Case Overview
In Luxembourg and Amazon v Commission (EU General Court, 2021), the Court annulled the Commission’s State aid decision on a Luxembourg tax ruling approving an IP royalty structure. The Commission had not proven that the ruling produced a selective tax advantage by reducing Amazon’s Luxembourg taxable base compared with normal taxation.
Key Facts
- A planned 2003 restructuring was implemented in 2006 using two Luxembourg entities: LuxSCS (a limited partnership) and LuxOpCo (an operating company).
- LuxSCS held European IP exploitation rights via buy-in and cost-sharing arrangements with US group entities and licensed those rights to LuxOpCo.
- A 6 November 2003 tax ruling approved the proposed royalty method and confirmed LuxSCS’s Luxembourg tax treatment as a transparent partnership absent a Luxembourg PE.
- The ruling was applied from May 2006 to June 2014.
- In 2017, the Commission ordered recovery estimated at ~€250m plus interest.
The Dispute
The Commission considered that Luxembourg, through the 2003 ruling (together with acceptance of annual returns based on it), granted Amazon a selective advantage under Article 107(1) TFEU by endorsing a royalty that shifted profits from LuxOpCo (subject to Luxembourg corporate income tax) to LuxSCS (not subject to Luxembourg corporate income tax at entity level). The royalty formula was designed to leave LuxOpCo with a “routine” return (including a floor/ceiling expressed as a small percentage of EU sales), with the residual profit paid to LuxSCS.
To show an advantage, the Commission re-ran the analysis using transfer pricing concepts drawn from OECD practice. In its primary line of reasoning, it treated LuxSCS as the least complex party and therefore the proper TNMM tested party, and it recalculated LuxSCS’s arm’s-length remuneration as essentially a pass-through of buy-in/CSA costs plus only a limited mark-up on certain “maintenance” costs. It also advanced three subsidiary findings (including that other methodological choices and a ceiling mechanism depressed LuxOpCo’s taxable profit).
Luxembourg and Amazon argued that the Commission’s functional analysis and recomputations were unreliable and did not demonstrate any actual reduction of LuxOpCo’s Luxembourg tax base under the domestic reference system.
Holding
The General Court annulled the Commission decision in full. It emphasised that the Commission bears the burden of proving an advantage and that identifying methodological flaws is not enough: the Commission must show that those flaws prevent a reliable arm’s-length approximation and that they actually lowered the tax burden under the national reference framework. On the facts, the Court found material errors in the Commission’s primary TNMM approach (including the tested-party switch and its recomputation of LuxSCS remuneration) and held the subsidiary findings also failed to prove advantage. The CJEU later dismissed the Commission’s appeal (C-457/21 P).
Takeaways for Practitioners
- In State aid disputes, link any transfer pricing debate back to the reference system and whether the alleged error changes the taxable base in practice.
- Tested-party selection and the reliability of a substitute computation can be decisive; document complexity, functions, and comparables.
- For intangibles/royalties, keep contemporaneous support for how residual profit is split (including floors/ceilings) and why it reflects the parties’ contributions.
Related Content
For practical guidance on defending IP outcomes, see Documentation for Intangibles. For core transfer pricing concepts that appear throughout both the Commission decision and the court litigation, revisit the Arm's Length Principle.
Frequently Asked Questions
Q1. What was the alleged State aid in Amazon Luxembourg?
The Commission alleged Luxembourg endorsed an inflated royalty from LuxOpCo to LuxSCS that reduced LuxOpCo’s taxable profits, conferring a selective advantage compared with normal corporate taxation.
Q2. Why did LuxSCS matter to the State aid analysis?
LuxSCS was a Luxembourg limited partnership treated as tax transparent (no Luxembourg corporate income tax at entity level absent a PE). Shifting more profit to LuxSCS via royalties could therefore reduce tax collected in Luxembourg.
Q3. What was the “tested party” issue under TNMM?
The Commission’s primary case depended on treating LuxSCS (not LuxOpCo) as the least complex entity and then applying a cost-based return to LuxSCS. The General Court held the Commission had not shown that approach produced a reliable arm’s-length benchmark or proved a reduced tax base.
Q4. Can the Commission apply an EU-wide arm’s-length principle or OECD Guidelines by default?
No. In the 2023 appeal, the CJEU stressed there is no autonomous EU arm’s-length principle for Article 107(1) TFEU; the analysis must be anchored in national law, and OECD Guidelines matter only if domestic law expressly refers to them.
Q5. Was this the final word?
Yes. The Commission’s appeal was dismissed on 14 December 2023 in C-457/21 P (EU:C:2023:985), leaving the annulment in place.