�ԹϺ�����

The acquisition of small, disruptive companies by big, established market players is a matter of significant concern for competition policy. This is particularly the case in the digital economy. Merger control has traditionally been enforced to predominantly tackle static concerns – the risk that a certain acquisition would remove a source of actual competition in the market ¹ Note that this does not necessarily need to be the case. Merger control at EU level, for example, envisages tools to deal with potential competition, and potential competition has played an important role in the analysis of several merger cases, for example Dow/DuPont (M.7932) or Novartis/GlaxoSmithKline (M.7275). Cases can be searched for at . . This emphasis on static concerns also tends to be reflected in the way merger control is designed. Merger notification rules, for example, require companies to flag their intention to pursue a merger to the competent authority only if the acquiring and the target companies are both established market players. In practice, that has generally meant placing much emphasis on companies’ current turnovers. 

Yet, turnover is not a good indicator of the potential of a company – especially tech or digital companies – to become a strong source of market competition in the future. A free smartphone application may generate no revenue today despite being downloaded by millions of users. The value of its business depends also on its growth potential, as indicated by the price that an established company may be willing to pay to acquire it. Limited revenue does not mean that it will not become a competitive threat to established applications. The business strategies of digital startups often aim to maximise adoption then monetise the product when the user base is big enough (Mariniello, 2022). When WhatsApp was acquired by Facebook in 2014, it had 600 million users but a turnover of only $10 million ² Jay Yarow, ‘WhatsApp, Facebook .

The tension between static merger control and dynamic markets is acute in economic sectors including digital and pharma, where the jump from potential to actual competition can be sudden and disruptive. ChatGPT, for example, in 2022 reached one million users in only five days ³ Katharina Buchholz, ‘Threads Shoots Past One Million User Mark at Lightning Speed’, Statista, 7 July 2023, . .

Under European Union rules, mergers and acquisitions are notifiable only if the turnover of the involved companies exceeds a certain threshold. EU countries, meanwhile, rely either on turnover thresholds or on a mix of solutions, including notification rules based on the price paid by the acquirer, the market share of the merged entity or an ad-hoc assessment by the competition authority of the potential significance of the merger (ie calling in acquirers to notify a merger if authorities think it might raise competition concerns). The European Commission, the EU’s competition enforcer, rarely has direct jurisdiction to vet acquisitions of small startups, but EU countries that have jurisdiction can refer mergers to the Commission. Consequently, occasionally, the Commission has scrutinised mergers that fell below the EU-wide radar.

This situation is far from ideal (addressing it is a priority for European Commission President Ursula von der Leyen ⁴ Javier Espinoza, ‘Brussels seeks powers to block ‘killer acquisitions’ in Europe and beyond’, Financial Times, 16 October 2024, . ). Some acquisitions might be not captured because the acquired company has no link to the national market where the acquisition of small-turnover companies must be notified. Sometimes, even if an EU country does have jurisdiction, it might opt not to refer the merger to the Commission for political reasons. A country might attempt to lure investment by signalling a more lenient merger policy, for example. The patchwork of notification systems and procedural uncertainty is not favourable to a healthy business environment.

Choosing the right policies to respond to these shortcomings is not straightforward. Doing more merger control entails administrative costs, for stakeholders and competition authorities. It also runs the risk of mistakenly blocking acquisitions that would foster, rather than reduce, competition or innovation, especially in digital markets, where the problem is most pressing.

The main EU-level tool to regulate digital markets, designate as ‘gatekeepers’ the largest digital firms and exercise control over their activities is the 2022 Digital Markets Act (DMA, Regulation (EU) 2022/1925). The DMA should be amended to empower the European Commission to scrutinise any acquisition by a gatekeeper ⁵ Gatekeepers are large, hard-to-avoid platforms that offer ‘core platform services’ such as internet search and social networks. Currently, the designated DMA gatekeepers are Microsoft, Google, ByteDance, Booking, Apple, Amazon and Meta. , whatever the levels of turnover involved. Moreover, the burden of proving that a merger is not harmful should be shifted to gatekeepers, with the goal of increasing the accuracy of the decision in a highly dynamic and uncertain environment by leveraging the knowledge of market players. 

The next section discusses the economic effects of acquisitions of small companies and why competition authorities should be concerned with them. Section 3 assesses the current system in the EU and its limitations. Finally, section 4 sets out fixes.

A relatively small group of largely digital companies has enjoyed spectacular growth in the last 20 to 25 years. Implementing an online platform business model, they have leveraged an ever-expanding user base to acquire and entrench dominant positions in key digital markets. Their dominance has rarely been challenged ⁶ Market definition – the definition of the boundaries of the markets in which such dominance is exerted – is not discussed in this paper. I assume that big-tech companies have strong quasi-monopolistic market power in the (correctly identified) relevant antitrust markets. .

Concentration has thus come to be considered an intrinsic feature of digital markets: concentration tendencies are explained well by the underlying economics. For this reason, market power does need not come from anti-competitive behaviour. Online platforms feature strong scale and network economies, leading to reinforcing feedback loops: the bigger they are, the better they get, and the more users they attract, the bigger they become (eg Varian et al, 2004).

