A. Introduction
Is the Apple of EU State aid law rotten, or merely bad? The question is more than rhetorical. In September 2024, the Court of Justice of the European Union (CJEU) delivered the European Commission (EC) its most significant State aid victory to date, upholding a recovery order requiring Apple to repay up to €13 billion in unpaid taxes to Ireland.Footnote 1 The judgment endorsed the Commission’s use of State aid control as a means of scrutinizing aggressive corporate tax planning through competition law.Footnote 2 Yet Ireland had fought for nearly a decade to refuse the money. Apple maintained it owed nothing. And despite the landmark win, no wave of enforcement has followed.
This disconnect—between judicially affirmed potential and persistent practical constraints—constitutes the “stubborn gap” at the heart of this Article. By the time of the Apple judgment, the Commission’s enforcement momentum had already waned, with no new headline investigations into advance pricing arrangements (APAs) launched after 2022.Footnote 3 The parallel cases—Fiat, Starbucks, Amazon, Engie—collapsed before the EU courts.Footnote 4 By the time the CJEU ruled, the Commission had largely moved on.
This Article argues that Apple was a warranted intervention that nonetheless exposed the limits of using competition law to reshape corporate tax governance in the EU. The Commission’s post-2013 campaign functioned as an experiment in institutional construction, seeking to connect firm-specific tax rulings to systemic distortions in competition across the internal market. Apple marked the apex of that experiment—it combined the most extensive evidentiary record with the largest recovery order and ultimately yielded the Commission’s only unqualified judicial victory. The case shows that, under exceptional conditions, State aid control can prioritize economic substance over formal legal structures. Yet the very specificity of those conditions helps explain why the approach has proven difficult to replicate.
The “stubborn gap” reflects structural constraints affecting both the enforcement and legislative tracks of the Commission’s campaign. From the outset, the Commission pursued parallel strategies: litigation against advance tax rulings and legislative harmonization through the Anti-Tax Avoidance Directives (ATAD),Footnote 5 Directives on Administrative Cooperation (DACs),Footnote 6 the Minimum Tax Directive, implementing Pillar Two,Footnote 7 the proposed Unshell DirectiveFootnote 8 and the Transfer Pricing Directive (TPD).Footnote 9 Yet both have encountered resistance. By 2025, the Unshell Directive had been abandoned, the TPD listed for withdrawal, and Pillar Two was reconfigured to exempt US multinationals from its core provisions. The political-economic forces that constrained enforcement—sovereignty, unanimity, geopolitical asymmetry—have proven equally capable of constraining legislation. These limits extend beyond State aid to reflect deeper structural constraints on the EU’s capacity to reshape corporate tax governance through competition law.
A Transformative Law and Political Economy (TLPE) lens helps explain this outcome. TLPE foregrounds the gap between legal possibility and institutional realization—the disconnect exemplified by the Apple saga—and directs attention to the form-giving role of law in structuring economic power. It clarifies why judicial victories alone rarely yield systemic transformation, and why both enforcement and legislative strategies encounter resistance from actors embedded in existing institutional and political-economic arrangements. In this Article, TLPE functions as an analytical framework for understanding how legal institutions both enable and constrain attempts at transformative change.
This Article makes three contributions. First, it offers a TLPE reading of the Apple litigation, treating the case as a test of institutional reimagination and of the limits of repurposing competition law to tackle corporate tax abuse. Second, it provides a doctrinal reconstruction of the Apple saga, engaging with critiques of the CJEU’s reasoning concerning the use of the Authorized OECD Approach (AOA), intra-entity profit attribution, and the relevance of a potential US permanent establishment (PE). Third, it situates the Apple case within the broader trajectory of EU tax governance by analyzing the parallel legislative track and the structural constraints it has faced. This Article concludes by showing how a legal framework attentive to political economy offers a more robust account of the EU’s capacity and limits in governing corporate tax avoidance.
B. The TLPE Framework: Law, Institutions, and Transformations
I. What TLPE Is
Transformative Law and Political Economy (TLPE) conceptualizes law as an instrument of institutional design. As Poul Kjaer argues, TLPE foregrounds law’s form-giving function: its capacity to designate, differentiate, interconnect, and constitute social processes as legally structured institutions.Footnote 10 This form-giving role extends to foundational elements of the economy, including competition, contract, and the corporation. Legal rules define what counts as a firm, how economic relationships are organized, and which forms of coordination and exchange are recognized or excluded. In this sense, law occupies a strategically central, infrastructural position in structuring the political economy and shaping the distribution of economic power and advantage.Footnote 11
TLPE understands legal structuring as the medium through which society sustains stability and enables transformation. Building on this insight and drawing on Roberto Unger, TLPE emphasizes that legal institutions, though often perceived as fixed or given, are contingent and contestable. They serve as foundations of the status quo and as potential sites for radical change.Footnote 12 In this Ungerian perspective, legal analysis provides the essential institutional imagination needed to envision and construct social alternatives.Footnote 13 TLPE thus attends to law’s (trans)formative power: its capacity to constitute social realities, embed normative values, and reshape power relations—while recognizing that such transformations confront entrenched resistance.Footnote 14
Three premises underpin the TLPE approach. First, contingency: arrangements that appear natural or necessary—for example, the corporation, tax residence, or the arm’s-length principle—are historical constructs that have emerged through political contestation and have become embedded as taken-for-granted background structures. Second, power: law does not operate on a level playing field, and legal change confronts actors with unequal resources and divergent interests in preserving or transforming existing arrangements. Third, form-giving: legal rules and institutions define what counts as a firm, a transaction, or a taxable presence, thereby shaping the terrain on which economic activity and distribution take place.
II. What TLPE Adds: Beyond Doctrinal Analysis
A purely doctrinal account of Apple can tell us what the CJEU decided.Footnote 15 It can identify the relevant reference framework, assess the application of the arm’s-length principle, and evaluate whether the General Court or the Court of Justice offered stronger legal reasoning. What it cannot do is account for the “stubborn gap” examined in this Article: the divergence between legal possibilities for transformation and the structural constraints that limit their institutional realization, in both enforcement and legislation.
TLPE provides that lens. It directs attention to the political-economic conditions that made Apple possible, including the doctrinal foothold in Section 25 Taxes Consolidation Act (TCA) 1997,Footnote 16 the Commission’s access to detailed evidentiary material, the post-financial-crisis political climate, and the cumulative pressure generated by investigative journalism. It also helps explain why the broader campaign has stalled, pointing to reference-framework disputes in Luxembourg, systematic evidentiary resistance by Member States, diplomatic pressure from the United States, and the unanimity requirement that constrains legislative alternatives. Whereas doctrinal analysis identifies the applicable legal rules,Footnote 17 TLPE explains how those rules operate within institutional and power structures, and why the EU’s attempt to reshape corporate tax governance through both litigation and legislation has met persistent resistance.
TLPE also illuminates contingency. Critics of the Apple judgment raised a range of objections: that the Authorized OECD Approach was improperly imported into Irish law, that the intra-entity focus ignored group–level functions, or that the Commission should instead have identified a US PE. Yet these critiques shared a common premise—that the legal structures sheltering Apple’s profits were fixed and legally compelled. The CJEU’s judgment unsettled that premise by treating those structures as contestable rather than necessary. In TLPE terms, Apple revealed the formal-allocation paradigm as contestable rather than necessary within the reference framework that had become procedurally final. Rather than independently resolving the content of Irish law, the judgment exposed the contingency of economic arrangements that had been presented as legally inevitable.
