Cornerstone Research, in collaboration with the Dynamic Competition Initiative (DCI), hosted its third annual Brussels Expert Forum, at which academic, regulatory and industry leaders gathered to discuss the rapidly evolving state of modern competition policy.
Three expert panels were held on 9 June 2026 that discussed the European Commission’s new draft merger guidelines, the treatment and role of dynamic competition, innovation and data-based intellectual property (IP) in competition enforcement, and the transatlantic contrasts emerging in the regulation of artificial intelligence (AI).
Diving into the Currents of Global Antitrust Enforcement Policy
The event opened with a discussion of the Commission’s draft merger guidelines, the most significant revision to EU merger control in the past two decades.
Central to much of the conversation was the Commission’s attempt to draw a concrete line between pro-competitive scale and anti-competitive market power. This framing comes in response to prior arguments shaped by the Draghi report and similar debates around competitiveness, while keeping in place the significant impediment to effective competition (SIEC) test as the binding legal standard. Reflecting on this development, the panel agreed the Commission’s attempt was not to diminish enforcement standards but to enact a more straightforward, consistent and resilient analytical framework.
The panel also spent time highlighting the draft’s convergence with the 2023 US merger guidelines, most notably the unified horizontal and non-horizontal structure, as well as expanded emphases on dynamic effects and explicit acknowledgement of theories of harm beyond price. However, the panel also identified and unpacked key areas where the draft diverges from its US counterpart, including its heightened attention to resilience and sustainability when considering efficiency, a more developed innovation theory of harm, and the draft’s references to the “margin of appreciation” that the Commission enjoys in merger assessment.
The speakers also discussed the draft’s introduction of the theory of benefit as a formal analytical concept and counterpart to the theory of harm, evaluating the role of efficiencies in the assessment process. On this point, some scepticism emerged as panellists questioned whether the evidentiary standards for theories of benefit could truly be considered symmetrical with theories of harm in practice.
Some concerns were also raised about the operationalisation of the innovation theory of harm, including the lack of clear limiting principles and resolution of the question of out-of-market efficiencies. Panellists noted key institutional implications of the draft, most notably that expanded factors and greater discretion require the Commission to be more transparent and rigorous than ever in showing its work to build durable case law and withstand judicial review.
Evolving Standards in EU Merger Review
While the first session took a macro view of EU merger guidelines, the second panel focused more on breaking down granular realities of implementation, examining specific theories of harm and benefit introduced or expanded on in the draft, before discussing exactly how these two are to be weighed together in practice.
In discussing the Commission’s approach to innovation and investment theories of harm, the panel established and drew a distinction between three analytically separate concerns. These included the elimination of a potential competitor; loss of innovation competition, primarily as it relates to the R&D process itself; and loss of investment competition, particularly in telecommunications and other network industries. On these points, the panel emphasised the importance of parties providing the Commission with a clear, provable account of transaction rationale as early in the process as possible, highlighting in unambiguous terms what is being combined, the incentive effects post-merger, and why each efficiency claim should be deemed credible and merger specific.
On the topic of common ownership, an additional concern was expressed that the new standalone theory of harm introduced by the draft, which posits a 5 per cent shareholding threshold, could be seen as an overcorrection that could unfairly penalise ordinary diversification by investors.
Speaking to the draft’s introduction of “the innovation shield,” which intends to offer protection in principle from an SIEC finding for the acquisition of small, innovative companies, many panellists viewed the eligibility criteria as too narrow. However, most agreed the shield was a welcome development, with some arguing it would in fact provide genuine safe harbour comfort if all criteria were met, but not without highlighting the need to resolve certain definitional ambiguities in practice.
Moving Towards Growth-Oriented IP Rules
To close out the day, the final panel explored the growing role of IP, and specifically data, as a strategic asset class when it comes to merger control and competition enforcement.
Specifically, panellists began by noting that, due to the draft’s new emphasis on dynamic competitive potential, innovation capacity, and long-term contestability, IP today is no longer a background factor but a major first-order consideration in competitive assessment. From here, the discussion evolved to include (1) the examination of how IP portfolios can be used as proxies for dynamic competitive potential in merger review, (2) how mandatory IP access doctrines have been transformed under EU competition law, as well as (3) the regional contrasts that are emerging in the influence of copyright law and regulations on how AI models are being developed today and in the future.
On the first topic, the panel was careful to note that while the economic literature does support a conditional link between patent portfolio strength and future competitive significance—i.e. firm value, innovative output and knowledge diffusion can all be reasonably predicted by citation-weighted patent counts—the relationship also depends highly on context and is sensitive to measurement approaches. In other words, although IP ownership can be informative about control positions and even innovation capabilities, it is in no way a guarantee of certain market outcomes. With this in mind, the panel agreed the appropriate response would be a rigorous, case-by-case assessment as opposed to mechanical reliance on aggregate IP metrics.
The panel then tracked the evolution of mandatory access, tracing a trajectory from the high threshold of Magill and IMS Health through the partial dilution of the standard in Microsoft I and II, to the now more permissive approach in Android Auto that considerably lowered the bar for access seekers when designing a platform. This relaxation of access standards in recent cases, according to some panellists, should be seen not as a doctrinal shift but rather as a judicial response to IP rights viewed as unreasonably broad compared to the actual creative or technical contribution they were drafted to protect.
Finally, the panel closed with an in-depth discussion on AI and copyright law, an area the speakers agreed is the sharpest current illustration of transatlantic divergence in IP regulation. For example, in the US, the flexible fair use doctrine allows AI training on copyrighted material to be adjudicated on a case-by-case basis, whereas the text and data mining framework in the EU’s Copyright Directive allows rights holders to either opt out or opt in to such use. Some panellists highlighted the concern that this asymmetry could place EU-based AI developers at a clear structural disadvantage while doing little to achieve the Commission’s underlying policy goals.
Learn more about Cornerstone’s European Competition practice.
Los Angeles