There is little doubt, however, that digital markets would benefit from increased competition, even accounting for their features evoking quasi-natural monopoly dynamics ⁷ Natural monopolies are markets in which the supply of a product by multiple firms is less efficient than the supply of the same product by one firm. , and despite the suitability of any potential regulatory framework, such as the DMA, tailored to mitigate big tech’s exercise of market power. The position of the Trump administration ⁸ The White House, ‘Fact Sheet: President Donald J. Trump Issues Directive to Prevent the Unfair Exploitation of American Innovation’, 21 February 2025, . , aimed at shielding American companies from regulatory enforcement, makes it all the more necessary to address the issues structurally, protecting markets’ contestability.

Policymakers should not expect to tame big tech only through regulation. Rather, they should work proactively to establish the conditions for competition to flourish in digital markets. So far, however, credible competition has not emerged. Why? One factor relates to mergers and acquisitions. 

A challenge to an incumbent’s market position can materialise in two main ways: (1) an existing company develops or acquires the means to compete with the incumbent in that market, or (2) a new company supplying a product that can be considered a close substitute for the incumbent’s product acquires sufficient scale to challenge the incumbent on its turf. A combination of (1) and (2) is also possible, if a company operating in a market related to incumbent’s market buys the target to use its base or knowledge to produce a substitute for the incumbent’s product ⁹ Countering such a scenario partly motivated Facebook’s acquisition of WhatsApp and Instagram. Salvador Rodriguez, ‘As calls grow to split up Facebook, employees who were there for the Instagram acquisition explain why the deal happened’, CNBC, 24 September 2019, . . Mergers and acquisitions can be pro- or anti-competitive, depending on the market circumstances.

2.1 Pro-competitive acquisitions

In a number of cases, acquisitions of small companies could have pro-competitive and thus welfare-enhancing effects.

• The acquired company’s product increases the acquirer’s efficiency or helps innovate and improve the quality of its products. For example, acquiring an artificial intelligence start-up could help Google improve its organic search engine, ultimately benefitting Google’s users, to the extent that they would not have benefitted to an equal degree from the AI services of the purchased company. More generally, vertical or conglomerate mergers may make supply more efficient when the merged companies supply complementary products (Teece, 2020) ¹⁰ A qualitative analysis of several acquisitions by big tech (Crandal and Hazlett, 2022) concluded that merger outcomes were often pro-competitive, increasing the efficiency of the incumbent or its innovation potential (though suggesting that in some cases the outcome was not necessarily pro-competitive). . 

• The buyer is dominant in one market, but it acquires the target to challenge the dominance of another incumbent in another market. For example, Amazon might one day buy a promising European startup/social network to challenge Meta’s Instagram dominance in social networking ¹¹ For a discussion, see Bryan and Hovenkamp (2020). For a theoretical model, see Motta and Peitz (2020). Often the loss imposed on the incumbent is greater than the gain yielded to the entrant, when moving from a substantial monopoly to a duopoly. If that is the case, the incumbent will always have the incentive to outbid any potential acquirers willing to use the target to enter its markets. A telling example was Google outbidding Apple and Facebook when acquiring Waze (Motta and Peitz, 2020). . That explains why a thorough merger check is necessary: a merger may be blocked if performed by the incumbent to pre-empt future competition, but should be cleared if it is instead functional to enter other markets ¹² A potential example of a pro-competitive merger was Google’s acquisition of Android in 2005. At the time Android was only two years old. Thanks to Google’s acquisition, it grew to a fully-fledged competitor to Apple’s iOS in the mobile operative systems market (Crandal and Hazlett, 2022). .

  This scenario also suggests that competition authorities should not treat mergers differently based on the acquirer’s geographical origin. A foreign company could acquire a small domestic company to improve its ability to challenge a gatekeeper. This would increase value for domestic users and have positive effects on the domestic economy (in that respect, the European Commission’s plans to tackle foreign killer acquisitions, as laid out in European Commission President von der Leyen’s brief to Competition Commissioner Teresa Ribera ¹³ Ursula von der Leyen, ‘Mission letter to Teresa Ribera Rodríguez, Executive Vice-President-designate for a Clean, Just and Competitive Transition’, 17 September 2024, . , are worrying).

• The prospect of being acquired is a catalyst for innovation. Assume that the target would successfully challenge the incumbent’s market power once it scales up. If the target’s expected profits in a duopoly are lower than the additional profits the incumbent makes by keeping competition out of the market, then there is room for a win-win merger deal. The pie to share is bigger with the merger. That is why innovative efforts and venture-capital investment can increase with the prospect of acquisition ¹⁴ Similarly, entrepreneurs may create a startup to flag their ‘talent’ to big tech, hoping to be ‘acquihired’ once the startup has proved successful (Benkert et al, 2023; Bar-Isaac et al, 2024). Acquihires may be problematic in themselves, for example by granting the acquiring company monopsony power over specialised talent. . Rasmusen (1988) called this “entry for buyout”. Letina et al (2024) showed that a blanket merger ban would have overall weakly negative effects on innovation. An analysis of more than 30,000 VC deals and nearly 400 startup acquisitions by big tech found that acquisitions of European target companies on average increase by 9.5 percent the number of VC deals and by 34 percent the total amount of VC funding in the first quarter following the acquisition (Prado and Bauer, 2022). This effect is however short lived, as it fully dissipates in the following quarter (similar effects are identified for acquisitions targeting US startups).

