One Year After Illumina/Grail – How Are EU Competition Authorities Now Dealing With Below-Threshold Mergers

About one year ago, the European Court of Justice (CJEU) ruled in its landmark Illumina/Grail judgment that the European Commission could not accept merger referrals from national competition authorities under Article 22 of the EU Merger Regulation (EUMR) unless those authorities had jurisdiction to review the transaction themselves (see our previous alert).

In this alert, we examine recent developments in EU merger control and take a look at how the Commission and national competition authorities are currently dealing with below-threshold mergers, and what the implications are for companies going forward.

I. Recap

In 2021, the Commission adopted guidance clarifying that a national competition authority could refer a transaction under Article 22 EUMR to the Commission for review, even if it was not notifiable under national merger control rules. This allowed authorities to “bypass” the system of EU and national merger control thresholds and enabled the Commission to investigate transactions that fell below these thresholds. Previously, below-threshold transactions had escaped any merger review, which was seen as a major loophole, allowing large multinationals to acquire nascent competitors with no substantial turnover (e.g., innovative startup companies) in deals that nonetheless had significant market impact (so-called killer acquisitions).

The 2024 Illumina/Grail judgment clarified that an Article 22 merger referral could only be made if the referring national competition authority had jurisdiction to review the transaction or in cases where there was no national merger control regime at all. This was clearly a blow to the Commission and Member State authorities, who had considered Article 22 referrals an effective mechanism to review killer acquisitions.

II. The Illumination of EU Merger Referrals

One year on, it is fair to say that Article 22 EUMR has survived the Illumina/Grail ruling to some extent.

Following the ruling, both the Commission and various Member State authorities swiftly published statements emphasizing their commitment to continuing referrals and investigating below-threshold mergers using all available tools. Some Member States amended or announced their intention to amend their national merger laws. These amendments would allow their competition authorities to either call in certain mergers or introduce alternative thresholds, such as market share or other non-monetary criteria.

There have also been two recent EU General Court cases that consider Article 22 of the EUMR post Illumina/Grail:

In the case of Brasserie Nationale/Boissons Heintz, the Luxembourg Competition Authority (LCA) referred a merger between Luxembourg’s largest brewery, Brasserie Nationale, and one of its main competitors, Boissons Heintz, to the Commission. Due to the significance of imports to Luxembourg in this sector, the Commission determined that trade between Member States could be impacted and accepted the referral in March 2024. Brasserie Nationale challenged the referral before the EU General Court, arguing that the referral request had been made after the expiry of the 15-day referral period. However, in July of this year, the General Court confirmed the Commission’s decision to accept the referral, stating that the initial information received from the parties had been insufficient for the LCA to assess whether the conditions for a referral were present. The Commission ultimately cleared the transaction, subject to divestment conditions. Although this case involves a referral from a Member State that does not yet have a merger control regime, and therefore undoubtedly falls within the scope of Article 22, the General Court's judgment still provides helpful clarifications regarding its application.

The Nvidia/Run:ai case is perhaps even more interesting in terms of the scope of Article 22, especially in the aftermath of the Illumina/Grail judgment. Following a referral from the Italian competition authority, the Commission investigated Nvidia’s acquisition of the AI startup company Run:ai. Given Run:ai’s low turnover, the transaction did not meet Italy’s regular turnover thresholds. Therefore, Italy used its new call-in powers, which allow it to require parties to notify the competition authority of their deal if the authority believes that the merger may harm competition and if the parties meet at least one of three turnover-related criteria. Although the Commission cleared the transaction unconditionally, Nvidia appealed the referral to the General Court in January 2025. Nvidia argued that the referral was not in line with the CJEU's Illumina/Grail decision.

Although the call-in powers used to catch the Nvidia/Run:ai transaction are relatively new, Italy is not the only country to have adopted such powers in recent years. These new regimes can be broadly divided into two groups: (1) regimes that allow for a call-in if certain lower-than-regular turnover or market share thresholds are met, and (2) regimes that allow for a call-in without any further requirements (other than a possible or likely impact on competition). Italy, along with Denmark, Hungary, Sweden, Latvia, and Slovenia, belongs to group 1, while Ireland, Lithuania, and Cyprus belong to group 2.

Additionally, the competition authorities in Belgium, the Netherlands, and France have begun advocating for call-in powers, even though their competition laws have not yet been amended.

Other approaches include introducing market share thresholds, similar to those in the Spanish and Portuguese merger control regimes, or transaction-value thresholds, like those in Germany and Austria. However, it is clear that call-in powers are currently favored because they give authorities the greatest flexibility to decide whether to exercise jurisdiction over a transaction.

III. No Merger Control, no problem – At least for the Enforcer 

However, the absence of flexible merger control rules does not mean that competition authorities leave transactions unscrutinized. Instead, they apply existing tools somewhat more creatively and sometimes interpret case law rather broadly.

