As artificial intelligence reshapes the global competitive landscape, antitrust authorities are ramping up scrutiny of deals and market practices in the sector. In the latest episode of our “Fierce Competition” podcast, antitrust/competition attorneys Mike Menitove, Antoni Terra and Iacovos Antoniou break down recent developments in antitrust enforcement in the fast-changing AI landscape. The team explores high-profile investigations, merger control challenges and the impact of new digital market regulations across Europe and the U.S.
Episode Summary
AI dominates headlines from both a deal and antitrust enforcement perspective. Antoni Terra, counsel in Skadden's Brussels office, moderates a discussion on recent developments and trends across Europe and the U.S. in the AI space with colleagues Mike Menitove, partner in Skadden's New York office, and Iacovos Antoniou, European counsel in Skadden's London office. The panel explores key merger control issues, the EU Digital Markets Act and U.K. Digital Markets Competition and Consumers Act, and ongoing litigation such as the U.S. Department of Justice’s challenge to Google's search business.
Voiceover (00:00):
Welcome to “Fierce Competition,” a podcast from Skadden's global antitrust and competition group that explores antitrust policy and enforcement around the world. Join our colleagues from across the continents as we discuss the latest developments and what they mean to you in an increasingly complex legal and regulatory landscape.
Antoni Terra (00:20):
Hi everyone, and welcome to this new episode of “Fierce Competition,” Skadden's antitrust podcast. Today, we will be discussing developments and trends in the area of artificial intelligence and competition law, a sector which is obviously driving massive investment across the globe. I'm Antoni Terra, counsel at Skadden Brussels office, and I'm joined by Mike Menitove, partner in our New York office, and Iacovos Antoniou, European counsel in our London office. In this podcast, we will delve into key merger control issues and current antitrust investigations across Europe and the U.S. in the AI space.
(00:57):
We'll also explore how the EU Digital Markets Act, the DMA and the UK Digital Markets Competition and Consumers Act may affect AI companies and other companies operating in the sector. And we will also touch upon ongoing litigation in the U.S. with a link on AI including the U.S. Department of Justice’s challenge to Google's search business. And then we'll conclude with practical takeaways for companies. Just to set the scene a little bit, AI indeed continues to dominate headlines left and right from both a deal and an antitrust enforcement perspective.
(01:32):
The EC, for example, has labeled AI as a priority sector, and then the French and the Italian authorities recently outlined examples of pro and anti-competitive conduct in the AI space. Also, the CMA in the UK pioneered a year or two ago these in-depth studies on AI foundation model. So, a lot going on in it.
Michael H. Menitove (01:53):
And from the U.S. perspective, AI and antitrust also were very much in the headlines. In July 2025, the Trump administration issued an AI action plan focusing on achieving AI dominance, global dominance for the United States, and the leaders of both the U.S. agencies tasked with enforcing the antitrust laws. The Federal Trade Commission and the DOJ's antitrust division have emphasized the importance of ensuring that AI markets remain competitive. So, I guess let's jump in and get started, and perhaps we can kick off the discussion with AI partnerships which have come under regulatory scrutiny both in Europe and in the United States.
(02:35):
So, Antoni, can you kick us off and explain what those AI partnerships are and how the European competition authorities have addressed them?
Antoni Terra (02:43):
Yeah, of course. So, values, agreements and collaborations that we've seen recently in the field of AI, sometimes getting access to valuable resources, distribution channels or in general know innovation, this has been the case for foundation models for AI, access to cloud, access to data services at an upstream level, and then downstream also getting access to AI applications. Now from a jurisdictional perspective, starting from Europe so far, or at least up until a while ago, the field was relatively predictable. We had our framework of filing thresholds based on quantitative metrics.
(03:25):
The issue has been though that competition authorities, including the European Commission in recent years, have been worrying about the so-called killer acquisitions or large companies acquiring nascent competitors that can have a lot of future market potential. And they started to think, how can we make sure that we capture these transactions as part of our merger control procedures and therefore we make sure that we can review them. We make sure that there is no elimination of relevant competition or innovation.
