1. Introduction

Merger control is the system of rules and procedures used to prevent corporate mergers and acquisitions that could significantly reduce competition in a market. In the European Union (‘EU’), this task is primarily carried out by the Commission under the Merger Regulation (No 139/2004)¹. Its role is to assess whether a proposed merger would harm competition, for example by creating or strengthening a dominant market position, and to block or impose conditions on deals that could lead to higher prices, lower quality or less innovation for consumers.

In recent years, concerns have frequently been raised that merger control in the EU suffers from underenforcement. The turnover thresholds defining the Commission’s jurisdiction to review concentrations are not broad enough to capture all transactions that may negatively affect competition, particularly in strategic sectors where turnover may not accurately reflect the companies’ market position (such as pharmaceuticals and digital technologies).

To address this enforcement gap, the Commission has sought ways to expand the scope of its jurisdiction under the Merger Regulation. Instead of revising the turnover thresholds set out in Article 1, the Commission in 2021 adopted a new interpretation of Article 22 on referrals from Member States. This allowed it to review concentrations that fell short of the EU and national turnover thresholds but were referred to it by national authorities.

However, in its judgment of 3 September 2024 in the Illumina case (Cases C-611/22 P and C-625/22 P), the Court of Justice (‘CJ’) ruled that this approach was unlawful. As a result, the Commission must once again seek alternative solutions to extend the reach of its jurisdiction.

2. The limits of turnover-based jurisdiction

The Commission’s jurisdiction to review concentrations, like that of most other merger control authorities, is determined by turnover thresholds. These are set out in Article 1 of the Merger Regulation².

If the companies involved in a merger or acquisition meet the thresholds under Article 1, the transaction cannot be implemented without the Commission’s approval (Article 7 of the Merger Regulation). Importantly, no additional merger filings are required within the EU, since all national competition authorities are prohibited from applying their domestic merger control rules to concentrations that fall under the Commission’s jurisdiction³. By contrast, if a transaction falls short of the turnover thresholds in the Merger Regulation, the Commission, in principle, lacks the power to intervene⁴.

The turnover-based test has clear advantages. It is highly predictable and easy to apply: companies only need to review the geographic breakdown of their sales from the previous year to know whether they require the Commission’s merger approval. If so, they can also be assured that no further merger control obligations will arise under EU national laws.

However, the test also has shortcomings. Turnover is a rough proxy for identifying transactions that may affect competition, since company size does not always correspond to market power, and vice versa. As a result, the test produces both false positives and false negatives.

False positives occur when transactions meet the turnover thresholds even though they pose no real threat to competition. For example, merging companies may have high turnover but operate in entirely different markets, so their combination does not reduce competition.⁵ To address this issue, the Commission introduced a simplified, fast-track procedure that allows unproblematic deals to be cleared in as little as 15 working days after notification (see OJ C 160, 5 May 2023).

False negatives, by contrast, occur when potentially problematic transactions fall below the applicable turnover thresholds and thus evade regulatory scrutiny. This situation arises when a company’s limited turnover fails to reflect its actual competitive significance. The phenomenon is particularly evident in so-called “killer acquisitions,” in which an incumbent firm acquires a smaller but innovative rival primarily to neutralize an emerging competitive threat.

For instance, in the digital sector, start-ups often rely on venture capital rather than revenues during their early years, developing significant assets – particularly intellectual property rights in new technologies –before monetizing their services. A recent example is the AI start-up Inflection, which raised more than $1.5 billion to develop its artificial intelligence chatbot. Despite reaching one million users, the product was offered free of charge and generated minimal revenues. Consequently, when Microsoft acquired Inflection and redirected its resources to its competing chatbot, Copilot, the deal did not meet the EU or any Member State’s turnover thresholds.

Similar concerns arise in the pharmaceutical sector. A company developing a promising new drug may appear small in turnover terms if it has not yet obtained marketing authorizations. This situation was exemplified by Spark Therapeutics, a U.S. biotechnology company specializing in gene therapies. At the time of its USD 4.3 billion acquisition by Roche in 2019, Spark generated no significant revenue from marketed products, as its therapies were still in clinical development. Its limited commercial footprint therefore placed the transaction outside the EU’s turnover thresholds for merger review⁶.

However, the issue of false negatives extends beyond the context of so-called “killer acquisitions.” A similar problem can also be observed in the less high-profile, yet economically significant, sector of staple consumer goods. These products are often manufactured and distributed across Europe by a small number of multinational firms, yet their unit prices remain very low. As a consequence, merging parties may hold substantial market shares without generating high turnover, with the result that their transaction may likewise fall below the Merger Regulation’s notification thresholds⁷.