III. The Meso-Level: Bridging Firm-Level Engineering and Systemic Market Power
A macro-level perspective is essential for analyzing corporate taxation. The fragmentation of national tax regimes across the internal market—through divergent residence rules, transfer pricing standards, and treaty networks—creates systemic conditions that firm-level tax engineering is able to exploit. Structures such as the “Double Irish” are micro-level arrangements, but their cumulative effects are macro, shaping where profits are recorded, which jurisdictions can tax, and how corporate wealth is distributed. Yet legal enforcement typically operates at the micro level, focusing on individual firms, specific tax rulings, and discrete recovery orders. TLPE addresses this disconnect through its meso-level focus on institutions that link firm-specific arrangements to broader patterns of market power and distribution.
From this perspective, the Commission’s post-2013 State aid campaign in the tax field can be understood as an experiment in meso-level institutional construction in the TLPE sense. The ambition was not merely to win individual cases, but to reconfigure how legal authority connects firm-level tax arrangements to the organization of competition in the internal market. By redeploying Article 107 TFEU as a bridge between bespoke tax rulings and systemic market effects, the Commission sought to use State aid law as a form-giving device, capable of reshaping the legal and economic conditions under which multinational firms allocate profits across jurisdictions.
Through a coordinated series of investigations, including those into Apple, Fiat, Starbucks, and Amazon, the Commission attempted to demonstrate that advance pricing arrangements negotiated between individual firms and national tax authorities were not merely technical tax determinations, but institutional practices that structured competition across the Single Market by enabling large-scale profit shifting.Footnote 18 If this recharacterization were sustained, State aid control would no longer operate only as a constraint on subsidies, but as a meso-level governance mechanism linking micro-level tax engineering to macro-level distortions in market integration.
This was institutional reimagination in the TLPE sense—the strategic repurposing of an existing legal instrument to perform a new form-giving function within the political economy.Footnote 19 Apple marked the high point of this effort, combining the largest recovery order with the most extensive evidentiary record and producing the Commission’s only clear judicial success. It therefore became the pivotal test of whether this meso-level bridge between tax law and market governance could hold.
IV. Apple as a TLPE Test Case
Why does Apple merit analysis through a TLPE lens? Three considerations are central. First, Apple was an episode of institutional reimagination. The Commission redeployed State aid control, until then a subsidy-policing instrument, into a governance device for disciplining multinational profit allocation, and in the process recast the purpose of the doctrine itself. TLPE captures this kind of institutional redesign because it treats legal instruments as form-giving infrastructures of the political economy, not simply as rule-sets to be correctly or incorrectly applied.
Second, Apple exposed the contingency of legal structures that had insulated aggressive tax planning. Advance pricing agreements and stateless head offices rested on formalistic interpretations of tax residence and profit attribution that appeared legally necessary within conventional tax doctrine. The CJEU’s judgment unsettled those assumptions by allowing a substance-oriented functional allocation to displace the rulings’ formal profit allocation. TLPE explains this not as a technical correction, but as the destabilization of a historically sedimented institutional arrangement—what Unger terms “false necessity”Footnote 20 —that had organized power and value allocation in the Single Market.
Third, and most importantly, Apple reveals the limits of institutional construction in a way that doctrinal analysis cannot. Doctrinally, Apple was a clear victory. Institutionally, it failed to generalize—no wave of enforcement followed, and parallel cases in Luxembourg, the Netherlands, and the United Kingdom collapsed. TLPE explains this divergence by foregrounding the political-economic constraints that operate alongside legal doctrine—fiscal sovereignty, unanimity requirements, information asymmetries, geopolitical pressure, and regulatory fatigue. A doctrinal account can tell us that Apple was won, while TLPE explains why that victory is unlikely to become a durable mode of governance.
C. The Apple Proceedings as a TLPE Test-Case: From Micro Tax Engineering to Meso-Level Legal Contention
This section traces the Apple litigation from its factual matrix through the judicial dialogue between the General Court and the Court of Justice in order to assess its significance as an episode of institutional reconfiguration in the TLPE sense. It asks whether the final judgment shows that State aid control can function as a meso-level legal device for translating firm-specific tax rulings into a mode of governance over corporate tax abuse within the internal market. The conditions that enabled this translation, and the structural constraints that prevented it from becoming durable, are examined in Sections D and E.
I. Factual Matrix: Micro Frictions Inside a Fragmented Legal Order
At the micro level, Apple Inc. deployed a bespoke version of the “Double Irish” structure. The classic arrangement used two Irish-incorporated companies, one resident and one not, with the resident company paying deductible royalties to the non-resident holder of the IP so that the income was taxed in neither place. Apple’s variant dispensed with the royalty leg altogether. It incorporated two Irish companies—Apple Sales International (ASI) and Apple Operations Europe (AOE)—and treated the head offices of both as being managed and controlled outside Ireland. Because Irish tax residence then followed management and control, both entities were classed as non-resident, and, having no tax residence anywhere else, their head offices were effectively stateless.Footnote 21 This structure was particularly effective because, under US tax deferral rules then in effect, profits earned by offshore companies were not subject to US tax until they were repatriated, allowing them to accumulate effectively tax-free.Footnote 22
Importantly, this legal structure was applied to companies with significant real-world operations. The Irish branch of AOE manufactured computer products for the entire EMEIA (Europe, the Middle East, India, and Africa) region from its factory in Cork,Footnote 23 while ASI’s Irish branch procured finished Apple devices from Asian contract manufacturers and sold them throughout EMEIA.Footnote 24 Yet under Section 25 of the TCA, companies, even if not tax residents, were subject to Irish tax on any profits attributable to the activities of their Irish branches. The crucial question was therefore not if tax was due, but how to allocate the profits from these real–world operations—particularly as Section 25 provided no specific formula for splitting profits between an Irish branch and its non-resident head office.
Apple used this statutory indeterminacy to frame the profit allocation exercise narrowly. Rather than inviting a holistic inquiry into where profits were generated, Apple presented the attribution of the royalty-free licence rights to the stateless head offices as a given and focused the negotiations with Irish Revenue on determining a routine return for the Irish branches.Footnote 25 This approach confined the tax base to a small cost-based markup, with the residual profits attributed to the head offices. On the Commission’s calculations, this resulted in approximately €111 billion of profits being allocated to the stateless head offices between 2004 and 2014.Footnote 26
This profit allocation was formalized in APAs issued by Irish Revenue in 1991 and 2007.Footnote 27 The rulings were underpinned by a cost-sharing agreement (CSA) under which ASI and AOE held royalty-free rights to exploit the Apple Group IP outside the Americas and bore associated costs and risks. Apple relied on that arrangement to portray the head offices as the “entrepreneurial,” or risk–bearing, drivers. Accepting this premise, the rulings endorsed a bespoke version of the Transactional Net Margin Method (TNMM). Under that method, the Cork branches were treated as routine “tested parties” and taxed only on a small markup over their operating costsFootnote 28 —a method which resulted in an effective tax rate on European profits falling to as low as 0.005% by 2014.Footnote 29
In TLPE terms, these advance pricing agreements performed a form-giving function. They legally constituted a distributional order in which the residual profits associated with Apple’s intellectual property were attributed to entities with no tax residence, thereby shaping both the location of taxable income and the competitive conditions under which Apple operated within the internal market. As outputs of a national administrative practice, the APAs defined the micro-level institutional arrangements that State aid control would later seek to contest.
