AI adoption in downstream markets presents both opportunities and challenges for competition. On the one hand, AI systems, in particular generative and agentic models, can lower barriers to entry by reducing labour and operational costs, enabling product differentiation and supporting innovation. So far, these effects are most visible in sectors with high exposure to cognitive tasks, such as professional services, software development and customer support. Modular and “plug-and-play” AI tools may allow smaller firms to scale more efficiently, while fine-tuning capabilities can support domain-specific innovation.
On the other hand, the competitive impact of integrating AI systems in downstream markets appears to be highly context dependent. Structural advantages in data, compute, and integration capacity may allow incumbents to capture or hold on to disproportionate gains. Where access to models or data is restricted, or where interoperability is limited, downstream contestability may be reduced. The paper highlights that model access, rather than openness alone, is likely to be a key determinant of differentiation and innovation. Similarly, data access – both upstream and operational – can shape firms’ ability to adapt AI systems to their needs. These dynamics may reinforce existing market power or create new bottlenecks, particularly in vertically integrated AI stacks.
From an enforcement perspective, AI systems may amplify traditional concerns such as collusion, exclusionary conduct, and foreclosure, while also introducing new challenges around attribution, transparency, and strategic adaptation. Generative and agentic AI systems may complicate the assessment of effects and liability, potentially requiring of competition authorities to also consider issues such as data governance, model explainability and design choices. The paper outlines how the use of generative AI may amplify risks such as personalised and dynamic pricing or vertical leveraging, and discusses how enforcement, advocacy, and regulation can help address these concerns.
To ensure that AI-enabled markets remain open and competitive, a multi-pronged approach seems to be needed. Enforcement may have to adapt to the realities of autonomous and opaque systems, potentially requiring a shift towards effects-based reasoning or even the development of new investigative tools. Competition advocacy and market studies can help identify emerging risks and inform regulatory design. Sectoral co-operation will be essential to align standards, support fair access to key inputs, and prevent regulatory fragmentation. Ultimately, the competitive impact of AI will depend not only on the technology itself, but on the institutional frameworks that govern its deployment and use.
Looking ahead, given the fast pace of developments in the sector – many of the references that this paper relies on were publishing during the time of writing – further empirical research is needed to assess sector-specific competitive impacts, particularly in health, finance, logistics, creative industries and professional services, where AI may both expand demand and reshape labour pipelines (Above The Law, 2025[130]). For instance, AI image generation is increasingly a substitute for stock imagery (concept art, product mock-ups, backdrops) and an input to, or substitute for, creative ideation pipelines. In July 2025, the fashion brand, Guess, sent shock-waves through the media world when its advertising in the magazine Vogue, featured a model generated by AI (BBC, 2025[131]). No human featured in their advertising campaign. Quite apart from the questions this may raise on ethical or cultural grounds, this demonstrates the power and potential of GenAI to structurally alter the competition dynamics of a sector.
Emerging questions around agentic AI also require close monitoring, as these systems may both lower search costs and risk biased steering of consumer choices (LinkedIn/Ipsos, 2025[55]). Comparative work across jurisdictions, such as the JFTC’s 2025 mapping of competition authority initiatives, offers valuable insights into common challenges and approaches.
At international level, co-ordination will be essential to address concentration risks arising from the global scale of compute and foundation models. Ensuring open and contestable markets for AI will therefore require a mix of national enforcement, cross-sectoral regulatory safeguards, and international co-operation.