Part 2: Monopoly agreements under new antitrust guidelines Mainland China’s State Administration for Market Regulation (SAMR) issued the final Antitrust Guidelines for Pharmaceutical Sector (“Guidelines”) on 24 January 2025, marking a significant step in the country’s ongoing efforts to strengthen competition compliance in this vital industry. Built upon a draft released for feedback last August,11 the Guidelines apply to traditional Chinese medicine, chemical drugs and biological products. Hardcore restrictions: what you need to know? In line with Article 17(1)-(5) of the PRC Anti-Monopoly Law (AML), the Guidelines highlight five “hardcore” restrictions that are presumed to violate the AML. Overcoming this presumption is particularly difficult as parties must meet a very high burden of proof to justify any exemption. The five “hardcore” restrictions are: price fixing, output restrictions, market division, joint boycotts and limits on new technology development. Since the SAMR’s consolidated antitrust enforcement to Q1 2025, six cases of horizontal agreements in the pharmaceutical sector have been penalised – mainly for price fixing, output restrictions and market division.22 (1) Price fixing Price fixing is by far the most common violation. Of the six cases, five involved companies colluding to coordinate prices – commonly sales price and bidding quote – through formal agreements or meetings, and also via messaging apps or third-party intermediaries. In some instances, like the Glacial Acetic Acid API case (2018), there was no written agreement at all; instead, companies reached a mutual understanding to fix price through communications facilitated by a third party.33 The Guidelines highlight that collusion can now occur through online platforms, apps or algorithms.44 In response, Chinese competition authorities are increasingly leveraging AI technologies to monitor pricing trends and potential alignment in key sectors such as pharmaceuticals – sending a clear message to companies about the importance of maintaining robust compliance practices. (2) Output restrictions The Guidelines clearly flag limits on product quantity or type – such as production caps, sales restrictions or bans on external sales.55 In past cases, indirect compensation schemes were used to hide cartel activities. To disguise the arrangement, companies were found using financial schemes such as selling products at artificially low prices and then repurchasing them at higher prices, sometimes with the help of downstream partners.66 The introduction of the Two-Invoice System in 2016, which restricts distribution to a single authorised intermediary, has increased transparency and made it harder to use such compensation schemes in public hospital sales.77 However, risks can remain – especially in areas like private hospitals or pharmacies, where distribution chains remain more complex.   (3) Market division Dividing either sales markets or procurement markets poses serious antitrust risks. This includes splitting customers, territories, market shares or even suppliers of raw materials and packaging.88 Market sharing also represents the most common form of misconduct – seen in five out of the six cartel cases – and typically involves: Agreeing not to compete in certain regions or with certain customers Allocating sales quotas or market shares Using exclusive distribution agreements to carve up downstream market In the pharmaceutical sector, limited suppliers and frequent use of tenders make the market sharing scheme more likely – sometimes through third-party distributors. Any form of market division, whether formal or informal, direct or through intermediaries, is high-risk. The Guidelines also make it clear that facilitators can be held liable. Companies should closely review their distribution, procurement and tendering arrangements to ensure compliance. (4) Joint boycotts Joint boycotts – collectively refusing to supply to or purchase from a particular business – are considered a “hardcore” antitrust violation when used to exclude competitors.99 This includes not only direct refusals to deal, but also agreements blocking cartel participants from working with specific competitors. Horizontal boycotts are generally presumed illegal (unless in certain research and development (R&D) collaborations that meet specific conditions), but vertical boycotts may be more justifiable – for example, jointly sourcing only from suppliers with higher environmental standards. Although there are no recent pharmaceutical cases, a 2023 Supreme People’s Court (SPC) decision in another industry illustrates the risks. In this case, rice noodle manufacturers colluded to force intermediaries and retailers to buy exclusively from them, penalising those who sourced elsewhere. This led to the exclusion and eventual foreclosure of a competitor, and the court found this anticompetitive.1010 For pharmaceutical companies, the message is clear: collective actions that limit competitors’ access to the market can pose a major red flag, and should be carefully reviewed, even if they are part of broader collaborations. (5) Limits on new technology development The Guidelines recognise that restrictions on technology acquisition, drug