Against this backdrop, the Commission’s investigation hinged on benchmarking these APAs against the arm’s length principle (ALP) to determine whether profits allocated to the Irish branches corresponded to what would have been obtained under normal market conditions.Footnote 30 To conduct this assessment, the EC applied the Authorized OECD Approach (AOA), which treats permanent establishments as separate entities and analyzes their functions, assets, and risks.Footnote 31 The Commission’s examination revealed that the stateless head offices performed no significant functions related to Apple’s IP licenses, while the Irish branches actively executed functions and assumed risks integral to managing these licenses—including activities critical to Apple’s EMEIA market presence.Footnote 32 The EC concluded that under arm’s length conditions, these branches would never have agreed to allocate profits from the IP licenses to essentially functionless head offices.Footnote 33
This misallocation led the Commission to conclude that the rulings constituted State aid within the meaning of Article 107(1) TFEU because: (i) they were imputable to Ireland and entailed a loss of State resources through foregone tax revenue;Footnote 34 (ii) they were liable to affect intra-Union trade;Footnote 35 (iii) they conferred a selective economic advantage on ASI and AOE;Footnote 36 and (iv) they distorted or threatened to distort competition.Footnote 37 Viewed through TLPE’s macro–meso–micro framework, the Commission’s decision represented an attempt to use State aid control as a meso-level instrument to intervene in the outcomes produced by a Member State’s micro-level tax administration practices. Formally the measures were individual, but the reasoning reached the ruling practice that produced them, with consequences for profit allocation across the internal market. On this basis, it ordered Ireland to recover up to €13 billion in unpaid taxes from ASI and AOE for the period 2003–2014, plus interest.Footnote 38
II. The General Court—A Jurisprudential Brake on Transformative Ambition
In challenging the Commission’s primary line of reasoning, Ireland, ASI and AOE argued that Section 25 TCA 97 permitted the attribution to the Irish branches only of profits arising from activities they themselves carried on.Footnote 39 The Commission, they claimed, had adopted an impermissible “exclusion approach”: it attributed the licence-derived profits to the branches because the head offices lacked employees and physical presence, rather than positively establishing that the branches performed the functions and controlled the assets responsible for generating those profits.Footnote 40 In this case, the GC sided with Ireland and Apple. Crucially, the GC did not challenge the Commission’s underlying methodological framework—it accepted both the ALP and AOA as valid analytical tools for determining selective advantage.Footnote 41 Rather, the GC found fault with the Commission’s specific application of these principles, identifying three key deficiencies in its analysis: (i) a misinterpretation of section 25 TCA 97;Footnote 42 (ii) failure to prove that the Irish branches actually controlled the IP licenses;Footnote 43 and (iii) excessive reliance on the absence of employees at head offices rather than conducting a comprehensive functional analysis as required by the AOA.Footnote 44
The GC clarified that the AOA requires a detailed analysis of the functions actually performed within the PE—something the Commission had not adequately demonstrated.Footnote 45 The Court’s reasoning on control hinged significantly on the Dataproducts precedent,Footnote 46 which established that profits derived from property not controlled by a non-resident company’s Irish branch cannot be attributed to that branch—even if the property has been made available to it.Footnote 47 According to the GC, “only the profit derived from the trading activity of the Irish branches, including that carried on on the basis of the Apple Group IP licences held by ASI and AOE, should be regarded as relating to the activities of those branches.”Footnote 48 What the Commission had done, in the GC’s view, was to improperly apply an “exclusion approach”—identifying functions performed by the company as a whole without analyzing branch-specific activities, then defaulting profits to the branches when those functions could not be allocated to the head offices.Footnote 49 This approach contradicted both Section 25 TCA 97 and the AOA’s requirements for functional analysis.
Doctrinally, the judgment turned on evidentiary standards and the interpretation of national tax law. Institutionally, however, it had broader implications. By reviving a property- and control-based allocation logic grounded in Dataproducts, the General Court reaffirmed a formalist approach that constrained the scope for State aid control to operate as a meso-level instrument of market governance. This reasoning treated existing legal structures—particularly the interpretation of Irish law and the link between legal ownership and profit attribution—as binding limits on intervention, even where those structures generated significant competitive and distributive effects. In TLPE terms, the judgment re-inscribed a form of “false necessity,” narrowing the capacity of competition law to confront entrenched profit-shifting arrangements.
III. The Court of Justice—Reasserting Transformative Potential?
The Commission appealed, aiming to dismantle the jurisprudential barrier the GC had erected. The appeal was based on two principal grounds, though the CJEU found the first sufficient to annul the GC’s decision and concentrated its analysis accordingly. This section analyzes that reasoning to assess, in TLPE terms, whether the Court re-established State aid law as a form-giving legal instrument capable of mediating between firm-level tax arrangements and the broader political economy of corporate tax abuse.
1. Challenging the “Exclusion” Approach Finding: First Part of Appeal
The Commission first contested the GC’s finding that it had improperly used an “exclusion” method to allocate profits to ASI’s and AOE’s Irish branches.Footnote 50 As noted above, the GC had concluded the Commission failed to adequately demonstrate that the branches performed functions justifying the allocation of IP-related profits, relying too heavily on the absence of activity at the head offices.Footnote 51
The CJEU approached this by reaffirming the fundamental State aid process: comparing the contested measure against the ‘normal’ tax system—the reference framework—Footnote 52 to identify any preferential deviation.Footnote 53 In this case, the benchmark was Ireland’s general corporate tax system, which aims to tax profits generated in Ireland.Footnote 54 Within this system, non-resident companies with branches, like ASI/AOE, are comparable to resident companies regarding profits generated locally.Footnote 55 Therefore, the crucial question was whether the profit allocations under the rulings departed from the arm’s-length benchmark that the General Court had identified under Section 25 TCA 97.
Crucially, the CJEU endorsed the Commission’s functional analysis—comparing the functions, assets, and risks of the head offices versus the branches regarding the IP licenses—as consistent with the first step of the AOA.Footnote 56 By linking the findings of (1) minimal functions/risks at the head offices and (2) significant functions/risks at the branches to justify allocating the IP profits to the branches, the Commission had not used a simplistic “exclusion” method, but rather conducted the type of substance-based analysis required.Footnote 57 The CJEU validated the Commission’s substance-based analysis, directly countering the GC’s formalism.
From a TLPE perspective, this reasoning is significant because it restores State aid control as a form-giving legal instrument capable of mediating between micro-level tax arrangements and their systemic effects. By allowing economic substance to displace formal allocation constructs, the Court reopened a space for competition law to address corporate tax abuse not merely as a matter of technical compliance, but as an institutional practice with consequences for the organization of market power within the internal market.