(03:55):
The Commission did that mostly through what we call the Article 22 of the EU Merger Regulation, which is a referral process from the national competition authorities in the European Union to the European Commission. And they issued guidance about Article 22 a couple of years ago, basically expanding the scope of these referrals, making it very easy for national competition authorities to refer deals to the EC. That was the situation, the case, that got Illumina/Grail in front of the European Commission, and that ended in a prohibition decision. That got appealed, went to the Luxembourg Tribunal, the Court of Justice.
(04:35):
The Court of Justice said that they were not okay with that expansive approach that the Commission was taking to the Article 22 referral. So, back to square one, we went back to the more restrictive Article 22 referral. Given this outcome, and in parallel to that outcome, many member states in Europe have been enacting the so-called call-in powers. Right now, we have about 10, 12 member states that have these call-in powers, which means that even for transactions that fall below the national thresholds, they can call in deals and potentially they could refer that deal to the European Commission.
(05:16):
And then also about 10 member states in the European Union are considering to implement and exercise these call-in powers. So, a lot moving in this space from a jurisdictional perspective. One recent example of this at the wake of Illumina/Grail has been NVIDIA-Run:ai – the Italian authority making use of below threshold call-in powers and then referring the deal to the European Commission, European Commission taking the case. One final note on this, which is the Towercast ruling in the European Court of Justice. This is another example of flexibility where basically from a merger control perspective, authorities can make use of rules and of use of dominance to review those cases.
(05:59):
And then finally, from a digital regulation perspective, EU and UK require designated companies, so gatekeepers or companies with strategic market status to new reporting obligations about their deals. Now the second topic, apart from below threshold considerations is below control deals. So, as you know, the European Commission has a very standard way of looking at how you can capture concentrations, which is acquisition of control. In the UK and in some member states in the European Union, that's a bit different. It's a lower threshold based on influence. And in Austria, for example, they have a looser consideration or definition of control.
(06:45):
The idea here is to also from a concentration definition perspective, try and capture deals and be able to review that. However, from an AI point of view, the deals and the partnerships that Mike was referring to that we've seen in the last year, so Alphabet Anthropic, Amazon Anthropic, Microsoft Mistral, Microsoft OpenAI, the European Commission has been looking at them, again, through the prism of control and they understood that given the amount of investment, the positive and negative rights provided to the buyer or to the investor in these transactions, that was not enough to consider.
(07:24):
That was an acquisition of control and therefore all these transactions have not been actually reviewed by the European Commission. With that, Iacovos, over to you to discuss some aspects about the CMA.
Iacovos Antoniou (07:36):
Thanks, Antoni. The first transaction that became the subject of regulatory review was Microsoft's investment in OpenAI. It was a $13 billion investment, and OpenAI's known to everyone as the creator and owner of ChatGPT. This led to one of the longest pre-notification processes in the CMA's history. The CMA concluded after investigating that the transaction was not subject to review as there was no change from a material influence to de facto control by Microsoft over AI. However, the CMA acknowledged that careful consideration of commercial realities beyond the formal terms of the agreement was warranted in this case.
(08:14):
In another similar partnership, the EC and the CMA also concluded that Microsoft's 15 million Euro investment in French AI company Mistral, which is known for its open source models, did not qualify for review. In this instance, the CMA noted that Microsoft had not acquired the ability to material influence Mistral's AI policy. Similarly, the CMA's decisions in cases involving Amazon and Alphabet and their partnership with Anthropic, which is a company behind the Claude AI model, offer some practical guidance for companies.
(08:44):
So, the Amazon-Anthropic partnership involved a $4 billion investment, which was structured as convertible non-voting equity accompanied by extensive commitments for cloud computing provision by Amazon to Anthropic. Alphabet's investment was fairly similar. It also involved acquiring a non-voting minority stake and no board representation in Anthropic. Both companies received consultation rights via their partnerships, and the CMA found that this can be seen as relevant criteria in the assessment of material influence. Even if the CMA ultimately concluded that it did not have jurisdiction to review these two cases. Mike, over to you for jurisdiction issues in the US.