For instance, the Belgian Competition Authority (BCA) applied the CJEU’s Towercast ruling in two recent cases. According to that ruling, a transaction that is not required to be reported to either the Commission or the competition authority of a Member State may still be investigated by the latter under Article 102 of the Treaty on the Functioning of the European Union (TFEU), which prohibits the abuse of a dominant position.

The first case relates to Proximus/EDPNet: Proximus, the incumbent Belgian telecommunications network operator, sought to acquire EDPNet, a broadband communications services provider. The transaction was not notifiable. In March 2023, the BCA opened an ex officio investigation into the acquisition as potential abuse of dominance, explicitly referencing the Towercast ruling. In June 2023, the BCA imposed interim measures on Proximus to ensure the viability and separation of EDPNet during the investigation, for a maximum period of 15 months. However, the transaction was abandoned in November 2023, so the BCA ended its investigation.

In early 2025, the BCA announced another ex officio investigation, again referencing Towercast but this time applying Article 101 TFEU, which prohibits restrictive agreements between companies. According to the BCA, Towercast should also be applicable in Article 101 scenarios and the proposed acquisition of Ceres's artisanal flour business by its main competitor, Dossche Mills, could be characterized as an anticompetitive agreement. The parties eventually abandoned the transaction, causing the BCA to end the investigation.

The French Competition Authority (FCA) has also referenced the Towercast judgment in a recent investigation. In May 2024, the FCA closed an investigation into several asset swaps that had taken place in 2015 between three French meat-cutting companies. Although none of these asset swaps was subject to notification under EU or French merger control rules, the FCA investigated whether the mergers constituted a restrictive practice under Article 101 TFEU. Ultimately, the case was dismissed due to a lack of evidence that the companies had used the exchanged information for purposes other than preparing the transactions.

Applying Article 101 TFEU to below-threshold mergers significantly expands the powers of competition authorities to review and potentially block those mergers. Indeed, Article 101 TFEU enables competition authorities to intervene even if the acquirer does not hold a dominant market position pre-merger as illustrated by the BCA’s investigation into the Dossche/Ceres merger. According to the BCA decision, Dossche held a market share of 25-35% on the relevant market before the proposed transaction, which is clearly below the dominance threshold. The application of Article 101 TFEU to below-threshold mergers will likely be tested in court. 

IV. Acqui-hires – The new Killer Acquisitions?

Recently, companies—particularly Big Tech companies—have started hiring a target company’s personnel rather than acquiring the company itself. This development is referred to as “acqui-hire” and is primarily seen in the AI space, where a company's value largely consists of its expert staff. For instance, Microsoft acquired nearly the entire Inflection AI team, including two of its founders, in a deal that also involved licensing Inflection AI’s technology. The Commission initially investigated the acquisition and determined that, since it involved all the assets necessary for Microsoft to take over Inflection’s market position in generative AI foundation models and AI chatbots, it constituted a concentration under EU merger control rules and was thus within the scope of those rules. However, the EU merger control thresholds were not met, and, following the judgment in Illumina/Grail, the Member States that had referred the case to the Commission withdrew their referrals, as the transaction was also not notifiable in any of them. This forced the Commission to end its investigation. Subsequently, the German competition authority picked up the case and concluded that the transaction resulted in a de facto acquisition of Inflection under German law. However, the authority also had to discontinue its investigation because the transaction did not meet the German turnover or transaction-value thresholds, which require substantial domestic operations of the target.

Other examples of acqui-hires involve Google, Amazon and Meta. Google made a deal with Character.AI. Through this deal, Google received a non-exclusive license to Character.AI’s large language model technology, and a brough a team of Character.AI employees, including two of its founders, on board. Similarly, Amazon hired most of Adept's team, another AI startup, while Meta invested USD 14.3 billion in Scale AI and hired its founder. None of these transactions were investigated under EU merger control rules.

V. Implications

All of the developments described above increase the complexity and uncertainty for companies when assessing the regulatory implications of M&A transactions. The days of straightforward merger filing analyses are over. Today, failing to meet the relevant monetary thresholds does not automatically mean that a transaction will escape antitrust scrutiny. Instead, companies must carefully assess the impact of their transaction on the market on a country-by-country basis to anticipate whether a competition authority might take a closer look, either by exercising call-in powers or initiating an ex officio antitrust investigation. This requires an in-depth analysis that takes into account the relevant sectors in which the parties are active and their market positions and examines whether the transaction may lead to disgruntled competitors or other market participants who may bring it to an authority's attention.

Furthermore, although competition authorities may currently be struggling to review acqui-hire transactions, it is likely that someone will soon test the legal waters by applying call-in powers or the Towercast judgment to such a scenario.

We would like to thank Cyriel Danneels for his contribution to this alert.