In recent years, the Commission has sought ways to assert jurisdiction over certain problematic transactions that fall below the turnover thresholds established in Article 1 of the Merger Regulation. The solution it ultimately adopted, however, did not involve revising or reinterpreting Article 1. Instead, the Commission relied on another, seemingly unrelated, provision located near the end of the Regulation: Article 22, which governs the referral of cases from Member States to the Commission.

3. The Commission’s “New Approach” on Article 22 and its application to the Illumina/GRAIL transaction

Article 22 empowers Member States to refer to the Commission transactions that do not meet the Merger Regulation turnover thresholds but could significantly impede effective competition within their territory and affect trade between Member States.

This has long been a “shapeshifting” provision. Over the decades, it has been revised and reinterpreted to pursue different objectives. Although its text does not expressly state this, the provision was originally intended to allow Member States lacking a national merger control regime to rely on the Commission to examine potentially harmful concentrations on their behalf. In the years that followed, however, most Member States established their own national merger control systems. This development produced the opposite difficulty: rather than an absence of review, now multiple national authorities often required parallel approvals for the same transaction. To remedy this, Article 22 was amended in 1997 to permit referral of cases from several Member States to the Commission, with more detailed procedural rules subsequently introduced in the 2004 Merger Regulation. During this second phase, Article 22 functioned as a mechanism to replace multiple national procedures with a single procedure before the Commission – the so-called “one-stop-shop” principle – thereby promoting procedural efficiency and reducing the regulatory burden on businesses. Given that the objective at that time was to limit the proliferation of national investigations, the Commission discouraged referral requests from Member States that lacked jurisdiction over a transaction under their own merger control regimes.

Finally, in 2021, the Commission once again repurposed Article 22 as a tool to tackle underenforcement. To extend its jurisdiction to transactions below the turnover thresholds, the Commission made a U-turn on its past interpretative practice. Instead of discouraging Member States from referring transactions over which they had no domestic jurisdiction, the Commission now explicitly encouraged such referrals. On the basis of this new interpretation, concentrations could be referred to the Commission regardless of whether either the Member State or the Commission itself had jurisdiction under their respective turnover thresholds. Crucially, this reform was ushered in without any formal amendment to the text of Article 22, only a change in its interpretation (in OJ C 113, 31 March 2021).

The acquisition of GRAIL by Illumina marked the first test of the Commission’s new approach to Article 22. GRAIL, a U.S. healthcare company, was developing blood-based cancer detection tests using genomic sequencing and data-science technologies. The significant market potential of these products was reflected in Illumina’s USD 7.1 billion offer. Nevertheless, because GRAIL’s products were still in development, the company generated no turnover – either within the EU or elsewhere – and the transaction therefore did not meet the turnover thresholds laid down in Article 1 of the Merger Regulation. As a result, the deal fell squarely within the category of “false negatives” that the Commission’s new approach sought to address⁸.

In line with the Commission’s new approach to Article 22, France in March 2021 requested a referral of the transaction to the Commission, subsequently joined by Belgium, Greece, Iceland, the Netherlands, and Norway. The Commission accepted the referral requests, noting that Member States could ask it to examine under Article 22 any concentration “over which they do not have jurisdiction” (Illumina, para. 27).

Illumina soon challenged this decision before the General Court (‘GC’). In its July 2022 judgment, the GC upheld the Commission’s interpretation (Case T-227/21). It reasoned that Article 22 could function as a “corrective mechanism” to the turnover-based thresholds in Article 1, allowing the Commission to intervene in cases that might otherwise escape review, provided a Member State requested it. Illumina appealed the judgment before the CJ in September 2022.

But in the meantime, the Commission pressed ahead. After adopting interim measures to against Illumina’s acquisition of GRAIL⁹, it prohibited the transaction in September 2022 and, in October 2023, ordered Illumina to divest GRAIL. In August 2023, the Commission also accepted referrals of two further transactions under the new Article 22 approach: Qualcomm/Autotalks and EEX/Nasdaq Power.

Just as businesses and practitioners were bracing for a new era of enforcement powers under a reinterpreted Article 22, Advocate General Emiliou poured cold water on the Commission’s strategy. In his Opinion delivered on 21 March 2024, he issued a strong rebuttal of the General Court’s judgment, arguing that Article 22 is intended to allocate jurisdiction between Member States and the Commission, not to create new powers over transactions falling outside both levels of jurisdiction. He found no legal basis in the legislative history of the Merger Regulation to support the Commission’s position and concluded that its approach exceeded the EU legislature’s original intent. The final word thus rested with the Grand Chamber of the Court of Justice, marking the last chapter in this judicial saga.

