May 5, 2026
By Vanguard Enterprise Intelligence Unit with the work of Daron Acemoglu, David Autor, Thomas Piketty, Mariana Mazzucato, and Erik Brynjolfsson.

Executive Summary

Artificial intelligence is not only a productivity story. It is an ownership story.

The public debate around AI often centers on whether the technology will eliminate jobs, improve productivity, or create new forms of work. Those questions are important, but they do not fully capture the distributional issue. The larger economic question is who captures the value AI creates.

If AI raises productivity across the economy, the gains may still be distributed unevenly. Capital owners may benefit through rising valuations. Large firms may benefit from scale, proprietary data, compute infrastructure, and customer networks. High-skill workers may benefit if AI complements their judgment and expands their output. Lower-skill workers, young workers, routine knowledge workers, and firms without access to advanced tools may see fewer gains or greater insecurity.

This is the inequality problem of the AI age. A technology can be economically transformative while socially destabilizing if its returns are concentrated.

The issue is not theoretical. AI development is already capital-intensive. The leading firms require advanced chips, large-scale cloud infrastructure, proprietary models, data-center capacity, specialized talent, and access to enormous datasets. These inputs are expensive and difficult to replicate. As a result, AI may strengthen the advantage of companies that already have capital, data, distribution, and computing power.

The central question for leaders is therefore not whether AI should be adopted. It will be. The question is whether adoption can be structured to create broader economic participation rather than narrower economic concentration.

The New Inequality Mechanism

AI affects inequality through several channels at once.

The first channel is labor. AI changes which skills are rewarded, which tasks are automated, and which workers become more productive. In some roles, AI may raise the productivity of less experienced workers by giving them access to drafting, analysis, coding, research, or decision-support tools. In other roles, it may increase the premium for experts who can use AI outputs more effectively than others. The worker who knows how to question, edit, validate, and apply AI-generated information may become significantly more valuable than the worker who merely accepts it.

The second channel is capital. AI is built on capital-heavy infrastructure. The companies that own data centers, chips, models, cloud platforms, software ecosystems, and distribution channels may capture a disproportionate share of the gains. If AI productivity shows up primarily as higher profits, higher margins, and higher equity valuations, wealth will flow first to shareholders and founders.

The third channel is data. Firms with proprietary customer data, transaction histories, enterprise workflows, search behavior, user networks, financial records, logistics data, or medical datasets may train, fine-tune, and deploy AI systems more effectively than competitors. Data advantage can become market advantage.

The fourth channel is market structure. AI may reinforce winner-take-most dynamics. Large firms can spread AI investment across huge user bases, integrate tools into existing platforms, acquire emerging competitors, and negotiate better compute access. Smaller firms may adopt AI tools, but not necessarily own the infrastructure that determines long-term value capture.

The fifth channel is geography. Regions with strong technology ecosystems, power infrastructure, research universities, venture capital, skilled labor, and cloud connectivity may gain faster. Regions without those assets may fall behind.

These channels make AI inequality different from ordinary wage inequality. It is not only about workers earning different wages. It is about who owns the productive infrastructure of the next economy.

Skill-Biased Change Is Becoming AI-Biased Change

Economic history shows that technological change often rewards workers whose skills complement the new technology. Computers increased the value of analytical, managerial, and technical work. Automation reduced demand for certain routine jobs. The internet created new markets but also concentrated returns among platform companies.

AI may deepen this pattern, but in a more complicated way.

Generative AI can perform parts of work previously associated with educated professionals: drafting, summarizing, coding, financial analysis, legal research, market research, translation, customer service, and administrative coordination. This means AI is not limited to factory automation or low-skill task replacement. It reaches into white-collar work.

This creates two competing possibilities.

In the optimistic scenario, AI democratizes expertise. Workers with less experience gain access to tools that help them produce better work. Small firms gain capabilities once limited to large firms. Education becomes more personalized. Entrepreneurs build faster. Productivity rises broadly.

In the pessimistic scenario, AI concentrates advantage. High-skill workers use AI more effectively, large firms deploy it at scale, platform owners capture the economic rents, and lower-skill workers face displacement or wage pressure. The technology becomes a multiplier for existing inequality.