development or formulation improvements are generally anti-competitive. However, in the IP-intensive pharmaceutical sector, it is not uncommon for competitors in R&D partnerships to limit independent research within agreed areas during the collaboration to focus resources and avoid duplication. Article 18 of the Guidelines allows exemptions for R&D collaborations that expand drug variety, improve safety and efficacy, shorten development cycles, lower costs or ensure supply during public health emergencies.1111 That said, these restrictions must be assessed case-by-case. Limits that go beyond the agreed scope or persist after it ends may raise antitrust concerns. Reverse payment agreements – a growing antitrust concern (1) What are reverse payment agreements? Reverse payment agreements occur when brand pharmaceutical companies pay generic manufacturers to delay market entry or drop patent challenges, avoiding court risk. These arrangements raise competition concerns by eliminating (potential) generic competition, allowing patent holders to maintain high prices. Essentially, they function by sharing the market. (2) Guidelines on reverse payment agreements The Guidelines recognise that reverse payments between existing or potential competitors, can be anti-competitive when unjustified. The assessment considers whether: Compensation significantly and unjustifiably exceeds the reasonable settlement costs. The agreement effectively extends the patent exclusivity or delays generic entry. This follows the SPC’s 2021 AstraZeneca v Jiangsu Aosaikang Pharma (the “AstraZeneca Case”) ruling, which call for case-by-case analysis.1212 (3) SPC’s ruling in the AstraZeneca Case The SPC clarified that reverse payments are not per se illegal. Instead, their legality depends on: The likelihood of the drug patent being invalidated without settlement. Whether the settlement unjustifiably extends the patent holder’s monopoly or delays generic entry. The rationale behind these factors is as follows: Strong patents (unlikely to be invalidated): Reverse payments have minimal competitive impact since the patent holder would likely maintain exclusivity anyway Weak patents (likely to be invalidated): Such agreements can unfairly extend market exclusivity and delay generic competition, raising competition concerns (4) Key takeaways Reverse payment agreements are under heightened scrutiny in Mainland China, particularly when weak patents involved. Even if framed as legitimate settlements, they may violate the AML if they unjustifiably delays generic entry. Pharmaceutical companies should thus: Carefully assess the patent strength before settling. Ensure payments reflect reasonable settlement costs, and not for market exclusion. Avoid clauses that extend beyond the patent’s legal protection or restrict generic competition more broadly than necessary. Vertical agreements (1) Resale price maintenance (RPM) RPM agreements are presumed illegal under the AML and Guidelines unless the operator can prove they do not have anti-competitive effects. Direct RPM: setting fixed or minimum resale prices through agreements or notices. Indirect RPM: controlling profit margins, discounts or rebates, often enforced with deterrent mechanisms, including penalties, incentives and monitoring scheme.1313 To overcome the presumption, companies must demonstrate no harm to competition within or between brands, pricing or market entry – a high evidentiary burden to meet in practice. Recent pharma enforcement highlights ongoing risk, especially for finished drug products sold through complex distribution networks. These are typically formalised in written agreements with compliance measures. Regulators are also watching for more sophisticated RPM tactics, such as algorithm-based price monitoring. So far, past RPM cases have all involved finished products sold, rather than Active Pharmaceutical Ingredients (APIs). This reflects key differences in market dynamics: Finished drugs: more vulnerable to RPM because they are sold through multi-layered distribution networks, creating price volatility that manufacturers often try to control. APIs: typically sold in business-to-business transactions with higher market barriers and less price transparency, making them more susceptible to cartel or abuse of dominance rather than RPM. There is currently no safe harbour for RPM, and any form – direct or indirect – can trigger antitrust liability by Q1 2025. As of 3 June 2025, SAMR is seeking public input on the possibility of introducing a high-threshold safe harbour. But for now, companies should continue to remain vigilant.1414 (2) Non-price vertical agreements The final Guidelines take a softer approach to non-price vertical agreements in the pharmaceutical sector. While earlier drafts proposed rules on non-price restrictions such as where and to whom distributors could sell, exclusive territories, and requiring certain products to be sold, these provisions were removed from the final version.1515 This aligns with current enforcement practice – so far, there have been no cases penalising standalone non-price vertical restrictions. However, important caveats remain: stricter scrutiny applies if these restrictions involve