2. Excluding Apple’s Inc. Functions: Second Part of Appeal
The Commission also argued that the GC erred by improperly considering the functions of the US parent company, Apple Inc., in its assessment and by relying on inadmissible evidence related to Apple Inc.’s role. The CJEU first addressed the procedural error, agreeing that the GC should not have relied on evidence regarding Apple Inc.’s strategic decision-making submitted late in the process, as the onus was on Apple/Ireland to provide relevant evidence during the administrative phase.Footnote 58
More significantly, the CJEU addressed whether the GC erred substantively by including Apple Inc.’s functions in Cupertino.Footnote 59 While the GC identified the AOA as the proper framework for branch profit attribution, it deviated by factoring in Apple Inc.’s activities in Cupertino to conclude the IP licenses should not be allocated to the Irish branches.Footnote 60 The CJEU found that such an approach distorted the AOA’s purpose and improperly questioned the reference system itself. The correct application of the AOA’s first step involves comparing functions, assets, and risks within the same legal entity—ASI or AOE—comparing the head office and the branch. It necessarily excludes the role of separate legal entities like the parent company, Apple Inc.Footnote 61
The CJEU rejected Ireland and Apple’s reliance on Apple Inc.’s ownership and group-level functions to determine the allocation of profits within ASI and AOE.Footnote 62 Instead, it emphasized that the AOA’s functional comparison ensures profit allocation aligns with the actual distribution of functions and risks within the same entity, consistent with the ALP.Footnote 63 By including Apple Inc.’s activities related to strategic decisions and R&D, the GC had incorrectly applied the Significant People Functions (SPF) analysis outside its proper intra-entity context. This is because the SPF analysis under the AOA is designed for profit attribution within a single entity—where risks and assets implicitly follow functions due to the absence of legally binding internal agreements—and is not meant for allocating profits between separate associated legal entities.Footnote 64 In this context, the CJEU observed that Apple Inc.’s functions related to the intellectual property were legally separated from ASI and AOE’s profit attribution as a direct consequence of the Group’s own structuring choices—specifically through its use of a cost-sharing agreement.Footnote 65 On this basis, the CJEU upheld the Commission’s appeal: Apple Inc.’s functions could not be considered when attributing profits to the Irish branches.Footnote 66
From the TLPE perspective, this boundary-setting is critical. It reflects the form-giving role of law in determining which economic activities count for purposes of profit attribution and which do not. By rejecting the General Court’s reliance on Apple Inc.’s group-level functions, the Court of Justice reaffirmed that the legal architecture chosen by the firm—in this case, the separation created through the cost-sharing agreement—also structures the terrain on which economic substance is assessed. Global multi-jurisdictional corporate groups cannot invoke group-level economic reality to escape the distributive consequences of the legal forms they have adopted.
3. The Final Judgment Process: Res Judicata and Jurisprudential Tensions
Beyond the core substantive reasoning, the procedural path to the CJEU’s final judgment and its application of res judicata add another layer to the analysis of the case’s implications. Deciding it had the necessary information based on the exchanges before the General Court, the CJEU opted under Article 61 of its Statute to give final judgment itself,Footnote 67 rather than remitting the case back to the GC as the Advocate General had suggested.Footnote 68 Specifically, the CJEU noted that Ireland and Apple had not filed cross-appeals challenging the GC’s rejection of certain arguments, including those concerning the identification of the reference framework under Irish law and the applicability of the ALP and the AOA for assessing profits under Section 25 TCA 97.Footnote 69 Consequently, the CJEU stated that these specific GC findings had the force of res judicata for the purpose of the appeal, meaning the Court of Justice considered itself bound by them and did not need to rule on the substance of those particular complaints again.Footnote 70
This procedural dimension became central to subsequent debates about the judgment’s coherence and broader political-economic significance. As noted earlier, the interplay between procedural finality and substantive review feeds into the “stubborn gap:” Apple’s apparent legal victory appears less secure as a stable precedent if its jurisprudential foundations are seen as context-contingent. In invoking res judicata, the CJEU cited its earlier case-law in Fútbol Club Barcelona,Footnote 71 which established that General Court findings not challenged through a cross-appeal generally acquire binding force. However, the relevance of this precedent was questioned by critics,Footnote 72 who pointed out a key distinction: in Barcelona, the losing party failed to appeal, whereas in Apple, Ireland and Apple had prevailed before the GC—raising doubts about which findings they could or should have cross-appealed.Footnote 73
From the TLPE perspective, these procedural ambiguities matter because they shape the form-giving effect of the judgment. The reliance on res judicata insulated key elements of the Irish-law reference framework from further scrutiny, even as the Court set aside the General Court’s outcome. As a result, Apple’s significance as a precedent is partly contingent on the procedural posture in which it arose. This contributes to the “stubborn gap” identified earlier: the case produced a high-profile legal victory, but one whose institutional generalizability is constrained by the procedural architecture through which it was delivered.
4. The Final Judgment: TLPE Implications
Did Apple demonstrate that EU competition law can function as a meso-level lever? The answer is a qualified yes, and the qualification is central. The Court of Justice’s 2024 judgment represents the most significant challenge to date to the assumption that bespoke tax rulings are insulated from systemic scrutiny. By privileging functional and risk-based analysis over the formal attribution of the royalty-free licences, the Court treated the allocation of their associated profits to the stateless head offices as contingent rather than legally necessary. In TLPE terms, State aid control here operated as a form-giving device, reshaping the normative terrain of profit allocation within the internal market and requiring corporate structures to correspond more closely to economic activity.
At the same time, Apple now sits within a fractured and unsettled jurisprudential landscape. While the judgment confirms that State aid law can, in principle, connect firm-level tax arrangements to broader concerns about corporate tax abuse, the failures that followed the launch of the Commission’s post-2013 investigations in Fiat, Amazon, Engie, Starbucks, and UK CFC demonstrate how narrowly this possibility is confined. In those cases, the courts either raised evidentiary thresholds or deferred to formal national-law structures, signaling a doctrinal ceiling on the extent to which State aid control can be used to unsettle entrenched profit-shifting structures.
Even within Apple, procedural architecture limited the scope of its institutional reach. The Court’s reliance on res judicata insulated key elements of the national reference framework from reconsideration, allowing a clear outcome in this case while leaving its generalizability uncertain. The judgment therefore exposed the contingency of the legal structures that had sheltered aggressive tax planning, but it did not convert that exposure into a stable mode of governance. As TLPE would predict, legal innovation was sufficient to disrupt an entrenched arrangement, but not to reconstruct the institutional political-economic order in which it was embedded. Section D examines the doctrinal constraints that explain why this gap persists.
D. The Missing Wave—From Apple’s High-Water Mark to the Commission’s Defeats
This section analyzes the “stubborn gap” at the judicial level. It first identifies three points of doctrinal friction within the fractured jurisprudential landscape: the construction of the reference framework, the tension between functional and risk-based analysis and formal attribution, and the evidentiary burden created by information asymmetries. It then engages with leading critiques of the Apple judgment, addressing questions of technical coherence and institutional legitimacy. Clarifying these judicial constraints is essential for understanding the dynamics examined in Section E, which turns to the broader political and geopolitical forces shaping the political economy of EU tax governance.