  This dynamic is particularly relevant in the digital sector: more than 90 percent of VC-backed successful startups are acquired by other companies; in biotech or pharma, that falls to 50 percent of startups (Eisfeld, 2024). Note, however, that the more established an incumbent’s market power becomes, the lower its willingness to pay for new technologies (Bryan and Hovenkamp, 2020). Thus, in the long term, the relevance of entry for buyout should be expected to decrease, if competition does not increase in parallel.

2.2 Anti-competitive acquisitions

There are two necessary conditions for a merger to be considered anti-competitive and thus welfare-reducing (Motta and Peitz, 2020) ¹⁵ Technically speaking, a merger could lead to lower welfare levels compared to the counterfactual even if condition 2 does not apply, for example when an alternative buyer could have bought the target, leading to lower concentration levels but at least similar innovation/efficiency gains. In practice, however, it would be very difficult to legally deem a merger anticompetitive on that basis. :

1. The target company has the financial, managerial or marketing means to develop its product (including in the case of acquisition by another, less worrisome, acquirer) should the merger not take place; and 

2. The acquirer does not develop (or even discontinues) the target’s product after acquisition, or otherwise uses it for anti-competitive purposes. 

The first condition refers directly to the counterfactual, or what would happen if the merger did not take place. Of all acquisitions by DMA gatekeepers since 2000, only one has been prohibited by the European Commission (Booking/eTraveli, M.10615, in 2023) – thus the counterfactual condition cannot be analysed empirically. However, from a theoretical perspective, it can be assumed that a target’s product could have in some cases become an actual competitor to the acquirer. A rich literature specific to online platform markets shows that, in that case, any incumbent subject to this potential threat would have had an incentive to pursue the acquisition to preserve their market power (Rhodes and Zhou, 2019; Motta and Peitz, 2020; Prat and Valletti, 2022). Thus, insights from observed acquisitions cannot dispel the concern that some mergers have been anti-competitive.

The second condition has been discussed in the literature mostly in relation to the possibility of killer acquisitions, or acquisitions with the goal of eliminating the acquired company ¹⁶ Killer acquisitions were first identified in the pharmaceutical sector. Cunningham et al (2021) found that between 5.3 percent and 7.4 percent of acquisitions in the US pharmaceutical sector could be classified as killer acquisitions. . Gautier and Lamesch (2020) found that in 60 percent of 175 acquisitions by Facebook, Amazon, Microsoft and Google, the target firm’s brand was discontinued. Affeldt and Kesler (2021) found that of more than 50 GAFAM acquisitions ¹⁷ Acquisitions by Google, Amazon, Facebook, Apple and Microsoft.  involving mobile apps in the Google Play Store, half were discontinued, while the others tended to become free of charge while requiring more privacy-sensitive information from users. They also identified indirect effects on the competitive landscape: the acquired apps were 2.8 percentage points less likely to run an update after the merger. This suggests that acquisitions in the Google Play Store ecosystem had general negative effects on market quality and innovation.

Kamepalli et al (2020) noted that, when the entrant is likely to be acquired, venture capital incentives to invest in the entrant drop. This is because, if the target is expected to be acquired, potential adopters of the target’s technology are put off and thus investors expect decreasing profits. Kamepalli et al (2020) referred to “kill zones” around small companies that attract big tech and thus become less interesting to investors ¹⁸ In nine case studies, Kamepalli et al (2020) identified a drop in venture-capital investment in market segments more likely to be targeted by Facebook’s or Google’s acquisition strategies. .

Barsy and Gautier (2024) showed that, on average, acquisitions by big tech increase the innovation efforts of the acquirers but only temporarily. After 18 months on average, patenting decreases, especially if the acquired patent is in the same patenting category as the acquirer’s technology portfolio. Barsy and Gautier (2024) also noted that, meanwhile, the rest of the industry continues innovating, suggesting that if the acquirer discontinues its patenting efforts it is not because the technology has reached maturity.

Moreover, conglomerate mergers can entail the prospect of exclusivity behaviour. The incumbent may acquire a small target company that supplies an important complementary input in the value chain. The incumbent may anticipate that a credible challenger to its market power will one day require access to that input in order to succeed. Demoting the input early on defuses the risk of entry. In the US vs. Microsoft case ¹⁹ See . , Microsoft enacted a strategy to undermine Netscape because of concerns about Netscape’s potential to incentivise the development of middleware that could have threatened Windows’s dominance (Motta, 2023) ²⁰ For a discussion of conglomerate effects, see also Bourreau and De Streel (2019) and Motta and Peitz (2020). .

2.3 Historical analysis 

An examination of the evolution of digital markets in the last 20 years leads to three observations.