abuse of dominance, support price-fixing or RPM, or are clearly anti-competitive (such as blocking passive sales or cross-supply between distributors). Situations not constituting monopoly agreements The Guidelines clarify that pure agency agreements in the pharmaceutical sector fall outside antitrust regulation if: the agent does not take ownership of products, the agent does not bear sales risk, and the arrangement is not a disguised exclusive distribution or resale. Typical examples include: agents selling products with prices and other key terms set by the supplier, pharmaceutical companies bidding or negotiating prices in centralised procurement with agents selling at agreed prices, or agents providing support services like importation or invoicing without bearing commercial risks. This distinction is especially relevant in Mainland China’s centralised pharmaceutical procurement system, where government-led intermediaries organise tenders and negotiate prices but do not bear financial or supply risks and charge fixed commissions. These intermediaries are generally considered genuine agents under antitrust rules. However, under Mainland China’s Two-Invoice System, delivery companies often assume inventory, financial and commercial risks, making them more independent and subject to antitrust scrutiny. Companies should thus carefully evaluate their agency arrangements to ensure they meet the above-mentioned criteria and avoid unintended antitrust risks. Going forward The pharmaceutical sector remains a key focus for antitrust enforcement, and the Guidelines now offer clearer direction for companies to assess and manage antitrust risks in their operations. In the next chapter, we will examine notable enforcement practices from other major jurisdictions. These cross-border perspectives will help clarify complex issues around monopoly agreements and provide practical insights for companies aiming to stay off the regulators’ radar. As you read on, consider: How do major antitrust regulators approach reverse payment agreements? How do regulators assess R&D collaborations between pharmaceutical companies? When do price-related agreements raise red flags for antitrust regulators? To support your understanding, we also include an appendix with relevant case examples for each identified risk area, offering practical reference points as you navigate these issues. Appendix – Relevant Cases (2018 to Q1 2025) 1. Price fixing Table 1: Price fixing cases (end 2018 to Q1 2025) Please click on the superscript numbers 1616, 1717, 1818, 1919, 2020, 2121 for details of the corresponding remarks in the table. View the table in actual size 2. Output restriction Table 2: Output restriction cases (end 2018 to Q1 2025) Please click on the superscript number 2222 for details of the corresponding remark in the table. View the table in actual size 3. Market division   Table 3: Market division cases (end 2018 to Q1 2025) View the table in actual size 4. RPM practices Table 4: RPM cases (2018 to Q1 2025) Please click on the superscript numbers 2323, 2424, 2525 for details of the corresponding remarks in the table. View the table in actual size

Part 1: Mainland China enforcement action The pharmaceutical sector has long attracted intense scrutiny from competition regulators worldwide – and Mainland China is no exception. This was clearly underlined in March 2025 when the Shanghai Administration for Market Regulation (“Shanghai AMR”) imposed a record RMB 223.4 million in fines in a high-profile pharmaceutical cartel case (the “Decision”).11 Notably, this case also sets a precedent for individual accountability: the executive responsible for orchestrating the violation was personally fined RMB 500,000, marking the first time in Mainland China that an individual was held directly liable in a monopoly agreement case. This Decision therefore sends a strong message that Mainland Chinese competition authorities are resolute in addressing misconduct. It also offers critical lessons for businesses operating in the country’s pharmaceutical sector and beyond. 1. Key takeaways (1) Personal liability for cartel involvement PRC law allows fines up to RMB 1 million for individuals involved in monopoly agreements, including legal representatives, senior management, or other directly responsible persons.22 The Decision highlights that liability may arise where individuals: Propose or instruct anti-competitive agreements Draft key agreement terms Organise meetings to exchange sensitive information Oversee or facilitate implementation of collusive strategies Notably, liability is not limited to senior management – any individual with a leading and substantive role in a cartel’s formation or execution may face penalties. (2) Ringleaders and leniency Mainland China’s leniency policy offers strong incentives for cartel whistleblowers33: First applicant: Up to 80% fine reduction or complete exemption Second: 30-50% reduction Third: 20-30% reduction It marks a key distinction from Hong Kong, where ringleaders are barred from applying for a leniency marker and can only seek up to a 50% reduction through a cooperation agreement.44 However, Mainland China allows ringleaders to apply for leniency and receive