4. The Court of Justice’s reasoning: does Article 22 reach below national thresholds?

In interpreting Article 22 of the Merger Regulation, the CJ employed its familiar toolkit of literal, historical, contextual, and teleological interpretation.

On the literal meaning of Article 22, the Court agreed with the GC that the expression “any concentration” in the first sentence¹⁰ indicated that the provision could apply irrespective of the existence or scope of national merger control rules. This supported the Commission’s view that Member States lacking jurisdiction over a transaction under existing domestic laws could nonetheless refer it to the Commission under Article 22 (paras. 122-125). Yet, despite this “allegedly clear wording,” both courts considered it necessary to proceed to “contextual and teleological interpretation, as informed by the legislative history of that provision” in order to clarify its scope (para. 128).

At this point, however, the paths of the two courts diverged. The CJ overturned the various elements of the GC’s reasoning, not because they were entirely inaccurate or immaterial, but because they failed to provide a definitive answer to the central question at issue: can a Member State that lacks jurisdiction under its own existing merger control rules nevertheless refer a transaction to the Commission under Article 22?

With respect to legislative history, the GC was correct in observing that Article 22 was not intended to exclude referrals from Member States with their own merger control regimes, and that the provision’s objectives had evolved over time. But none of these developments – nor any legislative documents surrounding the adoption of the Merger Regulation – shed light on whether transactions falling short of national thresholds may nonetheless be referred (paras. 136-143).

As to contextual interpretation, the GC was again right that the thresholds in Article 1 apply “without prejudice to … Article 22”¹¹. But this offers no guidance as to which transactions are eligible for referral under the latter. It is equally true that Article 22 applies to cases in which “no notification is required”¹², but this does not necessarily encompass transactions falling short of national thresholds: it may instead refer to referrals from countries lacking a merger control regime, or with a voluntary regime where no binding filing obligation exists. Similarly, the GC was correct in observing that Article 22’s prohibition on a referring Member State applying its own competition law to a referred transaction does not, in itself, imply that national thresholds must be met. Yet this observation does not support the opposite conclusion either (paras. 162-165, 170, 215 and 216).

On teleological interpretation, the CJ dismissed the GC’s reliance on Recital 11 of the Merger Regulation, which describes referral provisions as “an effective corrective mechanism.” For the Court, this Recital refers only to the one-stop-shop principle designed to avoid multiple parallel national notifications, not to referrals of transactions below national thresholds. Nor were the references to “all concentrations” in Recitals 6 and 24 of any greater use: those expressions are confined to concentrations with a European dimension – that is, those meeting the turnover thresholds in Article 1 (paras. 192-195).

What, then, remains after the Court of Justice’s systematic dismantling of the General Court’s arguments? If there are no conclusive grounds for accepting referrals of transactions falling below national thresholds, are there any conclusive grounds for rejecting them? The Court advanced four main arguments – none of which, however, appears entirely immune from criticism.

First, the CJ emphasised that transactions referred under Article 22 – unlike those referred under Article 4(5) – are not deemed to have a European dimension. Consequently, the Commission’s jurisdiction under Article 22 extends only to the territory of the referring Member States, not to the EU as a whole. According to the Court, this means that the Commission’s review under Article 22 “replaces” that of the national authority, which in turn presupposes that the referring Member State must itself be competent under its own merger control rules (paras. 177-180). Yet this interpretation finds no support in the text of the Regulation: only Article 4(5) – and notably not Article 22 – requires that a referred transaction be “capable of being reviewed under the national competition laws.” Moreover, the Court’s reasoning fails to account for situations in which referrals originate from Member States that have no merger control regime at all. In such cases, there is no national competence to “replace,” yet the applicability of Article 22 remains undisputed.

Second, the Court observed that Article 1(4) and (5) of the Merger Regulation establish a specific legislative procedure for revising the turnover thresholds. Consequently, the principle of institutional balance dictates that it is for the EU legislature, rather than the Commission, to reassess those thresholds or to introduce any corresponding safeguard mechanism (paras. 183-184 and 216). This argument, however, merely assumes – without demonstrating – that Article 22, which was likewise enacted by the legislature, cannot extend to transactions falling below national thresholds¹³.