Michael H. Menitove (09:22):
Yeah, so in the United States in the technology space, the DOJ and FTC have an interesting recent history of allocating responsibility for investigations and enforcement. In 2019, the agencies agreed the DOJ would take the lead on investigating Google and Apple and the FTC would take the lead investigating Meta and Amazon. And ultimately, all four of those investigations have led to high-profile monopolization lawsuits that actually remain pending to this day.
(09:53):
As it relates to AI, it was publicly reported in June of 24 that the US agencies agreed that the DOJ would take the lead investigating NVIDIA, the chip maker that's now become the most valuable company in the world because of demand for its GPUs. And they also agreed that the FTC would take the lead investigating Microsoft and OpenAI. As it relates specifically to the context of AI partnerships, FTC initiated a study in January of 2024 regarding partnerships between cloud service providers and AI developers, and specifically the FTC studied some of the investments Iacovos was just talking about – Microsoft's investment in OpenAI and Google's and Amazon's respective investments in Anthropic.
(10:42):
That analysis was conducted pursuant to section 6B of the FTC Act, which authorizes the FTC to conduct studies that allow enforcers to gain a deeper understanding of market trends and business practices. So, the FTC used that power to issue subpoenas to each of those three cloud service providers, Microsoft, Google, and Amazon, and the two AI developers, OpenAI and Anthropic. And the results of the study were published in January of 2025, about three days before the transition from the Biden administration to the Trump administration. Because we just touched on reporting requirements in Europe, I wanted to briefly cover that ground in the United States.
(11:23):
In the U.S., the Hart-Scott-Rodino Act or HSR Act requires that parties notify the DOJ and FTC of proposed mergers, acquisitions of stock, or other business combinations that meet certain thresholds. The notification process allows the antitrust agencies to conduct reviews of proposed transactions before they're completed. Even if a transaction does not need to be reported under the HSR Act, the DOJ and FTC still have the authority to investigate any transaction, even after it closes, if they conclude that it may substantially affect competition in the United States.
(11:58):
There has been some reporting. For example, Microsoft's investment in OpenAI, which was not subject to HSR reporting requirements here in the U.S. and that's because under the HSR Act investments in non-corporate entities like OpenAI only trigger a filing requirement if the investor obtains control of the non-corporate entity. It sounds similar to the European analysis because Microsoft only obtained a 49% stake in OpenAI. It didn't obtain control and the HSR filing requirement wasn't triggered.
(12:43):
So, another category of AI investment that reportedly has been structured to avoid HSR filing requirements as the so-called aquihire, where the investor hires employees and licenses intellectual property from the target company, but does not acquire equity or other assets. So, why don't we take a moment to discuss these. So, maybe Antoni, can you tell us what's happened in Europe with regard to aquihires to date?
Antoni Terra (13:11):
Yeah, definitely. The key case that we have in Europe about aquihires is Microsoft hire for more than $600 million of most of the staff at the company Inflection AI around 70 people, as well as a non-exclusive IP license deal. So, as Mike was saying, really no assets involved. It's more the know-how and the people working at the company. The EC looked at that deal. They said that the substance of the deal, the grounds, the aspects of the aquihire fall under what can be considered a concentration per the EU merger regulation. But given that filing thresholds were not met in that transaction, they were not reviewing the deal, the European Commission. What about the aquihires, Iacovos, in the UK?
Iacovos Antoniou (14:06):
Thanks, Antoni. In this instance, the CMA emphasized that depending on the facts, an aquihire alone can be a merger that is subject to CMA review. So, in this case, the CMA noted that the acquired staff was at the core of any business seeking to develop foundation models or chatbots. While the transaction was cleared in Phase 1 as Inflection's products were not considered to pose a competitive constraint on Microsoft in the development supply of consumer chatbots, the comments in this case are still helpful in anyone seeking to engage in similar transaction structures in the future.
