Europe and Britain are moving from retroactive antitrust enforcement to upfront rules for digital gatekeepers, while the United States relies on case-by-case enforcement; the resulting regulatory fragmentation creates uncertainty over whether tighter rules will curb platform power without stifling innovation, particularly around generative AI, data portability, and 'killer' acquisitions.

Main Finding

The paper argues that digital transformation has outpaced traditional antitrust tools, prompting a global shift toward proactive, ex-ante regulation (notably the EU’s DMA and the UK’s DMCC) to supplement or replace slower ex-post enforcement (the current predominant US approach). Key digital market features—multi‑sided platforms, strong network effects, data as a core asset, zero‑price markets, algorithmic pricing, and high switching costs—create entrenched market power and novel harms (often non‑price), requiring new regulatory remedies (data portability, interoperability, merger‑control reform) and international coordination. Emerging technologies, especially generative AI and foundation models, intensify these concerns.

Key Points

• Core characteristics of digital markets that challenge traditional antitrust:

  • Multi‑sided platforms with strong indirect network effects leading to “tippy” markets.
  • Data as a reinforcing competitive moat (more data → better services → more users).
  • Zero‑price markets where harms are often non‑price (privacy, quality, innovation).
  • Algorithmic pricing that can enable tacit algorithmic collusion hard to prove under existing law.
  • High switching costs and lock‑in (technical and social), increasing barriers to entry.

• Comparative regulatory responses:

  • EU — Digital Markets Act (DMA; in force Nov 2022): ex‑ante obligations for designated “gatekeepers” (7 designated firms as of 2023), rules on self‑preferencing, interoperability/data access, fines up to 10% of global turnover; early structural effects (e.g., third‑party app stores).
  • UK — DMCC Act (in force Jan 2025): hybrid model; Digital Markets Unit designates Strategic Market Status (SMS) and can impose tailored conduct requirements and tougher merger scrutiny (transaction‑value thresholds to catch “killer acquisitions”).
  • US — enforcement‑led approach (Sherman/FTC Acts): increased high‑profile litigation by DOJ/FTC but no overarching ex‑ante regime; slower, case‑by‑case outcomes and reliance on courts to define harms.

• Emerging and persistent challenges:

  • Generative AI / foundation models: building and scaling models requires huge compute/data, raising concentration risks and potential foreclosure of smaller AI firms.
  • Killer acquisitions: many acquisitions of nascent competitors evade notification thresholds; prospective, forward‑looking assessments are difficult.
  • Data portability and interoperability: desirable remedies but technically, legally, and security‑wise complex; risks to incentives for incumbent investment.
  • Transatlantic divergence: differing philosophies (ex‑ante vs ex‑post) increase compliance costs and risk fragmentation.

• Early evidence and trade‑offs:

  • Ex‑ante rules can produce quicker change against entrenched practices (e.g., self‑preferencing) but risk rigidity and regulatory lag if not adaptive.
  • Enforcement approaches retain flexibility but may be too slow for rapidly evolving markets.
  • Balancing competition protection with innovation incentives remains central.

Data & Methods

• Research design: qualitative, descriptive synthesis.
• Methods:
  • Systematic literature review across Scopus, Web of Science, Google Scholar, SSRN and competition authority sources using terms like “antitrust,” “platform economy,” “algorithmic collusion,” etc.
  • Comparative legal analysis of primary documents: DMA text, DMCC Act, DOJ/FTC complaints, legislative proposals, impact assessments, and international agency reports (OECD, EC).
  • Synthesis of emerging policy discussions and enforcement actions (e.g., DMA gatekeeper designations, Apple App Store fines, major merger cases).
• Outputs: narrative synthesis, comparative tables (e.g., jurisdictional frameworks, DMA gatekeeper services), timelines and illustrative figures.
• Limitations: qualitative approach, rapidly evolving policy landscape, jurisdictional differences that complicate generalization.

Implications for AI Economics

• Concentration risks in AI:

  • Foundation models and generative AI require large datasets, specialized talent, and cloud compute—resources concentrated among incumbents—altering market structure and raising monopoly/foreclosure risks.
  • Economic research should quantify how data and compute scale economies translate into entry barriers and market power for AI firms.

• Merger policy and killer acquisitions:

  • Economists should develop methods for prospective valuation of nascent AI rivals and models to detect likely killer acquisitions (including use of counterfactual market simulations).
  • Consider policy tools such as lower notification thresholds for tech incumbents, shifting burdens of proof, or mandatory review of small‑value but strategically relevant deals.

• Data, interoperability, and access:

  • Need for empirical metrics to measure “data advantage” and the competitive impact of data portability/interoperability mandates.
  • Design and evaluation of standardized data formats, safe APIs, and governance that balance competition and security/privacy concerns.

• Algorithmic behavior and collusion:

  • AI‑driven pricing and recommendation systems create new mechanisms for coordination and market distortion; economic models must incorporate algorithmic dynamics and learning behavior in oligopolies.
  • Develop experimental and field methods to detect tacit algorithmic collusion and evaluate remedies.

• Consumer welfare framework:

  • Traditional price‑centric welfare measures are insufficient for zero‑price AI services—research must operationalize non‑price harms (privacy loss, reduced choice/quality, innovation suppression).
  • Incorporate multi‑dimensional welfare metrics in empirical antitrust work and policy impact assessments.

• Policy design and international coordination:

  • AI economics research should inform calibrations of ex‑ante vs. ex‑post regimes: when predictable rules (ex‑ante) improve welfare vs. when flexible enforcement (ex‑post) better preserves innovation.
  • Cross‑border harmonization (standards for data portability, interoperability, merger review cooperation) is critical; empirical work can estimate costs of regulatory fragmentation and benefits of alignment.

• Research agenda suggestions:

  • Measuring the competitive impact of data accumulation and model quality on entry and pricing.
  • Simulation models for merger counterfactuals in fast‑moving AI markets.
  • Field studies on the impact of DMA/DMCC-style interventions on innovation, consumer outcomes, and market structure.
  • Methods to quantify non‑price harms and long‑run dynamism in digital/AI markets.

Overall, the paper underscores that AI economics must expand its toolset—new empirical measures, forward‑looking merger models, and frameworks for non‑price welfare—to guide effective competition policy in the AI‑driven digital economy.