Third, the Court noted that Recital 15 of the Merger Regulation only makes sense if the referring Member State already has jurisdiction over a transaction under its own laws. This part of the judgment warrants closer attention, as Recital 15 provides the sole textual anchor in the Merger Regulation for the Court’s conclusion. Recital 15 states that “Other Member States which are also competent to review the concentration should be able to join the request [for referral]” (emphasis added). The fact that joining Member States must “also” be competent implies that someone else is too. For the Court, this can only be the Member State initiating the referral. This wording would therefore suggest – albeit indirectly and implicitly – that a referral cannot be made by a Member State lacking jurisdiction in the first place (para. 198). However, the Court’s reasoning again fails to account for referrals made by Member States without merger control regimes, which are never competent to review a transaction and yet are nonetheless entitled to submit Article 22 requests¹⁴.

Fourth, in the final part of its reasoning, the CJ finds that the Commission’s new interpretation of Article 22 jeopardised the effectiveness, predictability, and legal certainty owed to the parties to a concentration. In particular, the Court echoed Advocate General Emiliou’s concern that, in order to avoid an ex post review of a transaction, companies might feel compelled to submit informal notifications to all national authorities so as to trigger the 15-day deadline precluding referrals under Article 22¹⁵. Yet, only a few paragraphs later, the Court appeared far less concerned about potential risks to legal certainty when it suggested that ex post reviews of transactions could be initiated by any national authority under Article 102 TFEU. Such an outcome would, in fact, present a considerably greater threat to legal certainty: ex post scrutiny would become fragmented across multiple jurisdictions – rather than centralised under the Commission – and would not even be subject to a deadline that the parties could activate through the submission of an informal notification.

Are these arguments sufficient to conclude that the GC erred in law when it upheld the Commission’s interpretation of Article 22? As Professor Lindeboom has observed, the Court of Justice’s reasoning is neither more nor less convincing than that of the GC. The literal, historical, contextual, and teleological dissection of Article 22 could plausibly have led to either conclusion. What appears to have guided the Court instead is a broader, yet compelling, intuition: that a change of such magnitude in the application of Article 22 – extending the Commission’s jurisdiction to virtually any transaction a Member State might choose to refer – could not legitimately rest solely on a mere “change of mind” in the Commission’s interpretation of the provision¹⁶.

5. Procedural aspects: intervention and admissibility of evidence

The Illumina case also gave both the GC and the CJ the opportunity to rule on the right to intervene before them – though with diverging and even conflicting outcomes. Their orders highlight the courts’ different approaches to interpreting the rights of third parties (particularly business associations) in judicial proceedings, as well as the purpose of the intervention mechanism more generally.

The courts decided on applications by business associations seeking to intervene in support of Illumina’s actions. Business associations, like any other natural or legal person, may intervene in cases before the GC and the CJ if they “can establish an interest in the result of a case submitted” to either court. Case law has interpreted the concept of “interest in the result of the case” as an interest in the operative part of the judgment – that is, the actual ruling on the forms of order sought – rather than in the pleas in law or arguments within the reasoning (Case C-220/21 P(I), para. 18). In other words, an applicant must demonstrate a direct interest in the annulment of the specific act being challenged, one that would result in a change to its legal position (Case C-220/21 P(I), para. 19). According to the GC (though, as we will see, not according to the Court of Justice), a mere interest in the subject matter of the case – for example, because of its similarity to other cases involving the prospective intervener – is not sufficient to establish a right to intervene (Case T-691/18, paras. 19-21).

That said, the courts have developed more flexible case law allowing for broader admissibility of interventions by business associations. Specifically, associations may intervene if three conditions are met: the association represents a significant number of undertakings active in the sector concerned, its objects include the protection of the interests of its members, and the case may rase questions of principle affecting the functioning of the sector concerned and the interests of its members may therefore be affected significantly by the forthcoming judgment (Case T-892/16, para. 12).

Applying these principles, the GC dismissed the application of the Computer & Communications Industry Association (CCIA) to intervene in support of Illumina’s challenge to the Commission’s decision accepting the Article 22 referral. The CCIA contended that it had a legitimate interest in the proceedings, asserting that the digital sector – where its members operate – had been specifically identified by the Commission, alongside the pharmaceutical sector, as a field in which Article 22 referrals would be particularly appropriate (Case T-227/21, paras. 26-29). The GC was not persuaded. It held that CCIA had only an indirect and hypothetical interest in the pleas or arguments raised in this case, based on potential similarities with future referral cases that might affect companies in the digital sector. Neither CCIA nor its members had a direct interest in the annulment of the specific decision addressed to Illumina (Case T-227/21, paras. 31-35). The GC also placed particular emphasis on the fact that the case at issue concerned the pharmaceutical sector, while CCIA is active in the digital sector. However, the relevance of this distinction is debatable: the findings in the GC’s judgment are completely unrelated to the pharmaceutical operations of the involved companies. The same findings would apply in just the same way to companies active in the digital sector, which was explicitly targeted alongside pharmaceuticals in the Commission’s guidance on the new interpretation of Article 22¹⁷.

