Presumption of Guilt? The EU Commission's Draft Article 102 Guidelines and the Burden of Proof Problem After the Intel case

By Alex Feeney

Article 102 TFEU is one of the foundational pillars of EU competition law. It is designed to prevent powerful companies from abusing their dominant position in a market, its primary goal being to ensure that competition remains fair so that consumers and society benefit from lower prices, better quality, and more innovation. However, this article has operated for over 15 years without formal EU Commission Guidelines. Every dominant company has navigated abuse of dominance rules through a patchwork of inconsistent case law and a non-binding 2008 Guidance Paper that the Commission repeatedly misapplied in court. This year, the Commission is finalising its first ever formal Guidelines on exclusionary abuses, the most significant attempt to codify dominance enforcement in a generation. Yet what appears to be progress on the surface is actually greater tension in EU courts: rather than consolidating the effects-based approach the courts have demanded, the Guidelines introduce legal presumptions of harm that reverse the burden of proof onto dominant companies, in direct tension with landmark CJEU rulings. Ultimately, the Guidelines represent an attempt by the EU Commission to bypass evidentiary standards its own courts have set and, unless revised, will likely generate the litigation they are supposed to prevent.

The 2008 Guidance Paper introduced an effects-based approach, committing the Commission to proving actual harm rather than treating conduct as automatically abusive. Yet it was not formally binding, and the Commission consistently failed to apply it. In Intel, the Commission imposed a €1.06 billion fine presuming exclusivity rebates were inherently abusive; the CJEU annulled it in October 2024 because proper effects analysis had never been conducted. By 2024, the case for formal Guidelines was unanswerable: digital markets had created new forms of abuse the 2008 framework never anticipated, Article 102 remained the only major area of EU competition law without Guidelines, and inconsistent enforcement across 27 national competition authorities was producing divergent outcomes across the single market, a compliance vacuum multiple law firms had explicitly flagged.

The Guidelines introduce a two-step framework: conduct is abusive if it departs from "competition on the merits" and is capable of producing exclusionary effects. Conduct is then divided into three categories carrying different presumption regimes. Category I covers naked restrictions, conduct with no conceivable pro-competitive justification, such as paying a customer to delay adopting a rival's product, where the presumption of harm is practically irrebuttable. Category II covers common commercial conduct including exclusive dealing, loyalty rebates, predatory pricing, and certain forms of tying, where a rebuttable presumption applies, once the conduct is established, harm is presumed and the company must disprove it. Category III requires full effects analysis.

Category II's presumption of harm is precisely where the Guidelines collide with settled law. Just 84 days after the draft Guidelines were published, the CJEU annulled the Commission's Intel fine in Commission v Intel (C-240/22 P, October 2024). The court's central finding was unambiguous: where a dominant company submits economic evidence that its conduct was not capable of restricting competition, the Commission cannot rely on a presumption of harm, it must conduct a full as-efficient competitor (AEC) test and effects analysis. Exclusive rebates, the very conduct at issue in Intel, sit squarely within the Guidelines' Category II. If adopted unchanged, every enforcement decision relying on Category II presumptions faces immediate appeal on Intel grounds. The Guidelines risk becoming a litigation roadmap rather than a compliance framework.

The conceptual problems run deeper. “Competition on the merits” - the gateway concept of the entire two-step test - is never operationally defined in the Guidelines. Mario Draghi's September 2024 competitiveness report warned they leave "excessive discretion" to the Commission and lack the "operational value" needed for legal certainty. A concept the Commission can define differently in each enforcement case is not a legal standard; it is enforcement discretion dressed up as legal certainty.

The timing of the Guidelines could not be more politically contradictory. The conduct covered by Category II, loyalty rebates, exclusive dealing, product tying, is not inherently harmful. A presumption-based system that treats all of these as suspect will inevitably punish pro-competitive conduct, what economists call a Type I error. This directly contradicts the Draghi Report's central demand that EU competition law support dominant European companies scaling up to compete against US and Chinese rivals. The December 2025 Lukoil Bulgaria judgment (C-202/23) reinforces the point; the CJEU confirmed that even for privatised infrastructure cases, strict fact-specific analysis under the Bronner doctrine is required before Article 102 liability is imposed. Intel and Lukoil form a consistent and strengthening judicial signal: the CJEU will not accept enforcement shortcuts.

The Guidelines will be adopted in Q1/Q2 2026, that much is certain. The question is whether the Commission has revised the presumption framework in light of Intel or will publish a document the CJEU has already implicitly rejected. Get it right and the Guidelines close a 15-year enforcement gap. Get it wrong and they simply move the legal battle from investigation to appeal, leaving dominant companies, national competition authorities, and courts across 27 Member States navigating the same uncertainty the Guidelines were designed to end. The Commission needs enforcement wins. The CJEU demands proof. Until those two positions are reconciled, presumption of guilt remains a policy ambition the courts have consistently refused to endorse.

Edited by Artyom Timofeev