First, the number of acquisitions by big-tech firms, especially of promising start-ups, has been very high. Between 2000 and 2023, companies now designated as DMA gatekeepers successfully executed at least 683 acquisitions globally ²¹ This figure is taken from Bloomberg data, which does not contain mergers of very small size. This implies that the figure is conservative: gatekeepers may have executed more acquisitions than the those reported in the observed period. , with 67 acquisitions targeting an EU27 based company (Figure 1) ²² Note that, generally speaking, merger control can kick in also when all involved companies are non-EU based. What matters for merger control is to potential effects on the EU single market, ie whether at least one of the involved companies is active in Europe. . Microsoft made 187 acquisitions (21 in the EU), Alphabet made 211 (17), Apple 93 (9), Meta 82 (9), Amazon 81 (7), Booking 16 (4) and ByteDance 13 (0).

The median age of acquired targets between 2008 and 2018 was four years old for Google, 2.5 for Facebook (Meta) and 6.5 for Amazon (overall, 60 percent of the observed acquisitions involved a target no older than four) (Argentesi et al, 2020).

Figure 1: Global and European acquisitions by DMA gatekeepers since 2000

^{Source: Bruegel based on Bloomberg. Note: see footnote 21 on scope of the data.}

Second, the main digital market segments have become increasingly concentrated. Ederer and Pellegrino (2023) explored the evolution of “product centrality” in selected digital markets. Product centrality is an inverse function of markup and a direct function of product differentiation. When the markup is high, and product differentiation is low, product centrality is low. At the extreme, product centrality is zero and the company can set production output at the monopolistic level. Conversely, when product centrality is 1, the company takes prices as given. In other words, in the terminology used by Ederer and Pellegrino (2023), a ‘central’ product is at the core of market competition (ie it is subject to a high level of competition), while products at the periphery are insulated from the behaviour of other tech companies. Ederer and Pellegrino (2023) showed that Google, Apple, Meta and Amazon have, since 2001, increased their market power dramatically (as proxied by a huge drop in their product centrality within the technological sector). There is no reason to expect these trends to be different in Europe, as the European Commission analysis shows ²³ Under Article 3(1) of the DMA, the Commission designates an undertaking as a gatekeeper if it fulfils three cumulative requirements: (a) it has a significant impact on the internal market; (b) it provides a core platform service, which is an important gateway for business users to reach end users; and (c) it enjoys an entrenched and durable position in its operations, or it is foreseeable that it will enjoy such a position in the near future (for an explainer, see Anderson and Mariniello, 2021). Google, Apple, Meta and Amazon were designated gatekeepers accordingly, in line with the insights in Ederer and Pellegrino (2023). .

Figure 2: Centrality of big-tech companies’ products

^{Source: Ederer and Pellegrino (2023). Note: The centrality percentile reflects a firm's position within the market rivalry network. Low values indicate that a firm occupies a ‘peripheral’ position in the network, granting it monopolistic power due to the uniqueness of its products.}

Considered together, Figures 1 and 2 suggest that, if anything, it can be ruled out with sufficient confidence that the increase in acquisitions by big tech has positively affected the degree of competition in digital markets. Acquisitions by big tech have not been done to challenge each other’s market shares on their respective turfs ²⁴ It could be argued that this point disguises the belief that acquisitions lead to increased concentration. This would be an erroneous correlation-based causality inference. Here, however, the point is to emphasise that acquisitions did not lead to lower concentration rates. And since concentration levels are very high, this seems to be a relatively solid conclusion (in other words: it would be hard to argue that, in the counterfactual scenario, concentration levels would be even higher). .

Third, nearly all of these acquisitions were not scrutinised by competition authorities, since merger-control frameworks use filtering systems to focus only on a restricted number of mergers. Since 2000, the European Commission has opened only 27 investigations into acquisitions involving DMA gatekeepers (19 were notified directly to the Commission, while eight were referred by national authorities ²⁵ From the Commission’s database (), the full list of acquisitions by gatekeepers that led to the opening of an investigation at EU level (in italics, the cases that were referred by national authorities) is: Facebook/Meta: WhatsApp, Kustomer; Apple: KKR, Beats, Shazam; Microsoft: Blizzard, GitHub, LinkedIn, FIH Mobile, Nokia, General Electric, Skype, Yahoo, Time Warner, Nuance, Zenimax, Inflection; Booking: HotelsCombined, ETraveli; Google/Alphabet: Fitbit, Sanofi, Motorola, DoubleClick, ResMed, Photomat; Amazon: iRobot, MGM; ByteDance: none. ). Of these, two were withdrawn, one was blocked and the remaining cases were cleared ²⁶ The withdrawn mergers were: Microsoft/Time Warner/Contentguard (M.3445) and Amazon/iRobot (M.10920). The prohibited merger was Booking/eTraveli (M.10615). In one case, Microsoft/Inflection, the Commission withdrew its investigation after the EU Court of Justice judgement on Illumina/GRAIL (see section 3). . The absence of the enforcement of merger control may indicate that a subsample of acquisitions by big tech had anti-competitive effects: simply, they were not checked. In other words, had there been scrutiny of all mergers, and had merger control authorities had the ability to identify anti-competitive mergers with reasonable accuracy ²⁷ Notification thresholds are only part of the problem. Antitrust authorities also face the difficulty of running a complex potential competition analysis on uncertain future market dynamics. This is why several authors propose changes in the standard of proof (eg Motta and Peitz, 2020; Cabral, 2024). See section 4. ,today’s digital markets would likely be more competitive.