The CJ adopted a different, more liberal approach when another association, Biocom, sought leave to intervene in the appeal proceedings. Unlike the GC, the CJ found that Biocom had a direct and existing interest in the case in light of the impact the proceedings would have on the way in which certain concentrations in the life science sector, where Biocom’s members are active, are likely to be handled under Article 22 in the future (Illumina, paras. 16-17). In doing so, the CJ did not uphold (in fact it implicitly overruled) the GC’s position that similarity between the case at issue and other future cases involving the association’s members was insufficient to justify intervention.

This divergence between the two courts resurfaced in Illumina’s separate challenge to the Commission’s prohibition decision. That case did not concern the procedural issue of Article 22 referrals but instead the substantive competitive assessment of Illumina’s acquisition of GRAIL. Biocom applied for leave to intervene before the General Court, arguing that the case raised two substantive questions liable to affect the functioning of the life sciences sector in which it operates – namely the principle of territoriality and the standard of proof for efficiencies. The GC rejected this application, reasoning that both concepts apply equally across all economic sectors and were not specific to life sciences. Biocom therefore lacked a direct and existing interest sufficient to justify intervention (Case T-709/22, paras. 33 and 52).

The CJ overturned this decision on appeal. It took the view that the case concerned not the general application of the principle of territoriality, but its specific application in cases referred under Article 22. Since Biocom was entitled to intervene in cases concerning Article 22 (as established in the earlier order), the same conclusion was warranted here (Case C-523/23 P(I), paras. 34-38). In other words, under this expansive interpretation, Biocom was entitled not only to intervene in cases relating to the Article 22 referral mechanism itself, but also in cases involving substantive issues decided under that mechanism, such as the principle of territoriality.

These divergent outcomes may reflect the courts’ different understandings of the role of associations in judicial proceedings and of the objectives of the intervention mechanism. The GC appears to view intervention of business associations mainly as a tool of procedural efficiency, intended to avoid the proliferation of applications by individual association members. That objective of procedural efficiency would be undermined if associations with only indirect or hypothetical interests were allowed to intervene (Case T-227/21, para. 41). By contrast, the CJ places greater emphasis on the intervention’s function of assisting the court’s assessment of a case’s broader context, by offering the association’s overall perspective on the collective interests it represents (Illumina, para. 9).

The judgment of the CJ also provides an instructive clarification on the admissibility of new evidence in appeal proceedings. The Court reaffirmed its settled case law that, on appeal, its jurisdiction is limited to reviewing questions of law and does not extend to reassessing the GC’s findings of fact. (This limitation significantly constrains the scope of appellate review, an obstacle that appellants often seek to overcome through inventive legal arguments.) Ordinarily, this principle precludes the submission of new evidence at the appeal stage, since evidence is typically adduced to establish factual matters. However, the Court recognised an important exception in circumstances where the evidence pertains to questions of law. In particular, when the interpretation of EU law requires consideration of its historical context, the Court must be able to examine materials illuminating the legislative history of the relevant provisions. Accordingly, the CJ held that the travaux préparatoires relating to Article 22 could not be regarded as “purely factual matters” and were therefore admissible on appeal, notwithstanding that GRAIL had not produced them before the GC. Moreover, in line with the general principle iura novit curia, the Court observed that it would have been entitled to consider such documents of its own motion (Illumina, paras. 132-134).

6. What ways forward after Illumina?

The Court of Justice’s judgment in Illumina effectively brought an end to the Commission’s effort to address perceived “false negatives” in EU merger control through its revised interpretation of Article 22. In the aftermath of the ruling, the Commission withdrew all decisions adopted on the basis of that interpretation, including its prohibition of the Illumina/GRAIL transaction. Similarly, the referral requests submitted by seven Member States in the Microsoft/Inflection case were withdrawn.

At the same time, the closing paragraphs of the judgment appeared to suggest possible paths forward, indicating that alternative mechanisms could be explored to address the limitations inherent in turnover-based jurisdictional thresholds.

6.1. Reform of the Merger Regulation

As a first option, the CJ noted that the thresholds in the Merger Regulation are always open to revision (Illumina, para. 216). The Regulation itself provides a specific legal basis for their amendment by qualified majority vote. Lowering those thresholds would enable the Commission to capture a wider set of transactions.

