Research report · AI political economy · Distribution design

The AI Dividend

Who gets rich when everyone gets faster?

Central thesis: AI productivity does not automatically become middle-class income. The political economy fight is over claims on output: wages, profits, equity, lower prices, taxes, public return rights, and citizen capital.

A practical research report on productivity capture, labor bargaining power, and the new AI-enabled middle class.

AIWorkProductivityLabor bargainingOwnershipMiddle class
Global job exposure ~40% IMF estimate for AI exposure
Advanced economy exposure ~60% IMF estimate; higher exposure than emerging markets
U.S. data center electricity 4.4% DOE/LBNL estimate for 2023
Capital distribution scale $1T+ S&P 500 buybacks over 12 months ending Sept. 2025

Abstract

AI productivity is a distribution problem before it is a cash-transfer problem.

Research question

When AI raises output, which institutions determine whether the gain becomes wages, capital income, public revenue, lower prices, or direct citizen ownership?

Evidence base

The report synthesizes AI task-exposure research, labor-share cautions, market-concentration evidence, infrastructure constraints, household ownership data, and owner-payout channels.

Principal finding

AI productivity does not automatically become middle-class income. The decisive variable is the claim structure around compute, data, platforms, labor scarcity, customer access, and public return rights.

Policy implication

A broad AI dividend requires a portfolio of allocation rules: stronger wage claims, worker ownership, small-business leverage, competition policy, public return rights, and citizen capital accounts.

  • AI can raise productivity while weakening labor bargaining power if ownership and surplus-sharing rules do not change.
  • The first distribution shock may appear through task thinning, wage pressure, and reduced outside options rather than mass unemployment.
  • The largest default winners are owners of compute, cloud, frontier AI models, dominant platforms, data-center infrastructure, and broad equity portfolios.
  • A new AI-enabled middle class is possible if workers, creators, small firms, and citizens can convert AI productivity into durable claims: higher wages, profits, equity, dividends, and portable capital accounts.
  • Policy design should compare multiple pathways: profit sharing, worker ownership, wage subsidies, antitrust, public return rights, universal capital accounts, citizen wealth funds, UBI/NIT, and data-rights institutions.
Author Kevin L. Michel
Field Political economy of AI
Method Distribution-channel model + scenario analysis
Core unit Claims on productivity gains
Scenario status Illustrative indices, not forecasts
Access Public abstract + member report preview

Data and evidence base

Exposure, ownership, market structure, and infrastructure are distinct evidence classes.

The report separates task-exposure estimates from direct infrastructure measures, owner-payout proxies, market-structure indicators, and policy inferences. The categories should not be read as one scale.

Analytical frame

Treat AI productivity as a surplus-routing problem. Each gain can move through labor income, firm margins, capital payouts, consumer prices, public revenue, or citizen capital.

Scenario method

The 2026-2040 pathways use a normalized distribution index where 2026 equals 100. They compare institutional directions and are not macroeconomic forecasts.

Evidence hierarchy

Direct measures, task-exposure estimates, infrastructure indicators, market-structure signals, owner-payout proxies, and policy inferences are marked separately.

Limit

AI adoption, worker bargaining power, firm strategy, competition, and policy design are moving targets. The report treats uncertainty as part of the model.

Method and limitations

The analysis tracks claims on AI-generated surplus.

The central method is to treat AI productivity as a routing problem. A gain becomes socially broad only when institutional rules move it from raw output into durable household claims, public claims, consumer benefit, or small-business capacity.

Exposure is not destiny

The report treats AI exposure as a measure of task contact with AI, not a deterministic prediction of job loss.

Scenarios are illustrative

The 2026-2040 scenarios show policy pathways and relative dynamics. They are not macro forecasts and should not be read as precise projections.

Policy incidence matters

Taxes, levies, warrants, and public return rights can be shifted through prices, wages, or investment unless designed around market power and mobility.

Broader than one ideology

The report compares labor, capital, competition, public ownership, tax-credit, and cash-transfer pathways instead of treating one instrument as sufficient.

Capture map

AI productivity can route to six different claims.

The same productivity gain can become higher wages, higher margins, shareholder payouts, lower prices, public revenue, or direct citizen capital. The distribution is not automatic; it is designed by markets, contracts, law, and bargaining power.

Workers Firms Shareholders Consumers Public Citizens
Workers

Wages, bonuses, bargaining power

Firms

Margins, productivity, workflow control

Shareholders

Buybacks, dividends, capital gains

Consumers

Lower prices, better services

Public

Taxes, warrants, procurement upside

Citizens

Dividends, capital accounts, public funds

AI Dividend productivity capture map A central AI productivity node routes gains to workers, firms, shareholders, consumers, the public, and citizens. PRODUCTIVITY AI gain who gets the claim? Workers claim Firms claim Shareholders claim Consumers claim Public claim Citizens claim

Figure note: conceptual surplus-routing model. The channels identify possible claims; visual position, ordering, and compact cards do not encode measured shares.

Workers Wages, bonuses, bargaining power

Requires scarcity, voice, and sharing rules.

Firms Margins, productivity, workflow control

Default capture channel inside adoption.

Shareholders Buybacks, dividends, capital gains

Gains compound for households that already own assets.

Consumers Lower prices, better services

Depends on competition and pass-through.

Public Taxes, warrants, procurement upside

Must be designed before rents harden.

Citizens Dividends, capital accounts, public funds

Turns growth into visible household claims.

Member report

The full AI Dividend report is member access.

Join the Library to unlock the executive summary, evidence dashboard, capture table, bargaining model, scenario matrix, AI-enabled middle-class ladder, policy pathway matrix, named actor map, source notes, and implementation cautions.

Member research

Unlock the full AI Dividend report with Basic.

Trace where AI productivity goes: wages, margins, shareholder returns, lower prices, public revenue, or citizen capital.

  • Capture map
  • Bargaining model
  • Scenario matrix
  • Policy pathways
  • Evidence notes

Basic unlocks the full online report, full books, and all interactive research reports.

Optional PDF copy is separate. Basic unlocks the online report now.