There is thus a compelling case to consider expanding competition authorities’ scrutiny area to capture also smaller transactions not currently subject to scrutiny, though it should be noted that because of the potential pro-competitive effects of acquisitions, it could be argued that excessively zealous enforcement could ultimately reduce competition, instead of enhancing it.

In practice, however, these concerns have no empirical backing. The actual risk that a pro-competitive merger in Europe is blocked is currently close to zero because very few mergers are blocked in first place. European Commission (2024) found that concentration in the EU at both industry and market levels increased in the past 25 years, while business dynamism declined, shown by increasingly stable market shares ²⁸ For example, average industry concentration as measured by the sum of the top-four market players increased by 5 percentage points between 2000 and 2019 (European Commission, 2024).  Meanwhile, from 2000 to 2024, only 22 out of the 8289 mergers notified to the Commission (0.3 percent) were prohibited (117 notified mergers – 1.4 percent – were cleared with commitments by the merging parties) ²⁹ See summary statistics on merger cases at ; note that the reported figures do not include merger control at the national level. . It thus seems far-fetched to claim that over-enforcement is a concern. In fact, these figures suggest that stronger enforcement of merger rules could help preserve competition in increasingly concentrated markets.

At EU level, under the EU Merger Regulation (EUMR, Regulation (EC) No 139/2004), mergers and acquisitions must be notified only if the turnover of the involved companies exceeds a certain threshold (EUMR Art. 1) ³⁰ For details of the thresholds, see European Commission, ‘Mergers procedures’, undated, . . For the purpose of this analysis, it is enough to note that a necessary (not sufficient) condition for notification at EU level is that the acquired company has a turnover of at least €100 million (or at least €250 million if it does not make €25 million in annual revenues in each of at least three EU countries).

The European Commission is aware that gaps in the merger control framework are a risk for sectors where potential future competition is extremely relevant, especially the digital and pharmaceutical sectors. Until recently, it thus resorted to EUMR Article 22, which allows EU countries to refer to the Commission mergers that do not qualify for notification at EU level. This option was originally introduced to allow EU countries to ‘use’ the Commission to review mergers, even if they did not yet have a merger control system in 2004 when the EUMR was introduced. 

In 2021, the Commission published guidance on the interpretation of EUMR Art. 22 (European Commission, 2021). This implied that member states could refer a merger to the Commission even if they have a merger-control system and the transaction does not qualify for national notification thresholds. This interpretative stretch de facto allowed the Commission to capture any merger, the only condition being the willingness of national authorities to flag to the Commission the potential problematic nature of a transaction. This, in combination with the greater monitoring power the Digital Markets Act conferred on the Commission subsequently (DMA Art. 14 obliges gatekeepers to inform the Commission about any intended concentration, even if not notifiable, not even at national level), gave the Commission the tools to be notified about and potentially tackle any acquisitions by gatekeepers.

But in September 2024, the EU Court of Justice struck down the Commission’s interpretation ³¹ In the context of the Illumina/GRAIL case. See European Commission press release of 18 September 2024, ‘Commission takes note of the withdrawal of referral requests by Member States concerning the acquisition of certain assets of Inflection by Microsoft’, . , taking a position against the uncertainty for business entailed by the high degree of flexibility the Commission granted itself. The Court ruled that referrals to the Commission under EUMR Art. 22 can only be done if the merger qualifies for notification at national level.

This has reopened a loophole. The Commission is now dependent on member-state rules for notification of mergers to national authorities. However, national merger-control frameworks capture the acquisition of small companies only in some countries.

Figure 3 shows main merger notification criteria per EU country. Spain, Portugal and Slovenia have specific market-share thresholds for notification, and can thus require notification based solely on the acquirer’s already strongly cemented market position. Hungary, Cyprus, Ireland, Sweden, Denmark, Latvia, Lithuania and Italy grant their competition authorities the ability to request notification if certain conditions are met (usually referred to as call-in powers) ³² Luxembourg does not have a merger control framework yet. That paradoxically allows Luxembourg discretionary powers, since it can refer any transaction to the Commission pursuant to EUMR Art. 22. . Germany and Austria have thresholds based on the overall value of the merger deal, thereby extending their reach to highly valuable acquisition targets, even if they presently have zero turnover. In all other EU countries, the target’s turnover is a strict condition for notification ³³ With the exception of Poland, where mergers threshold are based on turnover, but a zero-turnover target is still notifiable.  (see the appendix for details).

The German and Austrian approach is particularly intriguing since the price paid by the acquirer can be a better indication than the target’s turnover of the target’s future potential. In case of deliberate anticompetitive intentions on the part of the acquirer (for example, in the case of prospective killer acquisitions), the price the acquirer would be willing to bid would not be lower than the loss it would expect to incur in the future, if the target’s product entered into competition with its product ³⁴ Fumagalli et al (2020) suggested that a high transaction value could be used not only for notification but also for the assessment of the merger’s effect on competition. . Compared to turnover however, transaction value-based notification thresholds are less able to capture the link between the involved companies and the competition authority’s jurisdiction (the price of the deal does not indicate its significance for the market over which the authority exerts its merger control). That is why transaction value may be used to complement, rather than fully substitute turnover thresholds.