Moreover, although turnover remains the most commonly used jurisdictional criterion, it is not the only available approach. Some Member States, such as Spain and Portugal, rely on market-share thresholds as an alternative. Market share is generally a more direct indicator of market power than turnover; however, this method complicates the filing process for undertakings, as determining market share requires defining the relevant product and geographic markets – an inherently complex and frequently contested task.

Another possible approach is to base filing obligations on the transaction’s value. This criterion rests on the assumption that the price an acquirer is prepared to pay more accurately reflects the competitive significance of a target than turnover figures alone, particularly in innovation-driven sectors such as digital and pharmaceuticals. Germany and Austria introduced such thresholds in 2017, set at €400 million and €200 million respectively. Nonetheless, this model carries the risk of producing false positives, potentially capturing a large number of transactions that do not raise substantive competition concerns. National courts have therefore sought to narrow the application of value-based thresholds, limiting their use to cases in which turnover is clearly an inadequate measure of a company’s actual market position¹⁸.

A more radical option would be to dispense with binding jurisdictional thresholds altogether, replacing them with a voluntary filing system. Under that regime, companies would face no mandatory filing obligations but competition authorities would retain the power to “call in” transactions for review on a case-by-case basis, as in the United Kingdom. The CMA’s enforcement record in cases such as Meta/Giphy and Roche/Spark illustrates how this tool can be used effectively to capture digital and biotech deals that would otherwise escape scrutiny.

Finally, rather than amending the turnover thresholds set out in Article 1, it would be possible to rephrase Article 22 to make it (even more) clear that “any transaction” indeed means “any transaction,” and that all concentrations may therefore be referred irrespective of whether the Member State has jurisdiction under its domestic laws¹⁹. However, Article 22 cannot be amended through the special accelerated procedure provided for in Article 1(5). Any such amendment would need to be adopted on the general legal basis of the Merger Regulation itself – namely, a combination of Article 103 and Article 352 TFEU. The latter would require unanimous approval by the Council. Given the unanimity requirement among Member States, obtaining agreement on such a proposal may prove particularly difficult.

6.2. Call-in powers at the national level

As a second avenue, the CJ pointed to the possibility of lowering filing thresholds at the national level (Illumina, para. 217). This would empower national competition authorities to review more cases themselves or to refer them to the Commission under Article 22. Provided that these authorities have jurisdiction under the lowered thresholds, these referrals would be consistent with the Illumina judgment.

Several Member States have already pursued this path by introducing “call-in” powers enabling their authorities to review transactions that fall outside notification requirements. While Lithuania, Norway, and Sweden have had such powers for over 20 years, nearly half of EU Member States have introduced them within the past four years²⁰. The Commission has actively encouraged this development, and it is estimated that by the end of 2025, two-thirds of Member States will have call-in powers in their merger control regimes²¹.

This approach will soon be tested before the EU courts in the Nvidia/Run:ai case. The Italian Competition Authority exercised its newly introduced call-in powers to review the transaction and subsequently referred it to the Commission under Article 22. Although the Commission ultimately cleared the merger unconditionally, Nvidia brought an action before the GC challenging the referral decision, arguing that Article 22 cannot be relied upon where the national authority’s competence is based solely on a call-in mechanism (Case T-15/25, still pending). Some commentators have contended that the use of such national call-in powers may be difficult to reconcile with the principle of legal certainty, as underscored by the CJ in the Illumina judgment. However, while that principle guides the interpretation of the Merger Regulation, there is nothing in EU law that prevents Member States from introducing domestic call-in powers if they so choose. Where such powers exist, the one-stop-shop principle underlying the EU merger control system would, in fact, support the admissibility of referrals under Article 22.

The increasing reliance on call-in powers at the national level may also increase the relevance of another referral mechanism under the EU Merger Regulation: Article 4(4). Like Article 22, this provision enables the referral of a concentration from national authorities to the Commission. However, unlike Article 22, an Article 4(4) referral is initiated by the notifying parties and requires that at least three Member States be competent to review the transaction. Even when a referral under Article 22 is no longer available – for instance, because the 15-day deadline for submitting such a request has expired – the parties may invoke Article 4(4) to avoid multiple national call-in proceedings and instead secure a single, centralised review at the European level. Compared with Article 22, this mechanism offers an additional advantage: once a transaction is referred under Article 4(4), it is deemed to have a “European dimension.” Consequently, the Commission’s jurisdiction extends to assessing the transaction’s effects on competition across the entire EEA, rather than being limited to the territories of the referring Member States.

