Seven years after the landmark digital competition reviews, generative AI and a shifting geopolitical landscape are testing the frameworks behind them. This report addresses the research and policy questions raised at a workshop on AI, digital markets and competition.
Preface
It is seven years since a series of landmark reports on competition in digital markets was published: the Furman Review Unlocking Digital Competition in the United Kingdom (UK); the EU’s Competition Policy for the Digital Era; and in the United States (US) the Stigler Center’s Committee on Digital Platforms. In each of those jurisdictions the expert analysis paved the way for legislation or judicial action. The UK passed (in 2024) the Digital Markets, Competition and Consumers Act (DMCCA) 2024. In the European Union (EU) the Digital Markets Act (DMA) was in place by 2022. During the Biden Administration Lina Khan as Head of the Federal Trade Commission and Jonathan Kanter at the Department of Justice took a more active approach to enforcement.
Much has changed since 2019. Generative Artificial Intelligence (GenAI) has dramatically altered the technological context in several ways, including new entrants (Anthropic and OpenAI), but also in terms of capabilities that are likely to reshape market structures and business models. The economics of generative AI differ from platform economics: frontier models require vast upfront capital spending, involve large economies of scale and, while not characterised by the same network effects, operate in complex ecosystems. The technology is progressing extremely rapidly, with agentic systems widely seen as the next application likely to be adopted across the economy.
The political and geopolitical context has also changed significantly. Chinese start-ups have come close to matching the technical capabilities of US frontier models, but with a commitment to smaller open weight models that can more easily be adopted. The dominance of the US and China has led many other countries to consider both their need for some AI ‘sovereignty’ or control, and what shape their industrial policies for AI ought to take. Competition, industrial and trade policies have converged in ways that challenge the traditional separation of policy analysis and practice (e.g. Coyle 2025, Coyle & Dahmen 2026).
We therefore held this workshop to try to identify the gaps in research and policy posed by this changed context. The discussion was structured under three headings: business models and market power; competition policy in practice and enforcement; and the geopolitical context. This document sets out the policy and research questions that arose in the discussions. The agenda that emerged is certainly not comprehensive, but we believe it identifies some key challenges generative AI poses for competition and market economies.
Diane Coyle
Bennett Professor of Public Policy, Cambridge