up to an 80% discount, though full exemption is off the table. This creates a strong incentive for companies in a leadership role of a potential cartel to come forward and cooperate with authorities. In these situations, it is essential to seek legal advice early to assess leniency opportunities and determine optimal timing for applications. (3) Identifying cartel red flags Companies must watch for signs of cartel activity. Red flags include: any competitor proposals or discussions about coordinating prices, discounts or fees; dividing customers or sales territories; limiting output and bid-rigging; or exchanging sensitive business information. If approached: immediately and proactively refuse, document your rejection, and consider reporting to competition authorities. (4) Growing focus on illegal gains confiscation The Decision also imposed confiscation of illegal gains – an enforcement measure that was less common in past cartel cases. This reflects a stronger regulatory focus on financial recovery in antitrust enforcement. Notably, in March 2025 the identification of methods to calculate illegal gains was listed as a legislative priority, signalling its growing importance in antitrust enforcement.55 This follows release of the draft Illegal Gains Identification Method by the State Administration for Market Regulation (SAMR) in December 2021.66 Once implemented, it is anticipated to provide clearer guidelines for calculating illegal gains and establish a more standardised approach to quantifying antitrust violations. What this means for businesses: Companies should proactively assess potential financial exposure in antitrust investigations. Strengthening compliance programmes is key to reducing enforcement risk. Importantly, illegal gains are not eligible for reduction, even if the company cooperates during the investigation. 2. Looking ahead in this series The remainder of Part 1 delves into the landmark Shanghai AMR’s Decision, presenting a comprehensive analysis of how the cartel scheme unfolded – ultimately resulting in the substantial penalty. Part 2 will analyse the SAMR’s latest Antitrust Guidelines for the pharmaceutical sector, released in January 2025, offering insights to help companies navigate pharmaceutical agreements. Part 3 will examine global regulatory developments, highlighting emerging antitrust issues and approaches that may influence future enforcement trends. This series aims to provide a comprehensive perspective on the evolving landscape of antitrust enforcement in the pharmaceutical sector, equipping businesses with the knowledge to manage risk and maintain compliance. 3. Background of case Following an investigation initiated on 30 April 2024, the Shanghai AMR discovered that three pharmaceutical companies – Shanghai Xinyi, Henan Runhong and Chengdu Huixin (collectively, the “Parties”) – had colluded to fix prices and share markets for their sales of neostigmine methylsulfate injection (“NMI”) products in the Mainland China market.77 NMI is a critical anti-cholinesterase drug mainly used in anesthesiology. The Parties are leading suppliers of these NMI products, and considered to be direct competitors in this space. 4. Two monopoly agreements Shanghai AMR’s investigation established that two monopoly agreements (together, the “Monopoly Agreements”) operated from January 2020 to December 2023.   Table 1: Two monopoly agreements Source: Summarised from the Decision88 View the table in actual size 5. Price-fixing in Second Monopoly Agreement Before the emergence of their agreement, the Parties had varying pricing strategies across different provinces. However, from April 2020 onwards, they began working together to systematically raise both their (1) hospital listing/negotiated price, and (2) price to delivery company for NMI.99 Pricing for hospital listing/negotiated price From May 2020 to 2022, following a series of meetings in multiple provinces, the Parties gradually raised their NMI listing prices for public hospitals. By 2022, Shanghai Xinyi’s price had jumped from RMB 3.56-6.8 to RMB 71.5 per unit (2ml, 1mg) across 31 provinces. Similarly, Henan Runhong and Chengdu Huixin raised their prices to RMB 36 (from RMB 1.78–5.6) and RMB 35.8 (from RMB 1.68–3.68) per unit (1ml, 0.5mg), respectively. Pricing for delivery companies The Parties steadily raised their NMI prices for delivery companies, as illustrated in Figure 1 below. By late 2023, all of them had achieved the price increase: Shanghai Xinyi: Raised prices to RMB 60–71.1 from RMB 3.23–6.15 per unit (2ml, 1mg). Henan Runhong: Raised prices to RMB 30–34.56 from RMB 1.5–5.1 per unit (1ml, 0.5mg). Chengdu Huixin: Raised prices to RMB 30 from RMB 1.6–3.58 per unit (1ml, 0.5mg). Figure 1: Raising prices for delivery companies (Unit: RMB/Unit) Source: Summarised from the Decision1010 6. Market allocation in Second Monopoly Agreement Shanghai Xinyi and Chengdu Huixin sharing private hospital market In April 2020, Shanghai Xinyi became the exclusive distributor of Chengdu Huixin’s NMI, which prevented Chengdu Huixin from making direct sales, including to both public and private hospital channels. Instead, all distribution to private hospitals was supplied by Shanghai Xinyi. In