In the US, there is currently a size-of-transaction notification threshold of $111.4 million, but also a threshold for the size of the acquiring and the target company, the ‘size-of-person’ threshold. A 2021 Federal Trade Commission (FTC, 2021) study pointed to “l�ǴDZ��DZ���� [in the US notification system] that are unjustifiably enabling deals to fly under the radar” ³⁵ Remark of the FTC Chair Lina M. Khan Regarding Non-HSR Reported Acquisitions by Select Technology Platforms (2021). See . Note that, in theory, the FTC has the power to unwind anticompetitive mergers that have already taken place ³⁶ The European Commission can also unwind mergers that have already taken place, but only if the merger was subject to notification. . It is, however, difficult to see how such an ex-post adjustment could play out in fast-changing markets: by the time the ex-post investigation has been completed and the merger unwound, markets may look completely different.

Figure 3: EU countries’ merger notification requirements 

^{Source: Bruegel. Note: For a detailed explanation of notification criteria see the appendix.}

Currently, even after the EU Court of Justice ruling, the Commission can still rely on national authorities to capture small mergers, provided that these are notifiable under national law. Note that, up to 2020 (ie before the publication of the Art. 22 guidance), the Commission reviewed 40 Art. 22 referrals from member states. 

For the mechanism to work properly, however, lagging countries need to update their notification frameworks to capture small mergers. More harmonisation of national approaches is indispensable to reduce uncertainty and minimise costs for cross-border businesses.

Yet, even if all EU countries reformed their merger laws, uncertainty and the risk of regulatory capture would remain. The Commission would remain at the mercy of the national authorities. EU countries may decide not to refer a merger for reasons of national interest, and companies could engage in forum shopping, locating where authorities are reputedly less eager to refer mergers to the Commission. It is telling, for example, that, in the aftermath of the September 2024 EU Court decision, only Italy referred the Nvidia/Run:ai merger (M.17766), despite the relevance of the merger ³⁷ The merger was cleared unconditionally by the European Commission in December 2024.  (as discussed above, other seven EU countries would have had the power to call-in the merger and refer it to Commission).

It should also be noted that the fact that the EU has adopted a fully-fledged regulatory framework constraining the behaviour of digital gatekeepers, the DMA, does not reduce the need to scrutinise mergers for potential anti-competitive effects. If anything, the DMA exacerbates concerns about potentially anti-competitive acquisitions. For example, the DMA may increase the incentives to carry out anti-competitive mergers for exclusionary purposes, since the DMA prevents the acquirer from engaging directly in exclusivity practices ³⁸ That is, the acquirer buys a startup that supplies key inputs to competitors and then closes it down. See Motta and Peitz (2020) for a general discussion on the impact that regulation has on the acquirer’s incentives to engage in acquisitions for exclusionary purposes. . Klein (2024) argued that the DMA is likely to undermine the Commission’s ability to impose structural remedies in case of potential anti-competitive mergers, since the DMA could cover the same sources of harm (the difference being though that the DMA imposes behavioural constraints, which are unlikely to bring more structural competition to the market).

Moreover, it is too early to say whether the DMA will be effective. It was drafted based on the Commission’s past antitrust experience and may be not suitable to address novel challenges (the DMA entered into force in 2023 and no infringement has yet been sanctioned) ³⁹ For a discussion of the challenges in DMA implementation, see De Streel et al (2024). .

Sticking to the status quo in terms of merger control in the EU is problematic. The significant research efforts of public bodies shedding light on this topic suggest that public authorities have been receptive to that warning call ⁴⁰ See for example Scott Morton et al (2019), Furman et al (2019), Crémer et al (2019), Pike (2020). . Mergers involving small companies can be anti-competitive, but the European Commission currently has very limited, indirect jurisdiction to vet them.

4.1 EUMR revision?

The cleanest option to fix the system would be to amend the EUMR by designing new notification rules based on a combination of the solutions so far explored at national level ⁴¹ Filtering systems are intended to prevent competition authorities being flooded with small merger notifications. It would thus be unimaginable to not use a filter to select the mergers that merit scrutiny: the administrative burden could not be sustained. , such as application of thresholds based on transaction value or the aggregated market share of the merged entity. This would increase certainty for business and equip the Commission with a general tool applicable to any company meeting the new criteria, regardless of its size or sector. It would be relevant for the pharmaceutical sector, for example, where killer acquisitions are a source of significant concern.