I. The Reference-Framework Battleground
A crucial point of divergence in State aid cases involving tax rulings lies in the identification and application of the “reference framework”—the benchmark of “normal” taxation against which a selective advantage is assessed.Footnote 74 In TLPE terms, the reference framework is itself a form-giving device: whoever defines “normal taxation” shapes the terrain on which selectivity is assessed, determining what counts as deviation and what counts as baseline. In Apple, the Commission advanced an arm’s-length reading of Ireland’s Section 25 TCA 97, governing the taxation of non-resident branches. The General Court held that the provision supported an arm’s-length allocation and that its application broadly corresponded to the functional analysis conducted in the first step of the AOA.Footnote 75 Because Ireland and Apple did not challenge those findings on appeal, the CJEU proceeded within that framework without independently resolving the content of Irish law.Footnote 76 Section 25 states that taxable profits include “any trading income arising directly or indirectly through or from the branch or agency, and any income from property or rights used by, or held by or for, the branch or agency.”Footnote 77 On that procedurally fixed basis, where Ireland accepted an allocation that ignored the branch’s core functions, including procurement, sales and distribution, and after-sales support, the Commission could compare the rulings with an AOA outcome and establish a derogation.Footnote 78
In the Luxembourg trilogy—Fiat, Amazon and Engie—the Commission’s attempts to replicate the Apple playbook faltered. This was primarily because Luxembourg’s specific national tax rules did not offer the same general textual basis for an inference of an implicit arm’s-length logic as Ireland’s Section 25 TCA 97 did—even though the Commission tried.Footnote 79 In Fiat, the CJEU ruled that the Commission erred by imposing an EU-wide arm’s-length standard not grounded in Luxembourg’s specific transfer pricing rules under Article 164(3) of the Luxembourg Tax Code (LTC),Footnote 80 rules the Commission had declined to take into account when defining the reference system.Footnote 81 In Amazon, the CJEU again stressed that the reference framework must derive solely from Luxembourg’s legal provisions, not external standards like the OECD Guidelines, which were not incorporated into domestic law at the time.Footnote 82 Similarly, in Engie, the Commission built a reference framework limited to Articles 164 and 166 LTC, treating the exemption for participation income as conditional on prior taxation at subsidiary level—a reading the CJEU held it had failed to establish as prevailing in Luxembourg law.Footnote 83
In Starbucks, the Commission’s difficulty arose one step later. Despite Dutch law incorporating the ALP, the General Court found the Commission’s evidence insufficient to establish a deviation from the reference framework.Footnote 84 The Commission could not prove that its preferred transfer pricing method (Comparable Uncontrolled Price) was more appropriate than the one Starbucks used (TNMM).Footnote 85 This highlighted a core issue: even when a Member State’s tax system explicitly aligns with the ALP, disputes over the right transfer pricing method can derail the Commission’s efforts to prove selective advantage. In UK CFC, the Commission misidentified the reference framework by treating the Controlled Foreign Company (CFC) rules as the baseline in isolation, when the CJEU held they formed an integral part of the general corporation tax system.Footnote 86 The Commission’s ambition to redefine what counts as “normal” taxation is bold—but it risks backfiring if the fundamentals are not nailed down. Legal reimagination? Yes—but within reason.
Transmogrification? No—superimposing an externally constructed or misconceived “normal” tax system stretches State aid control beyond its legitimate bounds.
This is what distinguishes Apple: Section 25 of the Irish statute provided an ample indeterminacy, giving the Commission’s functional arm’s-length reading a domestic foothold that the General Court accepted and that, once left unchallenged on appeal, the CJEU applied without independently determining the content of Irish law. In TLPE terms, Apple exposed the contingency of formal profit allocations, but that contingency was institutionally bounded. The apparent rupture in “false necessity” depended on the specific legal architecture of Irish branch taxation and could not readily be generalized to Member States with more tightly specified or formally anchored tax systems.
II. Function-and-Risk vs. Formal Attribution
The difference in national reference frameworks had direct consequences for the relevance of functional analysis. In Apple, the CJEU proceeded within a framework that permitted the rulings to be tested against an arm’s-length, intra-entity allocation broadly corresponding to the first step of the AOA. Within that framework, the Court upheld the Commission’s analysis on the combined basis that the head offices performed no relevant functions or risk management, while the Irish branches performed the functions and assumed the risks associated with the IP licences.Footnote 87 State aid, in this context, was reimagined as a meso-level lever—enabled by just the right alignment of legal ambiguity, domestic anchoring, and institutional momentum. By contrast, in the Luxembourg troika, the CJEU rejected the Commission’s benchmarks and interpretations as insufficiently grounded in national tax law.Footnote 88 Here, the reimagination of State aid collided with a culture of black-letter formalism, narrowing the legal space to challenge entrenched power and economic privilege. Without a benchmark or interpretation securely embedded in Luxembourg’s reference framework, the Court found no legal anchor to support a purely economic or functional analysis—thereby insulating formal allocation methods, however artificial, from substantive scrutiny.
From a TLPE perspective, Apple represents a moment in which State aid law briefly operated in a substance-oriented mode, allowing functional analysis to displace an ownership-based allocation in the governance of corporate tax arrangements. The subsequent Luxembourg cases marked a reassertion of formal legal boundaries grounded in national tax law, demonstrating how effectively actors invested in existing fiscal arrangements can mobilize legal form to resist redistributive intervention. Apple succeeded not because those constraints were absent, but because, under a particular statutory and evidentiary configuration, functional analysis could operate within them.
The resulting jurisprudence reveals a structural tension within EU law. In Apple, the Court permitted functional and risk-based analysis to guide profit attribution. In Fiat and Amazon, the Court insisted that transfer-pricing benchmarks remain anchored in the detailed rules of domestic law.Footnote 89 In Engie, it applied the same national-law discipline to the Commission’s economic reinterpretation of Luxembourg tax provisions.Footnote 90 The contrast reflects an unresolved conflict between the integrative ambitions of EU competition law and the continued centrality of national fiscal sovereignty in tax matters.
Seen through a TLPE lens, this tension limits the institutional capacity of State aid control to serve as a stable instrument for governing corporate tax abuse. Apple demonstrates that law can, in specific configurations, unsettle entrenched profit-allocation structures. The Luxembourg cases show that, absent a comparable legal and evidentiary anchor, legal form and national competence remain powerful barriers to institutional transformation. Together, they suggest that the reach of State aid law in this field is contingent, episodic, and politically bounded rather than systemically transformative.
III. Burden of Proof & Evidentiary Asymmetry
In State aid tax cases, the Commission shoulders the onus probandi: it must concretely demonstrate that a tax ruling confers selective advantage.Footnote 91 As noted above, this challenge is illustrated in Starbucks. Footnote 92 The GC emphasized that the Commission must have shown the taxpayer’s method was erroneous and resulted in a non-arm’s length outcome, not just that an alternative method might be preferable.Footnote 93 This evidentiary bar positions the Commission as a reviewer of national tax decisions, not a substitute tax authority, respecting Member States’ fiscal autonomy.
