision had referred to significant overvaluation, impairments of about EUR699mn in 2014, a further EUR270mn in the first half of 2015, expected additional impairments and a retransfer of approximately EUR2bn.

But proportionality is not satisfied by reciting a public interest. A proper analysis asks whether the measure was suitable, necessary and proportionate in the strict sense. In practical terms: did it address the identified balance-sheet problem; was there no less onerous alternative; and did the public benefit justify the private harm?

The judgment engages with these ideas but often at a high level of abstraction. It treats the existence of impairments as justification for retransferring the selected bonds. It does not appear to conduct a rigorous counterfactual analysis of alternative loss-allocation mechanisms, capital measures, asset sales, a different distribution of losses or the timing of the measure. Nor does it fully confront the investors’ argument that the amount retransferred was not transparently matched to losses concretely established at the relevant moment.

Financial stability is a legitimate and powerful objective. No serious observer would deny that. But precisely because “financial stability” is so powerful, courts must prevent it from becoming a master key. The stronger the public-interest claim, the more important the court’s duty to test whether the measure was actually tailored to that claim.

Trust, warnings and the phrase “at any time”

The courts also rejected the bondholders’ legitimate-expectations argument. Their reasoning is straightforward: the 2014 resolution expressly warned that Banco de Portugal could transfer or retransfer assets and liabilities “at any time”. That warning was repeated in later materials. Sophisticated investors, the court held, should have known the risk.

There is force in that answer. Institutional investors cannot plausibly claim to be ordinary consumers. They are expected to read legal documents and price risk. But the counterpoint is more serious than the judgment allows.

The investors were not merely saying that they never read the words “at any time”. They were saying that, for 17 months, the official and commercial environment around Novo Banco communicated stabilisation, continued operation and market normality. Bonds were carried as Novo Banco obligations, traded as such, and in some cases serviced as such. Against that background, a generic reservation of power did not necessarily warn the market that five specific senior series could be pushed back to BES at year-end 2015.

The legal footnote swallowed the market narrative. The court’s answer is essentially that a general “at any time” clause can defeat reliance on subsequent conduct and communications. That may be convenient for resolution authorities. It is unsettling for markets. If bridge-bank liabilities remain indefinitely vulnerable to selective retransfer despite a year of stabilisation messaging, investors will price not only bank risk but also legal-system risk.

A judgment that reads like institutional self-defence

The Supreme Administrative Court’s decision is not irrational. It relies on statutory text, earlier Portuguese case law and CJEU judgments. It is legally constructed. But its tone and structure create the impression of a court more concerned with preserving the resolution architecture than testing the boundary of administrative power.

Several features stand out. The court gives substantial deference to Banco de Portugal’s technical assessment. It treats the retransfer as corrective rather than as a fresh legal burden. It relies on the 2014 legal framework despite the 2015 date of the act. It frames equitable treatment broadly enough to allow significant differentiation among senior creditors. It rejects indirect discrimination without a full distributional inquiry. It refuses a CJEU reference in a case where EU-law uncertainty was at least arguable. It does all this in an unusually compressed timeframe.

None of this proves AI generation, nor should that allegation be made without evidence. The stronger point is institutional. The judgment appears overly derivative, overly deferential and insufficiently searching for a case of this magnitude.

A final-instance court may agree with the state. But where public power reallocates almost EUR2bn of private loss, the court’s reasoning must do more than confirm that the authority had a plausible reason. It must show, visibly and granularly, that the reason was tested.

The pilot-case problem

This was a pilot case. That matters. Pilot litigation is designed to produce efficiency in mass disputes. But efficiency becomes dangerous when the pilot judgment is thin on the issues that make the wider litigation significant.

As Antigua.news noted in its report on the Supreme Administrative Court ruling, the judgment is not formally binding on every pending Novo Banco/BES case. But pilot cases exert a gravitational pull. Once the highest administrative court has endorsed the state’s theory in a flagship appeal, lower courts and future panels may have little appetite to reopen legal framework, proportionality, discrimination or EU-referral questions.

For claimants in parallel cases, the risk is practical rather than formal. A judgment that can be used as an off-the-shelf dismissal template can narrow the path before their own facts have been fully heard. That is why the quality of reasoning in a pilot judgment must be visibly higher, not merely efficient.

The wider message to capital markets

The Novo Banco dispute is no longer only about five bond series. It is about how investors assess legal risk in European bank resolutions.

If senior bonds can be transferred to a bridge bank, trade for more than a year on the assumption that they sit in that bank, and then be selectively pushed back to a failed estate, investors will ask what “senior” means in practice. If national courts decline to ask Luxembourg for clarification, investors will ask whether EU banking law is as uniform as advertised. If domestic judicial review of central-bank action is deferential rather than intense, investors will ask whether rule-of-law risk is priced correctly in sovereign-linked financial markets.

Portugal had legitimate reasons to protect Novo Banco, depositors and financial stability. But the rule-of-law question is different: whether those reasons were tested with the depth required when the state reallocates almost EUR2bn of private loss.

The Supreme Administrative Court has given its answer. For international investors, the troubling point is how quickly, and how comfortably, it arrived there.