[By Soujanya Boxy]
The author is a student of National Law University, Odisha.
Introduction
There is a booming interest among tech giants worldwide to fuel their technological growth with the adoption of AI. In their drive to lead in the AI race, tech giants are pouring billions into AI start-ups, hiring their key employees and acquiring their valuable assets and expertise. However, global competition regulators’ participation in market studies and ongoing investigations into various AI mergers demonstrate their determination to tackle competition issues arising from the AI space.
In its recent ruling, the United Kingdom (UK)’s principal competition regulator, the Competition and Markets Authority (CMA) approved the Microsoft-Inflection AI merger, observing no realistic prospect of a substantial lessening of competition (SLC) as a result of horizontal unilateral effects. The ruling gathered newsworthy attention because it will likely have global implications for the AI market and market players seeking mergers with tech giants. Just a few days later, the European Commission (EC) too decided to terminate its investigation into this merger, citing insufficient jurisdictional scope.
The article analyses the CMA ruling, which has implications for competition in the AI landscape. This ruling could potentially encourage more diverse and collaborative AI development, boosting innovation.
Deep Dive into Tech Titans’ Quasi Mergers with AI Firms
Quasi-mergers are trending among merger arrangement options in the technology sector due to their unique characteristics and benefits. These kinds of mergers represent the middle ground between direct competition and takeovers. The key benefit is that the firms can join forces, without sacrificing independence. As per The Economist, these forms of partnerships prove valuable in the face of higher trade barriers, regulatory concerns, and high interest rates.
In the recent past, tech giants, notably Amazon, Microsoft and Google have been most engaged with quasi-acquisitions of some foundational model firms, like Adept, Inflection AI and Character AI. Some other AI firms being acquired by Microsoft, Google, Amazon, and Nvidia, include Mistral AI, DeepMind, Anthropic, Hugging Face.
The quest among the powerful seven- Apple, Alphabet, Amazon, Nvidia, Microsoft, Meta and Tesla to be at the forefront of AI development, is likely to spur technology dealmaking. Nvidia is a major player in the AI chip market, with its investments in five AI-related firms, as it disclosed in a regulatory filing early this year. One noteworthy investment was a US$675 million deal in Figure AI, an AI startup, which included Microsoft, making it the largest AI fundraising round of Q1. There has been a dramatic increase in spending on AI by tech giants, totalling $160 billion, in the first half of this year, highlighting the growing fervour among firms to strengthen their AI capacities. Besides external investments, these firms spend heavily on their own AI R&D. For instance: Microsoft’s $13 billion investment in OpenAI.
The current AI landscape provides competitive advantages to tech giants. It equally poses exit challenges for venture capitalists (VCs), making it difficult for them to realise returns on their investments. Tech giants have more than financial backing to offer like cloud credits business networks, and other resources that VCs may be unable to replicate. This reduced the pressure on AI startups to go public.
Considering the tech giants’ perspective, it’s pivotal to examine the reasons behind their large-scale AI spendings and the anticipated returns. Tech giant CEOs expressed that despite capital expenditure and uncertainty around returns, they strongly preferred overbuilding their AI capacities than risking underbuilding. According to them, AI demand is outpacing supply. I/O Fund further emphasised the primary risk of being not “early enough” to capitalise on AI trends. Another important reason is the effectiveness of quasi-acquisitions as an alternative to in-house innovation, which involves the risk of failure and first-mover challenges.
A Global Footprint of Competition in AI
Competition regulators in the United States (US) and the European Union (EU) have been actively engaging in investigations, workshops and other initiatives to determine potential competition risks across the AI ecosystem. The US, UK and EU competition enforcers are concerned about competition risks posed by AI. In particular, they noted the concentration of AI models, heavy reliance on already concentrated markets, such as cloud computing, and the control of key inputs by a handful of firms. They are wary of AI partnerships, as these might be used by large incumbents to entrench market power.
The Department of Justice (DOJ)’s Jonathan Kanter, highlighted concerns that acqui-hiring could enable tech giants to stifle competition by absorbing the expertise of smaller firms, without acquiring them outright. Furthermore, the DOJ and FTC have divided the regulatory responsibilities for AI regulation. The DOJ will oversee the conduct of a large chip manufacturer, and the FTC will investigate into anticompetitive conduct of major software firms.
The EC and national competition authorities in the EU have launched investigations into virtual worlds and generative AI. Their active participation in the regulatory drive is evidenced by the conclusion of a workshop, studies, and reports published to analyse competition concerns arising from the emerging AI market. The EC’s policy brief, ‘Competition in Generative AI and Virtual Worlds’, specified critical bottlenecks including data limitation, talent scarcity and hardware constraints. Other barriers are high switching costs, market concentration, facilitated by established ecosystems, network effects and economies of scale of tech giants.
Given the profound impact of AI on the competitive landscape for firms, it requires careful evaluation of market competitiveness to understand regulatory concerns. The Forbes’ sixth annual AI 50 identified the most promising privately-held AI firms. Data showed the AI market is highly competitive, having numerous firms developing innovative technologies. While OpenAI ($11.3 billion) and Anthropic ($7.7 billion) have gained considerable funding, other players like Character ai ($193 million), Adept ($415 million), and Figure AI ($754 million) are also making significant strides. Lower levels of funding for some firms do not necessarily indicate a competitive disadvantage. This dynamic market is attracting partnerships from firms like IBM and Salesforce, suggesting a strong demand for AI.
