Case Overview
In Joined Cases T-778/16 and T-892/16 (Ireland v Commission; Apple Sales International and Apple Operations Europe v Commission), the EU General Court (15 July 2020) annulled the European Commission’s Apple–Ireland State aid decision. The Court held that the Commission had not proven to the requisite legal standard that the Irish tax rulings for Apple’s Irish-incorporated, non-resident companies (ASI and AOE) conferred a selective advantage under Article 107(1) TFEU.
Note on later history: In 2024, the Court of Justice set aside the 2020 judgment and confirmed the Commission’s 2016 decision (Case C-465/20 P, EU:C:2024:724).
Key Facts
- Tax rulings: Irish Revenue issued rulings on 29 January 1991 and 23 May 2007 approving methods to attribute profits to the Irish branches of ASI and AOE.
- Corporate status: ASI and AOE were incorporated in Ireland but not tax resident in Ireland; each operated an Irish branch.
- Commission decision (2016): The Commission found unlawful and incompatible State aid and ordered recovery (Decision (EU) 2017/1283, adopted 30 August 2016).
- Amount and period: The Commission estimated ~€13 billion plus interest, with recovery ordered for 12 June 2003 to 27 September 2014 (time-bar/limitation rules meant earlier years were not recovered).
- Theory of advantage: The Commission argued that profits linked to IP licences were wrongly allocated away from taxable Irish branches to “head offices” with little/no substance.
- Procedure: Ireland and Apple challenged under Article 263 TFEU; the General Court annulled the decision (2020).
The Dispute (State Aid + Profit Attribution)
The Commission analysed the rulings using the familiar State aid framework: identify the reference framework (the “normal” Irish corporate tax rules) and assess whether the measure derogated in a way that conferred a selective advantage.
A central plank was Irish law on taxing non-resident companies trading through an Irish branch (discussed via section 25 TCA 1997 and related principles): the Commission argued that applying Irish corporate tax “normally” required an allocation of profits consistent with an arm’s-length outcome (with OECD materials used as guidance, including the Authorised OECD Approach on profit attribution to PEs).
The Commission’s three lines of reasoning
- Primary: The “head offices” of ASI/AOE had no employees/physical presence, so they could not control or manage the IP licences; therefore, IP-related profits should have been attributed to the Irish branches.
- Subsidiary: Even if IP stayed outside Ireland, the methods endorsed by the rulings (tested party, profit level indicator, return levels) contained methodological errors that reduced Irish taxable profits.
- Alternative: The rulings reflected discretion and constituted a derogation from the Irish reference framework (including section 25), producing selectivity.
Holding (2020 General Court)
The General Court annulled the Commission decision in full because the Commission did not prove that the rulings produced a selective advantage (Press Release No 90/20; Judgment ECLI:EU:T:2020:338).
Key points reflected in the Court’s reasoning:
- OECD/arm’s-length tools are not free-standing EU law. The Commission could not treat Article 107(1) TFEU as imposing an autonomous arm’s-length principle regardless of national law; any benchmarking had to be anchored in the identified Irish reference framework.
- Primary line failed: The Court found the Commission had not demonstrated that the profits the Commission targeted were in fact attributable to the Irish branches based on what those branches actually did (and it criticised allocating profits by default merely because head offices lacked employees).
- Subsidiary line failed: Even if the rulings were incomplete or inconsistent, the Court held that identifying methodological issues was not enough, by itself, without showing that the rulings actually reduced ASI’s/AOE’s tax burden compared with normal taxation.
- Alternative/discretion line failed: The Commission did not establish that the rulings resulted from discretion conferring a selective benefit.
Subsequent Development (2024 CJEU)
On appeal, the Court of Justice set aside the General Court’s judgment and gave final judgment, confirming the Commission’s decision and the recovery obligation (C-465/20 P, EU:C:2024:724; Press Release No 133/24). Practically, this means the Commission’s recovery order (estimated €13bn plus interest) ultimately stood.
Takeaways for Practitioners
- For branch/PE profit attribution, keep a clear, evidence-backed functional analysis (functions/assets/risks) showing what the branch does versus what is done elsewhere.
- Maintain governance evidence that supports the location of key decision-making relevant to intangibles and risk control (minutes, mandates, delegation documents).
- If using cost-plus/operating-cost-based rulings or APAs, ensure the method is defensible factually and can be shown not to understate the local tax base versus the domestic “normal” framework.
- Expect State aid scrutiny to focus not just on method labels, but on whether the ruling outcome can be justified by the applicable domestic tax rules.
Related Content
See Documentation for Intangibles and Arm's Length Principle. For comparable fiscal State aid litigation, see Fiat and Starbucks.
Frequently Asked Questions
Q1. What were the contested Irish tax rulings?
Irish Revenue rulings from 1991 and 2007 approving how ASI and AOE would compute Irish-taxable profits attributable to their Irish branches.
Q2. Why did the Commission view the rulings as State aid?
Because (on the Commission’s case) the rulings endorsed a profit allocation that excluded IP-related profits from the Irish tax base, giving Apple a selective reduction in taxable profits compared with “normal” Irish corporate taxation.
Q3. Why did Apple and Ireland win at the General Court in 2020?
The General Court held the Commission had not proven that the rulings conferred a selective advantage, including failing to show that the Irish branches should have been allocated the contested IP-related profits.
Q4. Did the 2020 judgment bar using arm’s-length/OECD concepts in State aid cases?
No. But it required that any arm’s-length benchmarking be grounded in the domestic reference framework, not treated as a free-standing EU standard.
Q5. What is the current status after the appeal?
In 2024, the Court of Justice set aside the 2020 General Court judgment and confirmed the Commission’s 2016 State aid decision (Case C-465/20 P), reinstating the recovery order (Commission estimate: €13bn plus interest)....