TLPE draws attention to the power asymmetries embedded in this evidentiary regime. This burden is compounded by a significant evidentiary asymmetry.Footnote 94 The Commission relies on Member States and multinational enterprises (MNEs) for critical documentation—data these parties have no incentive to disclose. In Apple, the Commission leveraged detailed records like CSA and subsidiary ledgers, which Ireland and Apple did not successfully rebut. Contrast this with Fiat. Luxembourg stonewalled from the start, redacting taxpayer identities and withholding information on fiscal-secrecy grounds.Footnote 95 Only after unanswered reminders did Brussels press the nuclear button of an information injunction under Article 10 of Regulation 659/1999.Footnote 96 Even so, Luxembourg insisted that part of the requested information constituted “business secrets” beyond its possession.Footnote 97
Moreover, when the Commission broadened its inquiry to analyze Luxembourg’s treatment of other intra-group financing rulings, it found deeper inconsistencies. Three rulings in the sample—Nos 1, 5, and 22—applied the Circular to intra-group financing arrangements without any accompanying transfer pricing report, which the Commission considered contrary to the requirements of Circular 164/2.Footnote 98 While those missing reports did not call Fiat’s individual file into question, they complicated the Commission’s effort to identify a coherent domestic benchmark, reinforcing its view that Luxembourg’s practice lacked a uniform, objective standard. If Luxembourg applies the Circular erratically, there is no stable “normal taxation” yardstick against which to measure selectivity.
The appeal, however, never reached that terrain. Applying the Circular, the Commission had derived FFT’s arm’s-length remuneration from external market data on required returns on equity, including BIS and ECB studies.Footnote 99 The CJEU held that, having dismissed the relevance of Article 164(3) of the LTC and Circular 164/2, the Commission had applied an arm’s-length principle different from that defined by Luxembourg law, confining itself to the abstract expression of that principle in the objective of the general Luxembourg corporate income tax system.Footnote 100 Because the reference framework was misidentified, the selective-advantage limb collapsed and the decision was annulled. Two years of contested disclosure, and the analysis it made possible, went unweighed.
Engie extended that evidentiary discipline from methodology to legal interpretation. When the Commission questions a Member State’s reading of its own tax provisions, it must in principle accept that reading where it is compatible with their wording, and may depart from it only on the basis of reliable and consistent evidence that another interpretation prevails in the case-law or administrative practice of that Member State.Footnote 101 The same logic already governed transfer pricing. Identifying a methodological flaw is not enough: the Commission must show distortion, not mere deviation.Footnote 102 Either way, the Commission must marshal sufficient legal and evidentiary material, anchored in domestic law, to sustain its case.Footnote 103
From a TLPE perspective, this evidentiary architecture operates as a structural filter. Only cases supported by unusually extensive, nationally anchored documentation can pass through. Strategic non-disclosure and informational control by tax authorities and multinational groups reinforce this asymmetry, systematically favoring existing arrangements. As a result, the meso-level potential of State aid control to address corporate tax abuse is confined to exceptional cases, pushing the Commission towards ex ante legislative instruments such as the ATADs and Pillar Two as more stable, if politically constrained, alternatives.
IV. Apple’s Critics and Its Discontents
The Apple judgment prompted substantial criticism from tax lawyers and scholars who are otherwise sympathetic to the Court’s doctrinal discipline.Footnote 104 Critics argued that the Court had effectively imported the Authorized OECD Approach into Irish lawFootnote 105 and had allowed the Commission to act as a de facto supranational tax authority.Footnote 106 These objections share a common premise: that the legal structures governing profit allocation in Apple were fixed constraints that the Court impermissibly displaced. A TLPE perspective reframes this claim. What appears in doctrinal critique as settled law is treated instead as a historically contingent institutional arrangement, sustained by legal form and administrative practice rather than by necessity.
This reframing does not dismiss the critics’ doctrinal objections, but it shifts the analytical focus from whether the Court departed from black-letter orthodoxy to whether that orthodoxy was ever as determinate as its defenders assume. Even on their own terms, the critics’ position is less secure than it appears. The doctrinal scaffolding traced in Section C. III shows that the Commission’s arm’s-length reading of Section 25 TCA 97 found a domestic foothold in the General Court’s conclusion that the provision’s application substantially corresponded to the first step of the AOA. Let’s recall that the GC had affirmed “essentially some overlap”Footnote 107 between the application of section 25 TCA 97 and the analysis conducted as the first step of the AOA. The CJEU, too, characterized that very statement as acknowledging “that section 25 TCA 97 corresponded in essence to the functional and factual analysis conducted as part of the first step of the AOA.”Footnote 108 Even critics of the CJEU’s application of the AOAFootnote 109 acknowledged that while the difference between those phrases might appear significant at first glance, “that conclusion would be excessive,”Footnote 110 noting that the GC had also agreed that the application of section 25 by the Irish tax authorities “overlaps, for the most part, with the analysis proposed by the AOA.”Footnote 111 Because those findings became final on appeal, the CJEU proceeded within the resulting framework rather than independently determining the content of Irish law. The resulting framework does not settle the critics’ retroactivity objection on the merits. It does, however, complicate the claim that the Court simply imported an autonomous EU arm’s-length principle into the Irish system.
A second cluster of criticism attacks the application of the AOA. One strand says the CJEU and Commission used an unduly narrow, intra-entity lens when attributing profits between ASI’s and AOE’s stateless head offices and their Irish branches. Richard Collier argues that, where appropriate, an AOA analysis should look beyond the single entity to group level activities.Footnote 112 Ryan Finley maintains that the Commission effectively locked Apple’s IP inside ASI and AOE, ran a superficial SPF face-off between the empty head offices and the Irish branches, and—because “something is more significant than nothing”—handed the spoils to the routine Irish staff, even though they created none of the Development, Enhancement, Maintenance, Protection, Exploitation (DEMPE) value and all real functions sat in Cupertino.Footnote 113
Similarly, Ruth Mason and Stephen Daly argue that a narrowly applied AOA—focusing only on the functions of head offices and branches—risks misaligning income allocation with economic substance. This would create a logical problem by ignoring the functions, assets and risks of the Apple group, where the most valuable functions were clearly performed by the parent company in Cupertino.Footnote 114 These objections, however, misconstrue the purpose and structure of the AOA in this context.Footnote 115 The AOA is designed to allocate profits within a single legal entity between a permanent establishment and the remainder of that enterprise, which is precisely the task posed by section 25 of the Irish tax code.Footnote 116 That provision taxes the trading profits of a branch, not those of the parent company, and the AOA’s second step expressly requires the identification and recognition of dealings between the permanent establishment and the rest of the same legal person. Within that framework, an intra-entity functional analysis is not a limitation but a methodological necessity. The choice to organize Apple’s intellectual property within ASI and AOE, rather than in Apple Inc., determined the legal and institutional boundary within which profit attribution had to be assessed. We leave readers to decide what is more economically real: the fiction of stateless head offices or the technical coherence of an intra-entity AOA analysis.