Overall, the AI Market is characterised by strong competition, with new entrants (OpenAI, Cohere, Anthropic) competing fiercely with tech giants. Similarly, in terms of hardware, Nvidia undoubtedly leads the GPU market, yet it faces stiff competition from both AMD and Intel. Google is also investing to expand its specialised chip production.
The FTC is particularly concerned that partnerships between big tech companies could give them an unfair advantage in the cloud computing market. The top three cloud services, which are closely linked to these partnerships, are Microsoft Azure, Amazon Web Services (AWS), and Google Cloud Platform (GCP). Market shares reveal that there is no monopoly with Amazon at 31%, Microsoft at 25% and Google at 11%. Smaller firms like IBM, Salesforce, and Oracle have market shares ranging from 2-4% but they are rapidly expanding and engaging in AI partnerships (IBM-Mistral). It is unlikely that these AI partnerships will dramatically increase tech giants’ market shares.
Characteristics of the AI Market and Partnerships
Despite the high costs and other risks, the AI market has promise for competition. There might be significant sunk costs, but the proprietary nature of AI technology and the potential for the resale value limit their impact, maintaining fair contestability in the market. Foreign competition and market dynamics further contribute to market contestability. The Forbes’ data on different firms, partnerships and investments clearly exhibits the thriving nature of the AI market.
Partnerships like Amazon-Anthropic can enhance innovation and competition by providing integrated solutions that lower costs and improve user efficiency. Interconnectedness resulting from partnerships helps maintain high performance across the value chain, and competitive pressure encourages firms to make comprehensive improvements.
Although vertical integration raises concerns about exclusionary conduct, its procompetitive benefits must be balanced with anticompetitive effects. Procompetitive benefits include innovation and efficiency, and its anticompetitive effects are market dominance and restricted access for competitors. Policymakers, rather than solely relying on the traditional measure for ‘openness’, should adopt a comprehensive definition that includes both system accessibility and AI license to promote the flow of information and innovation. Closed models can likewise provide user efficiency benefits and limit free-riding, incentivising innovation. Thus, a market with both closed and open-source models can be competitive, offering consumers a variety of choices.
CMA’s Ruling: Boost or a Barrier for AI Quasi-Mergers
Microsoft invested $650 million in Inflection AI, gaining access to its models through its Azure cloud computing platform, and acquiring key talent, including the firm’s co-founders. On July 16, 2024, the CMA investigation into this partnership was launched, following concerns about Microsoft’s non-exclusive licensing agreement and hiring of Inflection AI co-founders. Neither event implied common control or any change in their distinct corporate structures. Inflection AI’s shift to B2B licensing, aligned with market dynamics, and its buyback of equity reflected strategic realignment rather than anti-competitive conduct. Microsoft’s licensing agreement fostered competition by allowing Inflection AI’s model to be available on multiple platforms. Microsoft’s hiring was a normal occurrence in a competitive labour market and contributed to innovation and benefiting consumers with varied products and services. Therefore, the transaction did not warrant competition regulators’ attention in the first place.
The CMA’s investigation into Microsoft-Inflection AI ended favourably on September 4, 2024. Regardless of overlapping activities between the two firms, the CMA concluded no competition risks. The investigation focused on two products: foundational models (FMs) and consumer chatbots. The CMA noted that Inflection AI does not pose any competitive threat to its competitors in chatbots and its planned AI studio business for enterprise customers is still in its early stages of development. It observed that Inflection FM lacked distinctive features and faced fierce rivalry.
The CMA stressed the significance of ongoing developments in FM models. They recognised that expertise is crucial for the development in the field of chatbots and FMs. However, it cautioned that the mere aquiring of a team with relevant know-how, without any further assets, could fall within the scope of the investigation. This highlights potential partnerships that may face scrutiny.
The partnership is currently under the FTC scrutiny. Nevertheless, the CMA’ greenlight is a positive sign for not only Microsoft and Inflection AI but also for other tech firms seeking to maximise AI efficiency and returns. Overregulation could hinder competition among firms, reducing innovation and harming consumers. A careful assessment of economic factors and the impact of such partnerships on competition, consumers, innovation and the market is necessary.
Concluding Remarks
The CMA ruling provides a major boost for tech companies concerned about regulatory interventions hindering their pursuit of AI efficiency. In the face of the AI race, investors are likely to invest more fearlessly, benefiting both themselves and investees. Increased product innovation leads to rapid growth in sales, market valuation and employment. Smaller firms may be motivated to innovate more due to the potential for investment from tech giants.
The ruling sparks a debate on whether the competition regulators should aggressively investigate hypothetical competition risks. While market contestability and underlying factors must be considered, it is important to avoid mere reliance on traditional measures in the dynamic digital world. Each factor relating to competition and innovation must be independently addressed to effectively balance pro-competitive and anti-competitive effects.