However, while the EUMR may have been underenforced in the last 20 years, its structural design is good. The EUMR does not lack tools to perform accurate substantive assessments in the digital economy. Rather, the Commission may not always have the expertise to use them properly. EUMR reform would thus entail a risk of overshooting: in the EU legislative process, it is often difficult to anticipate how a core single market regulation could turn out, once it is opened up for revision and national interests come strongly into play. Such a reform would be done at a time when there is high pressure (particularly from EU countries) to soften merger control to favour the creation of national champions ⁴² For example, Giovanna Faggionato and Hans von der Burchard, ‘Germany and France push for mega-deals in competition overhaul’, Politico, 28 May 2024, . . New tools to block acquisitions of startups are unlikely to be worth opening the gates to harmful waves of national consolidation.

Less risky and more practical would be simply to amend the Digital Markets Act. Furman et al (2019) proposed picking certain market players that enjoy a “strategic market status” and making them subject to specific merger notification requirements. Luckily, the EU has already identified those players: the DMA gatekeepers. The Commission could propose amending DMA Art. 14 to allow for special scrutiny of any acquisition performed by gatekeepers ⁴³ Note that, with this solution, some potentially worrisome transactions in digital markets would not be captured if the acquirer is not a DMA gatekeeper. .

4.2 A special DMA merger control regime

Amending the DMA would be simpler than opening up the EUMR. The amendment could allow the Commission to obtain jurisdiction to vet any big-tech acquisition; it also could create a mechanism to increase accuracy in the authority’s decision-making. Of course, the extension of merger scrutiny carries risks. It would increase administrative costs, for example. At the same time, the theoretical analysis of the economic dynamics leading to harm (both in the case of anti-competitive mergers that go through and in the case of pro-competitive mergers that are prohibited), shows the complexity of competition supervision. The Commission, like other competition authorities, lacks the expertise to foresee the evolution of competitive dynamics in very new and complex markets and may be unable to spot when the acquisition of a small company can be problematic, even if the merger is notified.

Thus, it is advisable to shift the burden of proving that a merger should not be considered anti-competitive onto the player with the best view of the market – usually, the acquiring company, even though even the acquiring company may not be fully able to anticipate how the market will evolve. Motta and Peitz (2020) argued for such a reversal of the burden of proof. They noted that while it is important to recognise the existence of significant dynamic uncertainty in digital markets, it is “no secret that ��…]&�Բ�����; established market players have a lot of data and data analytics capabilities to sniff around and detect potential competitors”.

A special DMA merger control regime should thus be designed to minimise administrative costs for the Commission and the involved parties, while relying on market players’ knowledge to increase the accuracy of assessments.

The special regime would entail a three-stage process:

1. A gatekeeper would flag its intention to acquire a company. The gatekeeper would provide basic information about the target company without disclosing its business plans to the Commission initially. The Commission would then publish publicly disclosable information about the gatekeeper’s intention to buy the target, giving users/stakeholders the opportunity to provide information within a, say, 30-day deadline. This could be done via a simple form in which only basic information could be provided anonymously, particularly about any potential concern that the acquirer might downgrade the target’s product, or regarding the counterfactual, including whether other credible potential buyers might be interested in the target. Based on the information received, the Commission would decide whether the transaction is sufficiently suspicious to qualify for a deeper preliminary assessment.

2. In that case, the Commission would open a second phase during which the gatekeeper would disclose its plans to the Commission, explaining why the merger does not entail anti-competitive risks.

3. Should the preliminary dialogue between the gatekeeper and the Commission fail to dispel concerns, the Commission would then open the third phase, which would coincide with the standard merger procedure (pre-notification, notification, phase I and phase II, where necessary), except that the burden of proof of showing that the merger is not harmful would be on the gatekeeper.

This approach would have the advantage of defining a clear timeline that the competition authority must respect in order to minimise the impact on business. It would rely on market players’ superior knowledge of market dynamics and would shift the burden of mitigating concerns to the most informed player: the acquiring gatekeeper.

Increasing the accuracy of authorities’ evaluations of dynamic effects is the ultimate challenge for merger enforcement in digital markets. There is no straightforward way around it, no ready-at-hand tool to improve the ability of authorities to forecast the threat to competition a small company could manifest in future. However, any improvement in assessment is likely to depend strongly on the quality of the authorities’ access to the information available to market players. This proposed special regime should bring fresh opportunities for better, more thorough decisions by the Commission, in addition to simply increasing the number of deals subject to screening by the Commission.

At the same time, the special regime would not exclude continued reliance on EUMR Art. 22 in order to use referral by member states to gain jurisdiction on potentially problematic acquisitions of small companies by non-DMA gatekeepers (thus also covering economic sectors other than digital, such as pharmaceutical).

Together with amending the DMA, supporting and guiding EU countries towards harmonisation of national approaches to tackle acquisitions of small companies should thus remain a priority for the Commission.