Both dynamics can occur at the same time. AI can reduce skill gaps inside a narrow task while increasing inequality across firms, regions, and asset owners. A junior employee may write a better memo with AI, but the larger economic gain may accrue to the firm that owns the workflow, the software provider that owns the tool, and the investors who own the equity.

This is the paradox of AI inequality: the technology may make individuals more capable while making economic rewards more concentrated.

Data Monopolies and the Infrastructure of Advantage

AI competition is increasingly shaped by data and infrastructure.

The leading firms are not only competing on algorithms. They are competing on access to compute, proprietary data, talent, distribution, capital, and cloud ecosystems. These advantages reinforce one another. A firm with more users generates more data. More data improves products. Better products attract more users. More users attract more developers, customers, and partners. The cycle strengthens market position.

This is why AI may increase skepticism toward traditional capitalism. If the public sees AI-driven gains flowing mainly to a small group of technology companies, investors, executives, and highly compensated specialists, the social legitimacy of the system weakens. The promise of innovation begins to look like another mechanism for asset concentration.

The public concern is not only job loss. It is exclusion from upside.

A worker may continue to have a job but see little improvement in wages while the market value of AI firms rises dramatically. A small business may use AI tools but remain dependent on a platform that controls pricing, data access, and distribution. A region may host data centers but receive limited broad-based prosperity if ownership, high-wage jobs, and intellectual property are located elsewhere.

The ownership structure of AI matters because it determines who participates in the gains.

Policy Reckonings

AI is likely to intensify policy debates around labor, competition, taxation, education, data rights, and social insurance.

The first policy area is workforce transition. Governments will face pressure to support reskilling, apprenticeships, lifelong learning, wage insurance, career mobility, and stronger labor-market data. Traditional education systems may not adapt quickly enough to meet the pace of AI-driven change.

The second area is competition policy. Regulators will examine whether dominant firms use data, cloud infrastructure, acquisitions, bundling, or platform control to entrench their position. The policy challenge is difficult: overly restrictive regulation may slow innovation, but weak regulation may allow AI markets to consolidate too quickly.

The third area is data governance. Policymakers will debate who owns data, how it can be used, whether individuals and firms should have greater rights over data portability, and how to protect privacy without freezing innovation.

The fourth area is taxation and redistribution. If AI raises returns to capital more than labor, governments may revisit corporate taxation, capital gains treatment, profit-sharing incentives, worker ownership models, or public investment mechanisms.

The fifth area is public infrastructure. Broad AI access may depend on investments in education, broadband, compute access, energy systems, public-sector AI tools, and regional innovation capacity.

The central policy challenge is balance. Societies need innovation, but innovation without participation creates backlash. AI policy will therefore increasingly focus not only on safety and privacy, but on distribution.

The Corporate Role

Corporations cannot wait for public policy to solve the distributional problem. They are central actors in how AI affects inequality.

Companies decide whether AI is used primarily to cut labor costs or to augment employees. They decide whether productivity gains are reinvested in wages, training, service quality, innovation, or shareholder returns. They decide whether workers receive the tools and training needed to benefit from AI. They decide whether AI governance includes fairness, transparency, and accountability.

A narrow corporate approach treats AI as a margin-expansion tool. This may produce short-term efficiency, but it can also damage morale, reduce trust, and expose the company to reputational and regulatory risk.

A broader approach treats AI as an organizational capability. The company uses AI to improve productivity while redesigning roles, training workers, raising decision quality, and creating new forms of customer value. The gains are still economic, but they are less extractive.

The strongest companies will not frame inclusion as charity. They will frame it as resilience. Workers who trust AI adoption are more likely to use the tools effectively. Customers who believe technology is being used responsibly are more likely to trust the brand. Regulators are less likely to intervene harshly when companies demonstrate credible governance. Investors benefit from durable growth rather than fragile extraction.

Inclusive AI is not only a social objective. It is a strategic risk-management principle.

Leadership Interviews: What Executives Are Likely to Say

Across sectors, executives are likely to express a common tension.