Michael H. Menitove (14:39):
So, in the U.S., according to public reports, there've been at least a few investigations that the US antitrust agencies have conducted with regard to these aquihire deals, but there haven't been any enforcement actions that have been taken to date. We've just talked about Microsoft/Inflection. So, in June of 2024, it was reported that the FTC was investigating that deal between Microsoft and Inflection AI, including whether Microsoft had structured the deal to avoid the HSR reporting requirements. And to date, no action has been taken by the FTC.
(15:13):
As another example, that same month June of 2024, Amazon paid a $25 million licensing fee to adapt AI and hired about two thirds of its AI researcher employees. The next month according to public reports, the FTC had launched an investigation into that deal, but again, no details of the investigation or any enforcement action have been announced to date. Lastly, in August of 2024, Google announced that it had paid approximately $2.7 billion for a non-exclusive license to character AI's large language model technology and to hire certain of character AI's employees.
(15:55):
After that deal was announced in May of 2025, it was publicly reported that the DOJ was investigating Google's agreement with character AI, including whether the transaction was structured to avoid the HSR reporting thresholds and further government scrutiny. So, Iacovos, in the UK, there have been some significant legislative changes to the merger control regime that may have an impact on the AI sector. Could you tell us a little bit more about those developments?
Iacovos Antoniou (16:21):
Yeah, sure. So, the UK recently introduced a new hybrid merger controlled threshold in January 2025, and this threshold gives the CMA jurisdiction to review transactions involving a party with a 33% share of supply in the UK and turnover in the UK in excess of 350 million pounds. So, the justification behind introducing this threshold is that the traditional jurisdiction of thresholds in the UK, which are based on the target's turnover in the UK or the party's combined share of supply can sometimes fail to capture transactions that could significantly impact competition.
(16:56):
And so, there is a need to be able to review deals involving firms with significant market positions, even if these transactions do not involve an actual overlap in the UK. Also, relevant for important digital players in the UK is the UK's Digital Markets Competition and Consumers Act whereby all digital companies that are designated as having strategic market status abbreviated as SMS, are now subject to mandatory notification of certain transactions with the UK Nexus and a deal value of 25 million pounds or more. This is quite a big change from the voluntary nature of the standard merger control rules in the UK. Antoni, what about the theories of harm considered by the authorities in the AI space?
Antoni Terra (17:40):
I would say at the general level that authorities have been looking at issues relating to access to critical inputs and then also the typical foreclosure concerns on downstream and related markets including tying, bundling, self-preferencing, usual theories, but that have another component, another dimension when we look at AI. Let's maybe examine NVIDIA Run:ai reviewed by the European Commission. So, the concern by the EC was mainly that NVIDIA would have allegedly a strong market position in discrete GPUs for data centers, and then how can they integrate vertically the products provided by Run:ai, which is essentially GPU orchestration software.
(18:25):
The EC examined first how NVIDIA might leverage its alleged dominance in GPUs to deny, or degrade compatibility with competing GPU orchestration software providers. That was one. Two, they also assess whether post-closing the merge entity might restrict Run:ai's orchestration software compatibility with GPUs supplied by NVIDIA's competitors and especially whether the strategy would reinforce NVIDIA's alleged dominance, therefore harming competitors and also consumers. And finally, the Commission also looked at anti-competitive bundling strategies, assessing whether NVIDIA could unfairly bundle GPUs with Run:ai software.
(19:09):
The decision in the end was an unconditional clearance by the European Commission in Phase 1. The Commission said that essentially because of the existence of many competitive alternatives to Run:ai, they didn't find serious concerns about this deal. As we were saying Microsoft/Inflection now from the perspective of the CMA, Iacovos.
Iacovos Antoniou (19:30):
Thanks Antoni. Looking at the CMA's analysis of Microsoft's acquisition of Inflection, the key focus in this case was assessing whether the transaction would significantly reduce direct competition between Microsoft and Inflection within specific AI-related market segments. The CMA assessed the extent to which the parties were close competitors based not only on the target's current offering, but also its plans and product development pipeline. First, the CMA evaluated the market for consumer chatbots examining whether the acquisition would significantly diminish competition by removing Inflection as an independent competitor.