6.3. Investigations for abuse of dominance

Finally, the CJ underscored that transactions falling below national notification thresholds may still be subject to scrutiny under Article 102 TFEU, as reaffirmed in the Towercast judgment (Case C-449/21).

That case concerned TDF’s acquisition of Itas, both active in the French market for digital terrestrial television broadcasting services. Towercast, the only remaining competitor, alleged that the transaction was anticompetitive. Because the deal did not meet either EU or French merger control thresholds, Towercast requested that the French Competition Authority examine it as an abuse of dominance under Article 102 TFEU. The authority rejected the complaint, reasoning that the Merger Regulation establishes an exclusive ex ante control system. Towercast appealed, prompting the French court to refer the question to the CJ for a preliminary ruling.

In its Towercast judgment, the Court held that national competition authorities are indeed competent to apply Article 102 TFEU to transactions that fall below EU or national notification thresholds. This conclusion rested on the principle of direct effect: as a provision of primary EU law, Article 102 has direct effect within Member States’ legal systems. Accordingly, secondary legislation – such as the Merger Regulation, which provides for ex ante merger control – cannot preclude the ex post application of Article 102 to transactions lying outside its scope.

The Towercast ruling therefore offers a possible avenue for addressing anticompetitive concentrations that escape ex ante merger control, without imposing additional notification requirements on undertakings. However, this mechanism presents several practical and legal limitations. First, review can occur only after completion of the transaction, making potential remedies highly disruptive for the undertakings involved. This also diminishes the companies’ incentives for cooperation during investigations, complicating evidence gathering. Second, ex post proceedings are generally not subject to strict deadlines, unlike ex ante review, leading to prolonged uncertainty and potential instability for businesses. Third, there is a heightened risk of regulatory fragmentation, as multiple authorities could initiate parallel and inconsistent enforcement actions concerning the same transaction. Finally, Article 102 does not capture all anticompetitive concentrations. It applies only where a transaction constitutes or reinforces a dominant position and enables the dominant firm to eliminate effective competition. It cannot address cases where competition is significantly impeded without the creation or strengthening of dominance. Moreover, even the mere strengthening of an existing dominant position may not suffice, since the Towercast judgment appears to require “that only undertakings whose behaviour depends on the dominant undertaking would remain in the market” – a stringent and somewhat ambiguous standard (para. 52).

7. Conclusion

The judgment of the CJ in Illumina marks a decisive turning point in the evolution of EU merger control. It constitutes a significant setback for the Commission, which had invested considerable political and institutional capital in its 2021 reinterpretation of Article 22 of the Merger Regulation. That “flagship” policy – intended to address perceived under-enforcement by capturing low-turnover transactions – has now been comprehensively rejected.

The judgment also delivers a further blow to the GC, whose reasoning in Illumina was overturned scarcely a year after the CJ had reversed its much-debated ruling in CK Telecoms (Case T-399/16; pourvoi Case C-376/20 P). In both instances, the CJ intervened to correct what it viewed as misapplications of legal principles governing merger control.

For businesses, however, the outcome is not the unequivocal relief it might initially appear to be. The Court grounded its reasoning in the principles of legal certainty and predictability, confining Article 22 referrals to national authorities that possess substantive competence under their own laws. Yet, in practice, this shift replaces the limited uncertainty created by a handful of targeted Article 22 referrals with a more fragmented and unpredictable regulatory landscape. In the wake of Illumina, Member States have begun introducing new call-in powers, each governed by distinct jurisdictional criteria, procedural frameworks, and time limits. At the same time, the prospect of ex post scrutiny under Article 102 TFEU, recognised in Towercast, now hangs over completed transactions. The result is a patchwork system in which the same concentration may face divergent treatment across Member States, with little assurance of consistency at EU level. For merging parties, this development increases rather than reduces regulatory risk: instead of a single, centralised review in Brussels, firms may now confront multiple, parallel investigations across Europe under different legal instruments.

Nor does the judgment represent an unqualified victory for Illumina itself. By the time the CJ annulled the referral decision, Illumina had already complied with the Commission’s order to divest GRAIL. In December 2023, the company announced the completion of the divestment, transferring GRAIL’s shares to a trust for an independent sale process. Ironically, Illumina’s pursuit of GRAIL ultimately failed, and the market structure the Commission had sought to preserve remains intact, irrespective of the Court’s ruling.