November 2020, Chengdu Huixin ended the exclusive agreement with Shanghai Xinyi, and resumed independent sales, but most of its products still went to private hospitals, with only a small portion able to reach public hospitals. This arrangement significantly reduced Chengdu Huixin’s presence in the public hospital market and restricted competition in both sectors. Allocating public hospital market between Shanghai Xinyi and Henan Runhong Since April 2020, Shanghai Xinyi and Henan Runhong agreed to use the same provincial distributors to promote their NMI products. Although Henan Runhong partially deviated from this arrangement for promotion in some provinces in 2021, Shanghai Xinyi and Henan Runhong continued sharing the same provincial distributors in key regions like Guangdong Province until December 2023, helping to maintain a stable market share in public hospitals.   Figure 2: Timeline and process of Parties’ strategic market allocation Source: Summarised from the Decision1111   7. Personal liability of Shanghai Xinyi GM The general manager (“GM”) of Shanghai Xinyi’s Sales Agency Business Department played a pivotal role in executing and implementing the Monopoly Agreements, which ultimately resulted in his RMB 500,000 fine under the Decision.1212 How the GM drove Monopoly Agreements First Monopoly Agreement (January 2020) Instructed and facilitated the price-increasing scheme with Henan Runhong and Chengdu Huixin, and reported on the price coordination strategy at a joint meeting. Second Monopoly Agreement (March 2020) Proposed the nationwide agreement. Arranged compensation to Chengdu Huixin to ensure its cooperation. Increased Shanghai Xinyi’s listing pricing nationwide. Coordinated with Chengdu Huixin to share the private hospital market, and with Henan Runhong to share the public hospitals market. Actively worked to resolve conflicts and align the interests of the Parties. The GM’s active role in communication, negotiation and coordination among the Parties was key to the success of the Monopoly Agreements, which consequently led to serious legal repercussions. 8. Penalties The price-fixing and market allocation schemes were in this case classified as “hardcore” restrictions under Mainland China’s competition rules, resulting in direct liability. Justifying such conduct carries a high burden of proof and has a low likelihood of success. Undertakings found liable face serious legal consequences, including orders to cease illegal conduct, confiscation of illegal gains, and fines from 1% to 10% of their previous year’s sales revenue. Directly responsible individuals may also be fined up to RMB 1 million.1313 This case resulted in record total fines of RMB 223.4 million. The breakdown of fines and respective mitigating elements for each party are as follows:   Table 2: Penalties and mitigating factors Source: Summarised from the Decision1414 View the table in actual size   9. Going forward This case resulted in hefty fines and severe penalties to involved undertakings and a key individual, sending a clear warning to the pharmaceutical industry – and beyond – about the serious consequences of antitrust violations. In Part 2 of this series, we will examine the SAMR’s Antitrust Guidelines for the Pharmaceutical Sector, providing critical insights into compliance strategies and identifying potential risks in pharmaceutical agreements. Before we dive in, let’s consider a few important questions: What types of conduct typically constitute a direct violation of competition rules in the pharmaceutical sector? Are reverse payment agreements arising from patent disputes between pharmaceutical companies always anti-competitive? What factors are considered when assessing their risk under competition laws? How do Mainland China’s competition regulators approach vertical restrictions in agreements (which are common in pharmaceutical agreements between suppliers and distributors)? What situations are generally not considered monopoly agreements in the context of the pharmaceutical sector?

Artificial intelligence (AI) has become a defining focus of the technology industry and at the heart of this wave are foundation models (“FMs”) – large-scale AI models trained on vast amounts of data (such as text, images or code) to perform a wide range of tasks, such as GPT-4 and Claude. These models typically use advanced machine learning techniques, requiring significant computational resources to train. Access to these models is becoming crucial across industries, as they underpin a wide range of AI applications. A key input in FM development is the access to high-performance computing resources, typically delivered through cloud infrastructure. This reliance has driven the rise of vertical integration in the AI ecosystem.11 This legal update explores competition challenges arising from such AI partnerships, examining merger control issues and how strategic AI alliances are coming under the scrutiny of competition authorities. Figure 1: Overview of the supply chain of FM development (Source: Compiled from CMA FM Report) Vertical relationships in AI development One prominent form of vertical integration is the formation of strategic partnerships between cloud providers and FM