Affeldt, P. and R. Kesler (2021) ‘Big tech acquisitions—towards empirical evidence’, Journal of European Competition Law & Practice 12(6): 471-478, available at  

Anderson, J. and M. Mariniello (2021) ‘Regulating big tech: the Digital Markets Act’, Bruegel Blog, 16 February, available at /blog-post/regulating-big-tech-digital-markets-act 

Argentesi, E., P. Buccirossi, E. Calvano, T. Duso, A. Marrazzo and S. Nava (2021) ‘Merger policy in digital markets: an ex post assessment’, Journal of Competition Law & Economics 17(1): 95-140, available at  

Barsy, L. de and A. Gautier (2024) ‘Big Tech Acquisitions and Innovation: An Empirical Assessment’, CESifo Working Paper 11025, available at  

Benkert, J.-M., I. Letina and S. Liu (2023) ‘Startup acquisitions: Acquihires and talent hoarding’, mimeo, available at  

Bourreau, M. and A. de Streel (2019) ‘Digital conglomerates and EU competition policy’, mimeo, available at  

Bryan, K.A. and E. Hovenkamp (2020) ‘Antitrust limits on startup acquisitions’, Review of Industrial Organization 56: 615-636, available at  

Cabral, L. (2024) ‘Big tech acquisitions’, International Journal of Industrial Organization 103127, available at

Crandall, R.W. and T.W. Hazlett (2022) ‘Antitrust in the Information Economy: Digital Platform Mergers’, The Journal of Law & Economics 65(S2), available at  

Crémer, J., Y. Montjoye and H. Schweitzer (2019) Competition policy for the digital era, European Commission Directorate-General for Competition, available at doi/10.2763/407537 

Cunningham, C., F. Ederer and S. Ma (2021) ‘Killer acquisitions’, Journal of Political Economy 129(3): 649-702, available at  

De Streel, A., M. Bourreau, R. Feasey, A. Fletcher, J. Kraemer and G. Monti (2024) Implementing the DMA: substantive and procedural principles, Centre on Regulation in Europe, available at  

Ederer, F. and B. Pellegrino (2023) ‘The great start-up sellout and the rise of oligopoly’, AEA Papers and Proceedings 113: 274-278, available at  

Eisfeld, L. (2024) ‘Entry and acquisitions in software markets’, mimeo, available at . github.io/assets/docs/JMP_Eisfeld_TSE.pdf 

European Commission (2021) ‘Guidance on the application of the referral mechanism set out in Article 22 of the Merger Regulation to certain categories of cases’, 2021/C 113/01, available at  

European Commission (2024) Protecting competition in a changing world, Directorate-General for Competition, available at
15 Policy Brief | Issue n˚11/25 | March 2025

FTC (2021) Non-HSR Reported Acquisitions by Select Technology Platforms, 2010–2019, Federal Trade Commission, available at  

Fumagalli, C., M. Motta and E. Tarantino (2020) ‘Shelving or developing? The acquisition of potential competitors under financial constraints’, Economics Working Papers 1735, Department of Economics and Business, Universitat Pompeu Fabra, available at  

Furman, J., D. Coyle, A. Fletcher, D. McAuley and P. Marsden (2019) Unlocking digital competition, Report of the Digital Competition Expert Panel, available at  

Kamepalli, S.K., R. Rajan and L. Zingales (2020) ‘Kill Zone’, NBER Working Paper 27146, National Bureau of Economic Research, available at  

Klein, L. (2024) ‘Merger remedies in the era of the Digital Markets Act (DMA): the impact of the DMA on the EU Merger Control Regulation (EUMR) in designing commitments’, European Competition Journal, 1–27, available at  

Letina, I., A. Schmutzler and R. Seibel (2024) ‘Killer acquisitions and beyond: policy effects on innovation strategies’, International Economic Review 65(2): 591-622, available at  

Mariniello, M. (2022) Digital economic policy: The economics of digital markets from a European Union perspective, Oxford University Press

Motta, M. and M. Peitz (2020) ‘Removal of potential competitors – a blind spot of merger policy?’ Competition Law & Policy Debate 6(2): 19-25, available at  

Motta, M. (2023) ‘Self-preferencing and foreclosure in digital markets: Theories of harm for abuse cases’, International Journal of Industrial Organization 90: 102974, available at  

Pike, C. (2020) ‘Start-ups, Killer Acquisitions and Merger Control’, OECD Competition Papers, Organisation for Economic Cooperation and Development, available at  

Prado, T.S. and J.M. Bauer (2022) ‘Big Tech platform acquisitions of start-ups and venture capital funding for innovation’, Information Economics and Policy 59: 100973, available at  

Prat, A. and T. Valletti (2022) ‘Attention oligopoly’, American Economic Journal: Microeconomics 14(3): 530-557, available at

Rasmusen, E. (1988) ‘Entry for buyout’, The Journal of Industrial Economics 36(3): 281-299, available at  

Rhodes, A. and J. Zhou (2019) ‘Consumer search and retail market structure’, Management Science 65(6): 2607-2623, available at  

Scott Morton, F., P. Bouvier, A. Ezrachi, B. Jullien, R. Katz, G. Kimmelman, D. Melamed and J. Morgenstern (2019) Committee for the Study of Digital Platforms, Market Structure and Antitrust Subcommittee, report, Stigler Center for the Study of the Economy and the State, available at  

Teece, D.J. (2020) ‘Innovation, governance, and capabilities: Implications for competition policy: A tribute to Nobel laureate Oliver Williamson by his colleague and mentee David J. Teece’, Industrial and Corporate Change 29(5): 1075-1099, available at  

Varian, H.R., J. Farrell and C. Shapiro (2004) The economics of information technology: An introduction, Cambridge University Press