Another strand in the misapplication battlefield contends that the Commission failed to identify and allocate profits to an undeclared US PE of the Irish subsidiaries, ASI and AOE. According to this view, because the significant people functions associated with strategic and development decisions were performed in Cupertino, these activities de facto created a US branch of the Irish companies themselves.Footnote 117 Therefore, a correct AOA analysis would have attributed the lion’s share of the profits to this US PE, rather than arguing over a flawed allocation between the functionless head offices and the Irish branches. Whatever its conceptual appeal, this argument is not tenable within the legal framework of an EU State aid proceeding. It would have required the Commission to transform an EU State aid proceeding into an inquiry with the practical effects of US tax enforcement, well beyond its jurisdiction.
Ireland and Apple indeed pressed such a narrative, but they stopped short of pleading that ASI or AOE actually possessed a US PE.Footnote 118 The reason lies in Apple’s own corporate architecture. Through the cost-sharing agreement, the group allocated the costs and risks of developing the intellectual property to ASI and AOE, with the consequence that the associated profits were legally located within those entities and had to be allocated between their stateless head offices and their Irish branches. As Advocate General Pitruzzella observed,Footnote 119 and the Court of Justice confirmed, the State aid analysis had to proceed within the boundaries of that self-selected legal structure, “without regard to any role that may have been played by Apple Inc.”Footnote 120 To concede the existence of a US permanent establishment would have risked exposing Apple to substantial IRS scrutiny and would have undermined Ireland’s claim to taxing rights it was seeking to defend, while also pushing the case beyond the jurisdictional reach of EU competition law.
A third critique holds that the Apple judgment produces windfall revenues for low-tax jurisdictions and thereby entrenches distortions within the Union.Footnote 121 While economically salient, this objection misidentifies the function of State aid control. Article 107 TFEU enforces a form of horizontal neutrality within a Member State by preventing selective departures from national tax rules that favor undertakings in comparable factual and legal situations. The distributive location of recovered revenue is therefore orthogonal to the legality of the aid itself.
TLPE helps clarify this institutional mismatch. Apple represented the high–water mark of an enforcement strategy that relied on ex post competition law to confront corporate tax abuse, but its very exceptionality reveals the structural limits of that approach. State aid control can disrupt particular arrangements when legal and evidentiary conditions align, but it cannot provide a durable mechanism for macro-level redistribution. Apple therefore demonstrated not the scalability of judicial enforcement, but the need for ex ante legislative instruments such as the ATADs, the Directives on Administrative Cooperation, and Pillar Two to address the systemic dimensions of profit shifting.
The fragility of the Apple outcome underscores this point. The case succeeded because a specific configuration was present: a statutory foothold for arm’s–length reasoning, unusually extensive evidentiary access, and a political climate receptive to scrutiny of US multinational tax practices. In TLPE terms, these conditions briefly opened a window for institutional reconfiguration, but they did not constitute a stable governance architecture. Once any of these elements was absent, as in the Luxembourg, Dutch, and UK cases, the Commission again encountered the combined force of legal form, fiscal sovereignty, and informational asymmetry.
The Belgian excess-profit scheme makes the same point from the opposite direction. There the Commission ultimately prevailed, but only after the General Court annulled its decision in 2019, the Court of Justice set that judgment aside in 2021 and referred the case back, and the General Court upheld the original 2016 finding in 2023.Footnote 122 Six beneficiaries have since appealed, and in March 2026 Advocate General Kokott proposed that the appeals be dismissed; judgment remains pending.Footnote 123 That the Advocate General still had to address whether the Commission and the EU Courts are bound by a Member State’s reading of its own tax law shows how little the underlying question has settled.Footnote 124 A decade after the contested rulings, the legal position remains open. Winning, on this record, is not the same as building something that lasts.
It is against this background that the Commission’s parallel and ultimately more durable turn toward ex ante legislative instruments must be understood. With litigation proving effective only in narrowly defined and fact-specific circumstances, regulatory reform became the more plausible, if politically contested, vehicle for constraining corporate profit shifting.
E. From Momentum to Fragmentation: The Arc of EU Tax Governance after Apple
The Apple litigation unfolded within a broader institutional wave. From 2013 onwards, the OECD’s Base Erosion and Profit Shifting (BEPS) project, the Commission’s coordinated State aid investigations, and the adoption of ATAD 1,Footnote 125 ATAD 2,Footnote 126 and the DACs,Footnote 127 particularly DAC 6,Footnote 128 converged on a shared objective: constraining aggressive corporate tax planning. Enforcement and legislation advanced in parallel, with each reinforcing the other.
Apple therefore functioned both as proof of concept and as a political catalyst. It demonstrated that State aid control could, under exceptional conditions, pierce highly engineered tax structures. More importantly, the headline investigations, amplified by LuxLeaks, the Panama Papers, and related disclosures, helped shift corporate tax avoidance from the realm of “quiet politics” into a high-salience subject of public and political contestation.Footnote 129 They generated the public pressure and institutional momentum that made the ATADs, the DACs, and eventually Pillar Two politically feasible.Footnote 130 State aid enforcement was never intended to regulate transfer pricing at scale. Its function was to destabilize entrenched arrangements and create space for legislative reconstruction.
Yet the doctrinal constraints analyzed in Section D were only one part of the resistance the Commission encountered. The enforcement campaign also triggered sustained political opposition. Several Member States framed the State aid cases as an encroachment on fiscal sovereignty and a threat to their investment strategies.Footnote 131 Ireland’s protracted resistance to the Apple recovery order became emblematic of this stance.Footnote 132 At the same time, transatlantic pressure intensified. The US Treasury criticized the campaign as retroactive and discriminatory,Footnote 133 and the Trump administration framed it as an attempt to appropriate US tax revenues.Footnote 134 These dynamics curtailed the Commission’s willingness to pursue further litigation and help explain why no major new State aid investigations into tax rulings were launched even after Apple was upheld.