developers to streamline both model development and deployment. These partnerships often include investments and/or equity acquisitions in FM developers, though specific terms and structures may differ. Below is a non-exhaustive list of notable partnerships between cloud providers and FM developers. Figure 2: Exemplary partnerships between cloud providers (e.g. AWS, Microsoft) and FM developers (e.g. OpenAI, Anthropic) (Source: CMA Cloud Infrastructure Services) 22 Microsoft and OpenAI partnership overview One of the most high-profile examples of strategic partnership is between Microsoft and OpenAI, which has attracted significant regulatory attention – including investigations by the European Commission (“EC”), the UK’s Competition and Markets Authority (“CMA”) and the German Bundeskartellamt.33 While these authorities have concluded that the partnership does not qualify as a merger or confer control that would trigger a formal merger review, their scrutiny highlights the potential merger control concerns such alliances may raise. Microsoft and OpenAI first entered into this strategic partnership in 2019, with Microsoft serving as both funding investor and provider of cloud computing and supercomputing capacity. In return, OpenAI utilises Microsoft’s infrastructure to develop its FMs technologies and grants exclusive licence to commercialise these technologies, including through Azure.44 Since the initial agreement, the partnership has evolved with Microsoft becoming OpenAI’s largest investor, contributing over US$13 billion between 2019 and 2024. The diagram below illustrates the collaboration relationship between Microsoft and OpenAI. Figure 3: Collaboration relationship under Microsoft/OpenAI partnership (Source: Compiled from CMA Decision) Regulatory probe into merger control issue The Microsoft-OpenAI partnership has granted Microsoft material influence over OpenAI, as Microsoft itself acknowledges.55 However, merger control regulations are only triggered when there’s a change of control or acquisition of de facto control. As the CMA notes, there is no bright line between material influence and de facto control. Assessing whether that threshold is crossed requires looking beyond formal agreements to the commercial realities of the relationship.66 In December 2023, the CMA launched an investigation into whether Microsoft’s partnership with OpenAI, or any changes to it, could have resulted in an increase in Microsoft’s influence on the level of de facto control. The CMA focused its assessment on three key areas of potential influence: Investment and corporate governance Provision of compute capacity Licensing and commercialisation of intellectual property Financial investment and corporate governance Microsoft kept limited governance rights over OpenAI – its veto rights are mainly restricted to typical financial investor protections. Microsoft does not have a right to appoint a director to either OpenAI, OpenAI GP LLC (the entity that manages OpenAI) or OpenAI Nonprofit (the entity that controls day-to-day operation of OpenAI). While Microsoft briefly held a non-voting observer seat on the OpenAI Nonprofit board after a November 2023 leadership crisis, this seat was relinquished in July 2024.77 The November 2023 crisis – when Sam Altman was dismissed briefly and quickly reinstated following staff revolt and Microsoft’s supportive intervention – highlighted Microsoft’s material influence. However, the CMA concluded that Microsoft’s influence was important, but not the decisive factor in Altman’s reinstatement.88 Since then, OpenAI has demonstrated more financial independence by successfully raising over US$6.6 billion from a Thrive Capital-led investment round in October 2024, and securing another US$40 billion from a Softbank-led round in early 2025.99 In December 2024, OpenAI Nonprofit announced plans to restructure as a Public Benefit Corporation (“PBC”) to attract more conventional investment. This move could further diversify OpenAI’s funding sources and reduce reliance on Microsoft, though the future size and terms of Microsoft’s stake in the PBC remain unclear.1010 Figure 4: Microsoft’s governance rights over OpenAI (Source: Compiled from CMA Decision) Compute capacity supply Compute is a critical input for developing FMs and FM-based services. An agreement to supply compute – especially on an exclusive basis – could amount to acquiring material influence or control, or contribute to such a finding. Microsoft supplies compute and supercomputer capacity to support OpenAI’s technology development. However, OpenAI determines its research direction and model development strategy independently. Until January 2025, Microsoft served as OpenAI’s exclusive compute capacity provider for all computing needs. Following an amendment to their agreement, Microsoft now only retains exclusive rights for OpenAI API workloads, while holding a right of first refusal for any other compute capacity OpenAI may require.1111 Figure 5 sets out each party’s role in terms of the provision of compute capacity, highlighting how these roles changed before and after January 2025. Figure 5: Microsoft-OpenAI’s role regarding compute capacity supply and relevant amendments (Source: Compiled from CMA Decision) Licensing and commercialisation of IP Microsoft holds an exclusive commercial licence to OpenAI’s FM IP, allowing it to integrate OpenAI technologies into its products and services, except for certain artificial general intelligence (“AGI”) technologies, over which OpenAI retains exclusive rights. Microsoft is also the exclusive distributor of commercial API access to OpenAI’s models via Microsoft Azure, through the Azure OpenAI Service for customers. The two companies closely collaborate on developing AI supercomputing infrastructure and maintain mutual revenue-sharing arrangements, both benefiting from the commercial use of OpenAI’s IP.1212 Figure 6: Microsoft’s IP rights and commercialisation rights under its OpenAI partnership (Source: Compiled from CMA Decision) Assessment on merger control issue While Microsoft is able to exert significant influence over OpenAI, the CMA does not find this influence amount to control, nor de facto control, over OpenAI’s operations or strategic direction. OpenAI Nonprofit or OpenAI actively seek investment from additional partners to diversify funding sources.1313 OpenAI also retains autonomy over its research agenda and model development. Notably, following amendments to their partnership agreements in January 2025, OpenAI gained greater flexibility in sourcing compute from third parties. Microsoft’s exclusivity in compute provision was meanwhile scaled back to a first refusal right, reducing OpenAI’s operational reliance on Microsoft. Microsoft holds exclusive rights to distribute OpenAI’s models via the Azure OpenAI Service, but OpenAI itself retains the ability to offer non-commercial API access. Crucially, OpenAI retains exclusive control over IP rights related to AGI technologies – often seen as the “crown jewels” of AI development. As such, while Microsoft remains an influential partner, the current structure of the partnership does not support a finding that Microsoft exercises control over OpenAI. Therefore, their collaboration does not fall under CMA’s merger review scope. Concluding implications AI partnerships are becoming a trend in the sector, blending Big Tech’s resources with the talent and innovation of AI startups.1414 These collaborations offer mutual benefits, but their complexity poses real challenges for traditional merger control and competition assessment. Unlike classic mergers, these partnerships – as shown in the Microsoft/OpenAI partnership – often do not involve direct ownership or board appointments by Big Tech investors. The depth and scope of these alliances are constantly evolving, and while Big Tech investors can exert material influence, the extent of that influence remains ambiguous. This uncertainty means there is no universal answer to whether an AI partnership will trigger merger control. Much depends not just on how the partnership is structured, but also on the jurisdiction in question. In many jurisdictions, a transaction may be considered a notifiable concentration if it results in a change of control.1515 Typically, “control” goes beyond simply influencing a company’s strategy — it involves owning or having rights over all or part of its assets, or holding powers that decisively affect its governance, voting, or decision-making.1616 In contractual arrangements without direct share acquisitions, factors such as board representation or the ability to influence key financial or operational policies can be relevant to indicate control.1717 However, the US takes a more flexible approach under Section 7 of the Clayton Act, focusing less on strict control, and more on whether a deal could substantially lessen competition or create a monopoly.1818 That said, even if control is not acquired, a deal that gives the acquirer or investor the ability to influence the target’s strategic or operational decisions can still fall within the scope of the Clayton Act. This approach may become increasingly attractive to regulators as they are proactively adapting their competition framework to address the fast-evolving digital landscape. Recent developments include the EU’s consultation on revising its Merger Guidelines and the UK CMA’s taking steps to evolve its merger control scheme. 1919 As AI partnerships are proliferating and attracting the attention of competition regulators, it is foreseeable that competition authorities may further refine their merger control frameworks to better capture such partnerships within their regulatory scope. At the same time, competition authorities have also made it clear that this type of strategic partnership, particularly those creating vertical integration, remains firmly on their radar. Beyond merger control, these partnerships are being closely monitored by competition authorities for potential competition risks. 2020 Given the rapid development of the sector and the proliferation of new cooperation models, there is a real opportunity for practitioners like us to help clients in navigating how competition rules may apply to their specific circumstances. Staying proactive and well-informed will be key as authorities continue to refine their approaches in response to the evolving dynamics of collaborations in the digital sector.