(20:04):
The CMA considered the party's market shares, growth trajectories and innovation capacities, and concluded that Inflection held only a small market share in terms of chatbot use within the UK. Additionally, despite innovative ambitions, Inflection had struggled to materially increase or sustain user numbers. The CMA also evaluated Inflection's innovation pipeline and potential for future product differentiation. However, the assessment ultimately concluded that Inflection was not a critical source of innovation likely to substantially challenge Microsoft's market position.
(20:35):
There reason that Inflection's capabilities, while promising, did not yet demonstrate distinctive competitive features that couldn't be easily replicated by rivals. Another horizontal competition concern addressed was the potential loss of competition in the development and supply of foundation models. But since Inflection's plans for its AI studio business for enterprise customers were in the initial stages, it would have still involved the development and supply of foundation model.
Antoni Terra (21:02):
Still on merits or substantive analysis of deals, we obviously have a critical development in the European Commission, which is the general consultation of the review of the horizontal and non-horizontal merger guidelines that was launched in May 2025. It's been about 20 years since the issuance of the now-current merger guidelines. So, it's been a while since we've been using the same type of guidance from the European Commission. One of the topics in this consultation is specifically digitalization.
(21:36):
The EC has stated that a common business strategy of leading companies in the digital and tech sector has been to acquire complementary business, or key inputs and then that the aim is to strengthen their position in core markets. The EC acknowledges as part of this consultation, which is ongoing, that these may contribute to increase this innovation including for artificial intelligence, but that at the same time, this strategy can have negative effects in competition. Right now, as we say, the consultation is ongoing. The European Commission is planning to issue a draft of the revised merger guidelines at some point in the middle towards the end of next year.
(22:17):
Then the goal would be for the formal issuance of the new guidelines in 2027. So, still a while to go. We'll keep on top of this development, so please stay tuned for future editions of this “Fierce Competition” podcast because we'll certainly be speaking about the revision of the guidelines. Now, Mike, about the U.S. on theories of harm.
Michael H. Menitove (22:37):
Yeah, so in the U.S., while we don't have specific enforcement actions to point to, U.S. regulators have expressed concerns along the lines of what you were just describing, Antoni, about resources, inputs and the scale needed to compete in the generative AI space. Generative AI relies on massive amounts of data and computing power in the form of things like semiconductor data centers and cloud storage. Meanwhile, it seems a relatively small number of AI researchers have the ability to develop and train AI models. These scarce resources are in high demand and have become very expensive. So, at high level, US regulators have expressed concerns.
(23:21):
Similarly, it seems to what's been expressed in Europe – that some of these large tech players may be able to leverage their resources to stymie competition and new entrants in AI markets. Looking back to that 6B study that the FTC conducted beginning in January 2024 that I alluded to earlier, the FTC had studied the partnerships between cloud service providers, AI developers and in January of 2025, identified a few different potential antitrust theories both horizontal and vertical.
(23:55):
First, the study concluded that the partnerships between cloud service providers and AI developers might affect access to key inputs such as computing resources and AI researcher talent, which would impact competition for both AI developer partners and non-partner AI developers. For example, the FTC noted that the partnerships may create incentives for cloud service providers to limit access to computing resources, for AI developers other than their partners. The FTC also posited that the partnerships might consolidate access to that narrow talent pool of AI researchers. Second, the FTC noted that partnerships could increase the contractual and technical switching costs for AI developer partners.
(24:38):
Specifically, the FTC mentioned that exclusivity terms and other provisions and partnership agreements could prevent AI developers from obtaining computing resources from providers other than their cloud service providers. For example, the FTC referenced that OpenAI had agreed that Microsoft would be its exclusive cloud service provider. Now, that's no longer true today, which shows really how quickly the facts on the ground are changing. Since that FTC study was published in January of 2025, the relationship between Microsoft and OpenAI has evolved, at least a couple of times.
