The New Oligopoly
In recent years, the term "algorithmic collusion" has become firmly established in competition law. It refers to a variety of scenarios in which pricing software — increasingly widespread in digital markets — contributes in various ways to the emergence of collusive outcomes. Behind the term there is a well-known fear: that algorithms make collusion easier, more rooted and more difficult to detect, undermining antitrust tools built on assumptions that may prove increasingly outdated for digital markets.
Questions of Collusion
Not all forms of algorithmic collusion, however, pose the same challenges. When algorithms merely facilitate or implement collusion devised by humans — for example, by monitoring competitors' prices or automating retaliatory mechanisms — EU antitrust law already has the necessary tools to intervene. In these cases, Article 101 of TFEU — which prohibits agreements and concerted practices between undertakings aimed at restricting competition, particularly through price coordination — applies without any particular force: the agreement or concerted practice remains attributable to the undertakings, and the use of technological tools may, if anything, be relevant for assessing the gravity of the infringement.
Similarly, even in cases of explicit collusion implemented by algorithms — in which pieces of software interact autonomously with each other, giving rise to forms of market coordination that are functionally equivalent to a cartel — competition law is not devoid of answers. Through an evolutionary interpretation of the concept of "meeting of minds" (i.e. wills converging) and leveraging the principles of attribution of conduct to the company, it is possible to bring these phenomena within the scope of Article 101 of TFEU, even when coordination has not been planned in detail by human decision-makers.
When Algorithms Change the Rules of the Game
The real problematic issue, however, arises with reference to tacit collusion implemented by algorithms. In this case, software, acting autonomously and interdependently — that is, adapting its decisions to the moves made by competitors — achieves supracompetitive pricing outcomes without any kind of agreement, exchange of information or significant contact between the companies. This is, essentially, a form of oligopolistic interdependence mediated by technology. As is well known, tacit collusion — even when it engenders harmful effects for consumers — has traditionally remained outside the scope of Article 101 TFEU. And the same currently applies to algorithmic tacit collusion.
Yet, this is precisely where algorithms are changing the rules of the game. Unlike human tacit collusion, algorithmic collusion can be more rapid, more pervasive, and more stable. Algorithms reduce strategic uncertainty, increase market transparency and enable forms of coordination even in contexts that lack the classic characteristics of oligopoly. In this sense, algorithmic collusion rehashes — and simultaneously transforms — the traditional "oligopoly problem," extending it well beyond the boundaries within which antitrust law had sought to confine it.
Looking at the Effects
In the face of technological developments, simply noting the inapplicability of Article 101 TFEU risks creating a protection gap. Alongside traditional antitrust remedies designed to address the effects of tacit collusion — such as the extensive application of Article 101, the use of collective abuse of dominant market position or tools that control mergers — it then becomes necessary to consider complementary tools. From this perspective, a possible answer may come from the field of regulation. In particular, the idea of "visibility of outcomes" suggests shifting the focus from the identification of an agreement, or of direct or indirect contacts between companies — requirements that underpin the application of Article 101 TFEU and which are often absent in cases of tacit algorithmic collusion — to the observable effects produced by algorithms on the market itself. If companies benefit from the use of automated systems, they should also be held accountable for the outcomes such systems generate, regardless of whether the outcomes can be traced back to a form of coordination in the traditional sense.
In sum, algorithmic collusion does not necessarily require rewriting competition law, but it does increasingly force a rethinking of the latter’s assumptions. In an economy governed by algorithms, the problem of oligopoly hasn't disappeared: it has simply changed form.