Taken together, these developments confirm that the problem of under-inclusive jurisdiction in EU merger control remains unresolved. While the judgment forecloses one avenue – the Commission’s reinterpretation of Article 22 – it leaves the underlying policy challenge untouched. If anything, it heightens pressure on national authorities to broaden their competences and on EU institutions to consider structural reform of the jurisdictional framework. The Illumina saga, far from providing closure, stands as an invitation to reconsider the very foundations of jurisdiction in EU merger control.

---

* The information and views set out in this annotation are those of the Author and do not necessarily reflect the official opinion of the European Commission.

¹ Council Regulation (EC) No 139/2004 of 20 January 2004 on the control of concentrations between undertakings (the EC Merger Regulation), in OJ L 24, 29 January 2004.

² The provision sets out two alternative sets of thresholds: the first based on the global and EU-wide annual turnover of the companies involved, and the second also taking into account turnover generated in each Member State. A transaction need only meet one of these thresholds to fall under the Commission’s jurisdiction.

³ See Article 21 of the Merger Regulation, and subject to referrals of transactions from the Commission to Member States, as provided in Articles 4(4) and 9 of the Merger Regulation.

⁴ Subject to referrals of transactions from the Member States to the Commission, as provided in Articles 4(5) and 22 of the Merger Regulation.

⁵ This situation typically arises when large private equity firms acquire a company operating in a market segment in which none of their existing portfolio companies are active.

⁶ S. Chisholm, M. Bennett, Roche/Spark and pharma mergers: acquiring a ‘pipeline’ potential competitor and ‘killer acquisition’ concerns, in CRA Competition Memo, 2020.

⁷ These transactions may meet the thresholds in at least three individual Member States, allowing the parties to request a referral to the Commission under Article 4(5) of the Merger Regulation. However, companies may choose not to pursue this route, since such a referral would give the Commission jurisdiction to assess the transaction’s impact across the entire European Union, including in countries where national thresholds are not met.

⁸ The transaction did not technically qualify as a “killer acquisition,” as GRAIL was not a direct competitor of Illumina. See A. Giraud and Others, Illumina/GRAIL: Inside the Perfect Storm, in Competition Policy International, 2024. The Commission’s concerns were therefore not based on a loss of competition between the parties. Instead, they focused on the risk that, following the acquisition, Illumina might engage in a strategy of foreclosing GRAIL’s competitors by restricting access to its technology.

⁹ Illumina had in the meantime proceeded to acquire a controlling stake in GRAIL, thereby openly disregarding the standstill obligation under Article 7 of the Merger Regulation.

¹⁰ “One or more Member States may request the Commission to examine any concentration as defined in Article 3 that does not have a Community dimension within the meaning of Article 1 but affects trade between Member States and threatens to significantly affect competition within the territory of the Member State or States making the request” (emphasis added).

¹¹ The first paragraph of Article 1 reads: “Without prejudice to Article 4(5) and Article 22, this Regulation shall apply to all concentrations with a Community dimension as defined in this Article” (emphasis added).

¹² “Such a [referral] request shall be made at most within 15 working days of the date on which the concentration was notified, or if no notification is required, otherwise made known to the Member State concerned” (emphasis added).

¹³ J. Lindeboom, Illumina/Grail: flawed originalism and the judicial hunch, in Journal of Antitrust Enforcement, 2025, p. 223 ff.

¹⁴ Ibidem.

¹⁵ “One or more Member States may request the Commission to examine any concentration … Such a request shall be made at most within 15 working days of the date on which the concentration was notified, or if no notification is required, otherwise made known to the Member State concerned” (emphasis added).

¹⁶ J. Lindeboom, Illumina/Grail, cit.

¹⁷ The GC quickly dismissed this point, noting that “the lawfulness of the 2021 guidance, to which the CCIA refers to justify the functioning of the digital sector being affected…is not the subject matter of the present case” (para. 39). Yet CCIA’s arguments do not appear to dispute – and in fact presuppose – that lawfulness of that guidance.

¹⁸ M. Sura, E. Wiese, F. von Schreitter, No value in the future? The twists and turns of Germany’s transaction value threshold, in Hogan Lovelss Insights and Analysis, 2025.

¹⁹ J. Bethan, Loriot: “Nothing has been decided” on potential EU deal value thresholds, in Global Competition Review, 2024.

²⁰ D. Connolly and others, Predictably Uncertain: Managing merger control call-in risk at local level in the EU, in Kluwer Competition Law Blog, 2025.

²¹ J. Bethan, EU “actively encouraging” Member States to adopt call-in powers, Guersent says, in Global Competition Review, 2024.