(25:14):
Microsoft went from being OpenAI's exclusive cloud service provider to having a right of first refusal to provide computing power to OpenAI, and now that right of first refusal is gone, and we've seen OpenAI enter into some recent agreements with other players to obtain computing power. So, in addition to contractual barriers, the FTC also identify technical barriers that could hinder AI developers’ ability to switch to alternative cloud service providers. Specifically, it may be difficult to switch cloud service providers because they use different AI chips and the software used to interact with the AI chips can be specific to the chip type.
(25:51):
Lastly, the FTC concluded that the partnership agreements could grant the cloud service providers access to AI developers competitively sensitive information including with regard to how their generative AI models work. So, this is a type of horizontal competitive issue, because the cloud service providers are developing their own generative AI models that may compete with their developer partners models. The FTC also noted that as it's a platform for their AI developer partners, the cloud service providers may gain insights into metrics and other valuable information that are otherwise unavailable to rival AI developers.
(26:29):
So, I'd referenced earlier that the DOJ had been tasked with conducting an investigation into NVIDIA. And according to public reporting, the DOJ is examining whether NVIDIA pressures customers not to purchase chips from its competitors or ties chip sales to purchases of other products including NVIDIA's CUDA software, which is used to interact with its chips. The FTC began investigating Microsoft's AI practices during the Biden administration, and it appears that that probe has expanded under the Trump administration.
(27:01):
According to public reporting, the FTC sent a detailed civil investigative demand to Microsoft requesting information about everything from the company's cloud computing business to its cybersecurity offerings and to its AI products. Among other things, the FTC is reportedly investigating whether Microsoft has imposed punitive licensing terms to prevent customers from switching from its Azure Cloud service to rival platforms, and whether Microsoft's bundling of its office products and security software with its cloud offerings have harmed competition.
(27:32):
The FTC is also reportedly investigating whether Microsoft has throttled back its own AI model projects after striking its deal with OpenAI and whether that partnership between OpenAI and Microsoft was structured to avoid the HSR filing notification requirements. So, Antoni, can I look to you to tell us a bit more about conduct investigations in Europe?
Antoni Terra (27:53):
Yeah, there are various pending probes related to alleged abuses of dominant position across the AI stack in Europe. Mike, as you're referring to in relation to NVIDIA, also the European Commission, the French agency, the UK CMA based on public reports maybe investigating NVIDIA's practices, potential exclusionary behavior or restricting access to essential GPU resources, these investigations appear to be ongoing. But based on public reports, not lots of news about this. There is something a bit more recent.
(28:27):
In Italy, July 2025, it was announced that the Italian competition authority raided Meta in relation to the pre-installation of its AI service on the WhatsApp application, and the authority apparently is investigating whether meta may be "imposing" the use of its chatbot and AI assistant services on its users, again, in relation to the use of WhatsApp. Finally, from an antitrust perspective in Europe, it seems that there are complaints that have been filed with the European Commission and the CMA against Google in relation to Google's AI-powered search summaries
(29:03):
you can see at the top of their search pages. Publishers and a non-profit organization would be alleging that Google is abusing its dominant position and generating these AI summaries in the search results. These agents argue that this approach is basically unfairly appropriating content from professional news sources without compensation, without permission and, therefore, they are distorting competition and harming publishers economically based on these complaints. This appears to also be a developing story. Obviously we'll be following it closely. We've discussed about merger control, jurisdictional aspects and merits aspects.
(29:43):
We've discussed conduct or antitrust investigations, and now, Mike, an obviously very big topic in the U.S., which is litigation. It would be great to hear from you on that aspect as well.
Michael H. Menitove (29:54):
Yeah, thanks Antoni. I'll start out by talking about the DOJ's ongoing litigation against Google. Interestingly, when that case was filed, it didn't have anything to do with AI. In October 2020, the DOJ and various state attorneys general sued Google alleging that it monopolized U.S. markets for general search services and general search text advertising. A bench trial was then conducted in the fall of 2023 before Judge Mehta of the district court for the District of Columbia in August 2024. The court held that Google has monopoly power and markets for general search services in general search text advertising.