Legislative reform offered no straightforward alternative. The same sovereignty concerns that limited enforcement also constrained negotiations over harmonizing directives, while the legislative process faced its own institutional limits. The cumulative juridification of EU tax governance through overlapping directives generated increasing administrative and political fatigue. By 2025 this was visible in the Commission’s “Omnibus” and “Competitive Compass” initiatives, which explicitly prioritized regulatory simplification and burden reduction across a wide range of policy areas.Footnote 135 Gunther Teubner warned that such juridification can produce “legal pollution”—the destruction of the very economic life-world structures the law seeks to regulate.Footnote 136
The anti-avoidance agenda was no exception. On June 18, 2025, the EU Council abandoned the Unshell Directive after three years of failed negotiations.Footnote 137 Member States could not agree on substance thresholds or the interaction with existing DAC6 reporting obligations, and the proposal was shelved with only a vague commitment to incorporate some high-risk indicators into a future DAC6 revision.Footnote 138 Four months later, on October 21, 2025, the Commission listed the TPDFootnote 139 for withdrawal in its 2026 Work Programme, as part of its regulatory simplification agenda.Footnote 140 Meanwhile, Business in Europe: Framework for Taxation (BEFIT)—the proposed common corporate tax base framework—remains frozen,Footnote 141 with no clear roadmap for adoption. Footnote 142 Even beyond tax, the pattern held; the green transition agenda—once the centerpiece of von der Leyen’s political project—has been scaled back under pressure from industry and Member States invoking competitiveness concerns.Footnote 143
Most consequentially, Pillar Two was reconfigured to accommodate US opposition. On June 28, 2025,Footnote 144 G7 finance ministers announced a “side-by-side” agreement excluding US-parented multinational groups from both the Income Inclusion Rule (IIR) and the Undertaxed Profits Rule (UTPR) in respect of their domestic and foreign profits, in exchange for the United States withdrawing its proposed Section 899 retaliatory tax. The OECD Inclusive Framework adopted the Side-by-Side Package on January 5, 2026.Footnote 145 By April 2026, the OECD confirmed that the United States was the only jurisdiction added to the Central Record as having a “Qualified SbS Regime,” shielding eligible US-parented groups from the IIR and UTPR while leaving domestic minimum top-up taxes unaffected.Footnote 146 The US regime operating “side-by-side” with Pillar Two differs in key respects: it applies a 14 percent effective rate rather than 15 percent, and permits global blending of income across jurisdictions—allowing low-taxed profits to be offset against high-taxed profits elsewhere.Footnote 147 European competitors remain subject to foreign top-up taxation under the IIR and UTPR that their American counterparts avoid. The European Commission maintained that no amendment to Directive 2022/2523 was required,Footnote 148 relying instead on Article 32’s safe harbor mechanism to accommodate the exemption—a mechanism originally designed for administrative simplification rather than structural carve-outs.Footnote 149 The resulting framework thus privileges US multinationals over their European competitors while simultaneously—through its jurisdictional blending model—imposing heavier top-up taxation on cross-border operations within the internal market than on equivalent domestic arrangements, a design whose compatibility with the fundamental freedoms remains contested.Footnote 150
Against this backdrop, Apple’s legacy is necessarily ambivalent. The enforcement campaign did shift the baseline. Stateless head offices have disappeared,Footnote 151 DEMPE-based analysis is embedded in domestic tax practice, and general anti-abuse rules are now standard. But the more ambitious objectives—a common tax base, coordinated transfer pricing rules, and a universal minimum rate—encountered more durable political-economic resistance. Pillar Two survives only in truncated form, while the Unshell Directive and the TPD was listed for withdrawal.
What remains is a thinner anti-avoidance architecture. ATAD 1 and ATAD 2 continue to set minimum constraints through interest limitation, exit taxation, CFC rules, general anti-abuse provisions, and hybrid mismatch rules. The IIR and UTPR continue to apply to EU-headed and other non-US-parented multinational groups, with domestic top-up taxes capturing some revenue. But the three-sided structure that was meant to close the gaps left by enforcement—disciplining legal form (Unshell Directive), prices (TPD), and rates (Pillar Two)—now has two missing sides, and the third is selectively inoperative.
The TLPE perspective clarifies why both enforcement and legislation encountered structurally similar limits. In both tracks, law attempted to redesign the institutional foundations of corporate taxation by altering the form through which profits are legally constituted and allocated. State aid enforcement sought to reconfigure firm-level arrangements by exposing the contingency of transfer pricing and branch attribution. Legislative reform sought to entrench those insights ex ante through harmonized rules on substance, pricing, and minimum taxation. In each case, law functioned as a form-giving instrument aimed at redistributing market power and taxing rights.
In both tracks, however, institutional redesign confronted the same political–economic constraints. Member States defended fiscal sovereignty not only as a legal competence but as a revenue model and an investment strategy. The unanimity requirement under Article 115 TFEU transformed distributive conflict into institutional veto power.Footnote 152 At the global level, the asymmetric market and geopolitical power of the United States limited how far EU law could reallocate taxing rights without triggering retaliation. These constraints are not incidental. They are structural features of the political economy in which EU tax law operates.
TLPE therefore explains why Apple’s legal breakthrough did not translate into a durable governance regime. The judgment demonstrated that the formal attribution of licence-derived profits was contingent rather than necessary, but it did not alter the power relations and institutional dependencies that sustain them. Law could destabilize entrenched arrangements, but it could not, on its own, reconstruct the fiscal and political settlement that underpins them.
Apple thus occupies a distinctive position. It showed that State aid control can, under specific conditions, confront corporate tax abuse and prioritize economic substance over legal form. But it also marked the high–water mark of an institutional momentum that has since fragmented. The legislative turn that followed was not a rejection of Apple’s logic, but an attempt to stabilize it through more durable instruments. That attempt has proven politically and geopolitically constrained. What remains is a contested terrain in which legal capacity, legislative ambition, and political-economic resistance continue to interact. Apple revealed what was possible—the years since have revealed how difficult it is to institutionalize that possibility.
F. Conclusion
This Article began by asking whether the Apple judgment represents a failure or a success of EU State aid law. The answer is neither. Apple was a warranted intervention and a proof of concept. It showed what EU competition law can achieve when doctrinal footholds, evidentiary access, and political conditions align. By endorsing a substance-oriented approach to profit attribution, the Court of Justice confirmed that EU competition law can, under exceptional conditions, reach deeply embedded corporate tax structures.
But Apple was also more than an isolated legal defeat for a single multinational. It was the apex of a broader institutional experiment—the Commission’s attempt to use competition law to link firm-specific tax rulings to systemic questions of market power and profit shifting within the internal market. That experiment succeeded in this case. It did not, however, scale. Parallel investigations failed, enforcement momentum dissipated, and the meso-level bridge constructed through State aid law proved too fragile to sustain broader institutional transformation.
The “stubborn gap” traced in this Article is therefore not an anomaly to be resolved but a structural condition to be understood. The TLPE perspective clarifies why the enforcement strategy encountered doctrinal limits embedded in national tax law, while the legislative turn confronted the same political-economic constraints: fiscal sovereignty, unanimity, and geopolitical asymmetry. In both tracks, law operated as a form-giving instrument aimed at redistributing power and taxing rights, and in both tracks it ran up against the institutional and political boundaries that structure European tax governance.
The baseline has nonetheless shifted. Stateless head offices are no longer viable. DEMPE-based analysis is embedded in domestic tax systems. General anti-abuse rules are now standard, and Pillar Two persists, albeit in modified form. These changes matter. But the more ambitious objectives—a harmonized transfer pricing regime, effective substance rules, and a level playing field that includes US multinationals—have encountered more durable resistance.
Apple’s enduring legacy is therefore not a new equilibrium but a clearer understanding of what legal innovation can and cannot do. The case demonstrated that legal institutions can be reimagined and that entrenched arrangements are not immutable. It also revealed how deeply those arrangements are anchored in political economy. This is precisely what a TLPE perspective brings into view—law’s form-giving capacity to expose and unsettle existing distributions of power, and the structural constraints that limit how far that exposure can be converted into durable institutional change. What remains is a contested terrain in which legal capacity, legislative ambition, and political-economic constraint continue to interact. Apple showed what is possible—the period since has shown how difficult it is to make that possibility durable.
Acknowledgements
Earlier versions of this work build on feedback from Stephen Daly, Ruth Mason, Jean-Philippe Robé, Federica Casano, and Nessa Ní Chasaide, whose insights helped refine our analysis.
Competing Interests
The authors declare none.
Funding Statement
This research was supported by the European Research Council (ERC) under the European Union’s Horizon Europe research and innovation programme (Project: Democracy Challenged, Grant Agreement ID: 101125672). Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.
Open access
