It's been an unsettled end to the summer in many parts of Europe and the United Kingdom ...
...including when it comes to the frontiers of merger control enforcement!
The jurisdictional boundaries of both the European Commission (EC) and the Competition & Markets Authority (CMA) in the investigation of mergers have been tested in recent months, with divergent results.
In this week's blog post, we explore three important cases – Illumina, Inc. v. European Commission (an ECJ judgment, which followed the Commission's prohibition of Illumina / Grailand subsequent General Court judgment upholding the Commission's decision), Microsoft / Inflection AI (a CMA Phase 1 clearance decision, and an abortive European Commission Article 22 referral) and Amazon / Anthropic AI (a CMA 'found not to qualify' decision) – and consider what they foretell of the approaches in the European Union, the United Kingdom, and beyond, to tackling the elusive 'killer acquisition.'
First in the European Union, Where the Court Found That Two Wrongs Do Not Make a Right
Illumina, Inc. v. European Commission
In an important judgment of September 3, the European Court of Justice (ECJ) struck down the decision by the European Commission (EC) to investigate and prohibit Illumina's takeover of Grail three years ago.
That merger, originally signed in 2020, did not meet European Union Merger Regulation (EUMR) turnover thresholds. And no EU member State's national competition authority (NCA) had jurisdiction to review it under national legislation. Nonetheless, the Commission – departing from its long-standing practice of only accepting referrals under Article 22 EUMR from NCAs with jurisdiction to review the relevant merger – encouraged the French NCA, among others, to make a referral of Illumina / Grail even where no such jurisdiction arose. In the end, seven member States joined the referral, most without the power to investigate the merger in their jurisdiction. The Commission investigated and ultimately prohibited the deal in September 2021. It also published a new guidance note on Article 22, encouraging NCAs to make Article 22 referrals where they were concerned with mergers that did not meet their own jurisdictional thresholds.
That expansive approach to asserting jurisdiction over 'below threshold' mergers had introduced significant uncertainty into deal-making in the period since, including for many otherwise unproblematic transactions.
Now, three years later, the European Court of Justice (the highest Court in the European Union) has roundly rejected the use of Article 22 EUMR in this way, striking down the prohibition of Illumina / Grail and shredding the EC's Article 22 Guidance Paper. In a detailed judgment, it found that the so-called 'Dutch Clause,' designed to enable the EC to 'step in' to review mergers where no national merger control regime was in place (in 1989, when the predecessor to the EUMR was first introduced, the Netherlands had asked that the Commission hold such a power, since it lacked at the time a national merger control regime), could not be used in the way envisaged by the Commission. The ECJ considered both the literal, historical, contextual interpretation and teleological interpretation of Article 22, concluding that the existing framework of turnover thresholds must be respected and that the legislature had specifically reserved unto itself the right to revisit those thresholds in future if they were not working correctly. It was not for the Commission to usurp that function by extending its jurisdiction over mergers that otherwise did not qualify for investigation under the EUMR and the use of the Article 22 referral in this way was inconsistent with the statute's intention.
In essence, the ECJ held that two wrongs did not make a right, that not having jurisdiction under the EUMR or under national rules doesn't combine to give the Commission jurisdiction. Thus, the Commission's workaround solution to 'killer acquisitions' was dealt a 'killer' blow.
Meanwhile, in the United Kingdom... Two Marginal Cases That Clarify the CMA's Broad (but Not Limitless) Jurisdiction to Investigate Mergers at the Frontier
Microsoft / Inflection AI
The day after the ECJ's judgment in Illumina, Inc. v. European Commission, the CMA announced that it had found Microsoft's hiring of former employees from Inflection AI to qualify as a 'relevant merger situation' under its jurisdictional rules, since (i) the transfer amounted to two enterprises ceasing to be distinct, (ii) the share of supply test was met, and (iii) the merger had not completed or had completed less than four months from the date of the decision.
The CMA's decision that Microsoft / Inflection AI constituted an 'enterprise' is highly significant from a jurisdictional perspective, much more than the substantive outcome (i.e., clearance), which was expected given the fragmented nature of the AI sector. In its summary of the Phase 1 Decision (the full decision has yet-to-be published), the CMA reiterated the very low threshold for establishing what qualifies as an enterprise from a jurisdictional perspective, noting that there "is no particular combination of assets that constitutes an enterprise... [ ] it may include a group of employees and their know-how where this enables a particular business activity to be continued" and may include "acquiring a team with relevant know-how – even without further assets.. [ ]."
This case confirms the CMA's broad interpretation of its own jurisdiction, to potentially include the bare transfer of employees (since "know-how" can in its view simply include the 'collective' knowledge of the employees themselves, not necessarily any standalone IP or other technology).
In contrast, and to further compound the Commission's retrenchment post the ECJ's judgment in Illumina, Inc. v. European Commission, it was then announced – on September 18 – that the seven NCAs that had referred Microsoft / Inflection AI to the Commission had withdrawn their Article 22 referral (only France and Ireland, it appears, had had jurisdiction to review the transactions under national legislation and therefore could lawfully make the referral, the others could not). We now have only a Competition Policy Brief to suggest what the Commission might have explored had the Article 22 referral gone ahead. It's helpful but just a list of potential issues in the use and application of AI.
Amazon / Anthropic AI
Our third and final case provides further clarification on the CMA's view of its own jurisdiction (well, sort of). On September 27, in the Friday afternoon news cycle, the CMA announced that it would not be investigating Amazon / Anthropic AI (a £4 billion partnership between the two companies) as it did not qualify for investigation.
While, as with Microsoft / Inflection AI, only a summary of the decision is available, it shows that the CMA could not establish that the merger would satisfy its share of supply test (which requires a combined share of 25% or more on a market or markets in the United Kingdom, with an increment).