(30:33):
The court further held that Google unlawfully maintained its monopoly power through exclusive agreements with browser developers like Apple and Mozilla, OEMs and wireless carriers. So, at that point, the case was focused on Google's power as a general search provider and its conduct as it relates to general search. But fast forward to the remedies phase, having found Google liable for violations of section two of the Sherman Act for monopolization, the court proceeded to the remedies phase. So in the spring of this year, the court held an evidentiary hearing on remedies and ultimately issued a decision on remedies on September 2nd, 2025.
(31:14):
And notably, generative AI played a significant role during the remedies phase and in the court's ultimate decision. In fact, on the very first page of its remedies decision, the court wrote, "The emergence of GenAI changed the course of this case." The court noted that while no witness at the liability trial in 2023 testified that GenAI products posed a near term threat to general search engines, GenAI was identified as a nascent competitive threat during the remedies proceeding, in the spring of 2025. The court concluded that GenAI is highly competitive.
(31:49):
And commentators have posited that the court's conclusion in that regard about GenAI led it to impose a lesser set of remedies on Google than otherwise would've happened. Certainly, less than the DOJ sought remedies including a bar on Google being able to compensate its distribution partners. DOJ also sought to force Google to divest its Chrome browser and ultimately, the court held that those remedies were unnecessary. The court also explained though that the remedies proceedings had morphed – instead of being focused only on promoting competition among general search engine providers, the remedies proceedings also became focused on ensuring that Google's dominance in search doesn't carry over into the GenAI space.
(32:35):
So, accordingly, the court ordered remedies encompass not just Google's general search engine, which was the subject of the liability phase, but also extend to its AI products. For example, just as the court’s order prevents Google from entering into exclusive contracts relating to the distribution of its general search engine, it also prevents Google from securing exclusivity for its GenAI products, including its Gemini app. The order also requires Google to make certain search index and user-generated data available to qualified competitors, and the remedies order contemplates that those qualified competitors will include GenAI developers.
(33:12):
It's been reported that Google's likely to appeal both of those decisions and DOJ may appeal the remedies decision. So, more to come as it relates to DOJ and Google. Another lawsuit with potential implications for the AI space is a recent case that X, formerly known as Twitter, and xAI filed in August of 2025 against Apple and OpenAI in the northern district of Texas. X is challenging Apple's integration of ChatGPT into its iPhone operating system. It alleges that Apple and OpenAI have entered into an exclusive arrangement that harms Chat GPT competitors like xAI's Grok.
(33:52):
X alleges that this arrangement furthers Apple's alleged monopolization of the market for smartphones, and OpenAI's alleged monopolization of the market for generative AI chatbots. Apple and OpenAI have moved to dismiss arguing that among other things, X has failed to plausibly allege that their agreement is exclusive. And now this case is interesting because it may shed some light on how courts will analyze AI partnerships. It also could provide an indication as to how markets for AI products will be defined. So, as we get closer to wrapping up today's episode, Antoni, can you tell us a little bit about how the European regulators are positioning themselves when it comes to AI?
Antoni Terra (34:34):
In general, what we are seeing is a more wait-and-see approach from regulators when it comes to the antitrust regulation of AI. There is a growing sense that while generative AI is rapidly reshaping, its potential harm from an antitrust perspective remains unclear. The EC, for example, has been saying that it is closely monitoring AI integration in existing services such as search engines and social networks to ensure code that the AI use in the EU advances in line with the law. That comes from a letter in late September from Henna Virkkunen, the European Commissioner for Tech.
(35:14):
And at the same time, what are the implications at the policy perspective or at the policy level of AI in relation to the DMA? Olivier Guersent, the former EC director general for competition was saying that it was too early to regulate AI in a definitive way and more broadly, along the same lines as we were saying, that there was no consensus in this regard as to what would be needed, or to what would be required. The EC nevertheless has reiterated that they will use all available tools, including the DMA, to ensure that generative AI remains fair, competitive and contestable.
