CONCEPT

AI Industrial Policy (Mazzucato Framework)

The deployment of public investment, regulatory frameworks, and institutional design to direct AI development toward public purposes rather than leaving direction to commercial logic alone.

AI industrial policy, in Mazzucato's framework, is the deliberate use of state capacity to shape the direction, distribution, and institutional architecture of the AI transition. It is not regulation in the conventional sense — rules that constrain private behavior — but the active construction of the conditions under which AI development serves public purposes. The framework integrates mission-oriented investment, conditionality on public support, windfall taxation, public AI infrastructure, and return-sharing mechanisms into a coordinated industrial strategy. Mazzucato's argument is that the choice is not between innovation and regulation but between undirected innovation (which the commercial logic of platform companies will direct toward the highest private returns) and directed innovation (which public investment and institutional design can direct toward applications that address social challenges).

In the AI Story

Hedcut illustration for AI Industrial Policy (Mazzucato Framework) — AI Industrial Policy (Mazzucato Framework)

Mazzucato has been explicit about the institutional mechanisms required. In her February 2025 essay with Tommaso Valletti, she argued that while AI could deliver profound benefits for all of society, it is likely to do the opposite if governments remain passive bystanders. Policymakers must step in now to foster a decentralized innovation ecosystem that serves the public good. The critical word is decentralized. Mazzucato is not arguing for state control of AI development but for state direction — the use of public investment and institutional design to ensure AI capability flows toward societal needs.

The framework has four integrated components. Mission-oriented public investment defines specific societal challenges (climate adaptation, health equity, educational access) and directs research resources toward them. Conditionality attaches enforceable requirements to public support of private firms — interoperability, data sharing, fair pricing, worker protections. Windfall taxation captures a share of returns from AI commercialization to fund further public investment and transition support. Public AI infrastructure provides alternatives to private platforms, ensuring that access to AI capability does not depend entirely on commercial decisions.

The components are interconnected. Missions identify what should be built. Conditionality shapes how private firms engage with public support. Windfall taxation captures resources for further investment. Public infrastructure provides alternatives to private monopoly. Implementing any one component in isolation produces partial solutions that the other components would reinforce.

The international dimension adds complexity. AI industrial policy implemented in a single jurisdiction risks regulatory arbitrage — companies relocating to jurisdictions with weaker frameworks. Effective implementation requires coordination: shared standards for conditionality, coordinated competition policy, mutual recognition of governance frameworks. The EU, UK, and multiple Latin American governments have begun implementing versions of this approach, but global coordination remains rudimentary.

The critique of undirected AI development Mazzucato offers is grounded in the documented concentration patterns of the current AI economy. The overwhelming majority of AI investment flows toward applications with high commercial returns. The applications that would produce the greatest social benefit — AI-assisted healthcare in regions with inadequate medical infrastructure, agricultural optimization for smallholder farmers, educational platforms for underserved students — receive a fraction of this investment. The technical capability for these applications exists; what is missing is the institutional framework that would direct investment toward them.

Origin

The AI industrial policy framework emerged from Mazzucato's application of the mission economy framework to AI-specific challenges. Her engagement intensified in 2024–2025 with the Project Syndicate essays, the UK AI Action Plan critique, and the Paris AI Action Summit participation.

The framework has been adopted in various forms by EU policy initiatives (the EU AI Act complemented by EU AI Factories, the InvestEU program's AI focus), UK Labour Party policy proposals, South African industrial strategy, and multiple Latin American national AI strategies.

Key Ideas

Not regulation but direction. The framework goes beyond constraining private behavior to actively shaping the conditions of AI development.

Four integrated components. Missions, conditionality, windfall taxation, public infrastructure — each necessary, none sufficient alone.

Decentralized innovation ecosystem. Not state control but state direction — pluralistic innovation with public-purpose orientation.

International coordination required. Unilateral implementation faces regulatory arbitrage; coordinated frameworks are essential.

Directed outperforms undirected. The DARPA, Apollo, and Human Genome precedents show mission-oriented public investment expanding rather than constraining innovation frontiers.

Debates & Critiques

Critics argue that AI industrial policy risks picking losers and producing state capture. Mazzucato's response is that the absence of industrial policy does not produce neutral outcomes — it produces outcomes shaped by the commercial logic of the most powerful firms, which is itself a form of direction that excludes public input.

Appears in the Orange Pill Cycle

Mariana Mazzucato — On AI