Interestingly, the CMA did not conclude on whether the partnership itself amounted to 'enterprises ceasing to be distinct' (an element the CMA must establish by reference to whether one company exerts 'material influence' over the other, a lower bar for establishing 'control' than in the European Union). This suggests that such a finding was not straightforward for the CMA in this case and that – rather than conclude one way or another – they are keeping their powder dry for now. It also demonstrates that even the CMA cannot investigate every tech case.
As with Microsoft / Inflection AI, we will be reading the full decision of Amazon / Anthropic AI closely, when it is published, for clearer indications of where the CMA now sees its jurisdictional horizon.
Forecasting Merger Control Enforcement at the Frontier
As we move from September to October and Summer to Autumn, the picture in the European Union and United Kingdom as regards merger enforcement at the frontier is a mixed one, with a forecast of 'persistent uncertainty' and a continued focus on creative ways by enforcers, both sides of the Channel, to try to investigate 'tech' mergers.
Europe
In the European Union, we have a return to the practice pre-Illumina / Grail, namely, NCAs that do have jurisdiction under their national laws (including through the use of a below threshold 'call in' power, such as Iceland, Ireland, Italy, Norway, and Sweden) can choose to investigate those mergers on their own or join an Article 22 referral with similarly-placed NCAs to the Commission. And many that don't have that power already, want it (like the Netherlands). But, critically for business, post-Illumina / Grail, those NCAs that don't have jurisdiction under national laws cannot make an Article 22 referral to the Commission, and the Commission cannot accept such a referral.
Outgoing Executive Vice-President Margrethe Vestager made clear immediately following the ruling that the Commission was exploring alternative ways to call in problematic mergers falling between jurisdictional lines. How precisely that would be achieved has not been communicated, but it is unlikely to be via Article 22.
Of course, if the acquirer is a gatekeeper, it may need to notify the Commission of its mergers anyway under Article 14 of the Digital Markets Act.
And it remains possible, in a small number of cases, that NCAs may seek to rely on the direct effect of Article 102 TFEU – as per the novel approach established by the Belgian Competition Authority in the Towercast judgment – to (however imperfectly) investigate non-reportable mergers approaching monopoly, where they cannot otherwise rely on merger control-specific powers.
Set against all of that, the Draghi report, released in early September, calls for a fundamental rethink of the EU's approach to regulating businesses and calls for a more permissive approach to mergers. What that means in practice, remains to be seen, but could signal a more lenient approach to the enforcement of certain mergers (of, for example, national or European champions) and therefore a slight dampening of the risk of certain 'frontier' mergers that fit that profile from being called in.
The United Kingdom
In the United Kingdom, the bar for the CMA to assert jurisdiction was lower than most countries and Microsoft / Inflection AI has lowered it further still, demonstrating the CMA's willingness to interpret its jurisdictional rules expansively; in particular, when seeking to review tech mergers. This case establishes that even the bare transfer of employees may amount to a qualifying merger. And while Microsoft / Inflection AI was cleared quickly at Phase 1, the next example of this jurisdictional flex might – if the CMA does identify concerns – end up in the Competition Appeals Tribunal.
Similarly, while Amazon / Anthropic AI was found not to qualify, the published summary of the decision makes clear that the CMA is looking very closely at commercial partnerships of all shapes and sizes to see whether they also qualify for investigation under the UK's lowest test for a change of control – that of 'material influence.' The full decisions of both of these cases will make interesting, and important, reading, when they are eventually published.
This is, of course, in addition to new notification obligations shortly arising under the Digital Markets, Competition and Consumers Act, whereby mandatory notifications to the CMA of any acquisition (including minority stakes of just 15%) will need to be made by (i) any digital companies designated as having 'strategic market status' by the CMA, (ii) any foreign seeking to acquire an interest in newspapers, and (iii) any energy network enterprises, where the total consideration of the transaction is at least £25 million.
Altogether, this means that, for the foreseeable future, companies – particularly in the tech sector – will need to keep a careful eye on the forecast when revisiting existing, and commencing future, dealings in the market, and be prepared for the CMA to come a knocking.
The United States and Beyond
In the United States, guidance jointly issued by the FTC and DOJ in 2023 lowered the substantive (but not jurisdictional) threshold for intervention in merger cases, with the Merger Guidelines characterizing as presumptively anti-competitive mergers where the combined share of the parties exceeds 30%.
Thus, at least on paper, the enforcers' presumptions are that (i) mergers with a post-merger HHI less than 1,800 and a change in HHI less than 100 and (ii) mergers resulting in a combined post-merger share less than 30% and a change in HHI of less than 100, will not give rise to competition concerns.
For the time being, the turnover thresholds in the United States remain in force, with little sign of an interest or ability to engage in the kind of jurisdictional acrobatics by regulators in the United Kingdom and Europe (although, see our blog post from last week on how a Trump or Harris administration may change the longer-term approach to merger control).
Practical Advice for Businesses Contemplating M&A in the United Kingdom, Europe and Beyond
For now, as always, forewarned is forearmed.
- Consider the nexus of your businesses, the jurisdictions in which you operate, the turnover which you and your prospective target generate, and where, and the extent of any competitive overlaps or hotspots.
- Identify mandatory filings (alongside inward investment screening and foreign subsidy notification obligations).
- Seek advice early as to the risk of a call in by the CMA, other NCAs, or an Article 22 referral by European NCAs to the Commission.
Finally, with all of the above in mind, consider appropriate condition precedents in your transaction documents (and an appropriate long stop date) to ensure your risk is correctly mitigated in the context of your deal.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.