(35:53):
And in that same letter that I was alluding to from Ms. Virkkunen, she reminds DMA gatekeepers of their obligations regarding how they process, combine and cross-use data between services and the platforms should enable the option for default services, including embedded AI to be changed. Iacovos in the UK, what's the talk in town from a policy perspective?
Iacovos Antoniou (36:21):
The CMA has also made progress on implementing its new digital markets regime introduced by the Digital Markets Competition and Consumers Act or DMCCA. The CMA recently issued its first strategic market status designation decisions under the DMCCA against, well firstly Google for its general search and search advertising services as well as against Google and app respectively for their mobile platforms.
(36:45):
These companies which are now designated as having strategic market status in the UK are subject to the new mandatory notification obligations that we mentioned earlier, which means while the CMA regime remains voluntary, in general, when it comes to these companies, there are certain transactions that will need to be reported to the CMA to the extent that they fulfill certain transaction value criteria that we mentioned earlier.
Michael H. Menitove (37:10):
Can you tell us some more about your practical takeaways, that should be on a to-do list right now for companies that are active in the AI space?
Iacovos Antoniou (37:18):
Competition enforcement in the AI space is here to stay and potentially become more prevalent. So, it's definitely a good time to zoom in on competition principles and understand the regulatory direction. In terms of mergers, although AI partnerships, acquihires, or non-horizontal mergers have not undergone in-depth investigation so far, the early decisions that we discussed earlier provide useful insights. Transactions in the AI space are also being monitored on an ongoing basis, including by national authorities, which will not shy away from using their national merger control rules to review deals of interest as Antoni spoke about earlier.
(37:54):
Other factors that reflect national interest in AI antitrust enforcement are recent merger control threshold modifications. For example, in the UK, we outlined the new hybrid threshold, which extends the jurisdictional scope to more AI partnerships.
Antoni Terra (38:10):
When it comes to AI, it is important that companies and advisers take a holistic approach as to what is required to be looking at, to be monitoring in terms of compliance as well from a merger control perspective, from a conduct perspective with regards to the competition law implications of this. Think again, Towercast and all the conduct examples that we've been mentioning, deals can be examined from the perspective of Article 101 and 102 of the Treaty of Functioning of the European Commission. So, again, this really reminds everyone of this encompassing approach that regulators are taking for AI and a bit more generally also in the digital and tech space.
(38:55):
And now, as you can see in my background, sunset in Brussels, we wrap it up for this edition of the “Fierce Competition” podcast, merger control, antitrust enforcement, adapting to the AI sector evolving considerations around this. We are here to keep you updated, so please stay tuned for future editions of this podcast. Thanks everyone very much for listening. See you all next time.
Since the recording of this podcast, there have been a couple of AI developments in early December with respect to abuse of dominance investigations by the European Commission that we definitely wanted to mention here. So first, the Commission opened an antitrust investigation into a recent policy announced by Meta whereby AI providers will not be able to communicate with customers using WhatsApp's business solutions when AI is the primary service that they offer. Given the ongoing proceedings regarding Meta AI and WhatsApp in Italy that we referred to before, in this podcast, the EC investigation covers the EEA except for Italy. And then second, in relation to Google's AI power search capabilities that we have also alluded to earlier in this podcast, the EC is now formally investigating whether those services are based on web publisher's content without appropriate compensation, and also whether those publishers can refuse such use of their content without losing access to Google search. Similarly, as part of this investigation, the Commission indicated that they're exploring the conditions of Google's use of video and other content uploaded on YouTube to train Google's GenAI models. We'll keep you posted as we said before. Thanks all very much.
Voiceover (39:25):
Thank you for joining us for today's episode of “Fierce Competition.” If you like what you're hearing, be sure to subscribe in your favorite podcast app so you don't miss any future conversations. Additional information about Skadden can be found at Skadden.com.
Listen here or subscribe via Apple Podcasts, Spotify, YouTube or anywhere else you listen to podcasts.