Technology leaders will emphasize speed. They will argue that firms must adopt AI aggressively or risk falling behind. Their concern is competitiveness.

CHROs will emphasize workforce trust. They will warn that adoption without training and role clarity creates anxiety, resistance, and attrition.

CFOs will emphasize return on investment. They will want productivity metrics, cost savings, margin improvement, and capital discipline.

General counsel will emphasize governance. They will focus on bias, privacy, intellectual property, labor rules, and regulatory exposure.

CEOs will face the integration problem. They must decide whether AI becomes a tool for narrow efficiency or a platform for broader transformation.

The leadership issue is that each perspective is valid but incomplete. AI strategy requires all of them. The firm that moves too slowly loses competitiveness. The firm that moves too quickly without governance creates risk. The firm that cuts too aggressively loses trust. The firm that cannot measure impact wastes capital.

The CEO’s role is to convert AI from a technology program into an institutional strategy.

A Framework for Inclusive AI Growth

Companies should evaluate AI adoption through a seven-part inclusive growth framework.

1. Map Value Capture

Leaders should identify where AI-generated value is expected to appear: lower costs, higher revenue, faster cycle time, better quality, reduced risk, improved customer experience, or new products.

The key question is: Who captures the value created by AI?

2. Segment Workforce Exposure

Companies should identify which roles are augmented, transformed, reduced, or newly created by AI. This should be done at the task level, not only the job-title level.

The key question is: Which workers need protection, training, or transition support?

3. Invest in AI Literacy

AI literacy should not be limited to technical teams. Employees across functions need to understand safe use, prompt quality, output validation, privacy, bias, and escalation.

The key question is: Are workers equipped to benefit from the tools, or only measured against them?

4. Redesign Roles Before Reducing Headcount

Productivity gains should be used to redesign work before defaulting to layoffs. Some roles will disappear, but many can evolve toward higher-value tasks.

The key question is: What new work becomes possible when AI reduces manual work?

5. Share Gains Credibly

Companies should consider gain-sharing, bonuses, training investments, internal mobility pathways, employee equity, or wage progression tied to productivity improvements.

The key question is: Do employees see any upside from AI adoption?

6. Protect Competitive Fairness

Companies with platform power should avoid using AI to lock in customers, exploit data asymmetries, or eliminate fair competition. Durable advantage should come from better value creation, not abusive control.

The key question is: Are we building an ecosystem or extracting from one?

7. Report Progress Transparently

Companies should communicate how AI affects work, governance, productivity, and risk. Stakeholders will increasingly expect credible disclosure.

The key question is: Can we explain our AI strategy in a way that builds trust?

The Inclusive Growth Model

Inclusive AI growth does not mean slowing innovation. It means designing innovation so that economic gains do not become politically and socially unstable.

An inclusive growth model has four components.

The first is access. Workers, small firms, schools, and public institutions need access to useful AI tools, not only large corporations.

The second is capability. Access without training does not create inclusion. People need the skills to use AI effectively and safely.

The third is participation. Workers and communities should have pathways to benefit from AI-driven growth through wages, ownership, career mobility, entrepreneurship, or local investment.

The fourth is accountability. Firms and policymakers should measure not only adoption and productivity, but distributional outcomes.

This model recognizes that AI will create winners. The issue is whether it also creates enough pathways for others to participate.

The Risk to Capitalism’s Social License

Capitalism depends on more than efficiency. It depends on legitimacy.

If AI produces dramatic gains for a narrow segment of society while many workers experience insecurity, stagnant wages, or reduced mobility, public skepticism will intensify. Calls for regulation, taxation, labor protection, antitrust intervention, and public ownership of AI infrastructure may grow. This is not an abstract political risk. It is a predictable response when technological change appears to privatize gains and socialize disruption.

Executives should take this seriously. The social license for AI will depend on whether people believe the technology improves their lives, not only corporate margins.

The most successful companies will understand that legitimacy is an economic asset. Trust reduces resistance. Trust improves adoption. Trust protects brands. Trust gives firms more freedom to innovate.

In the AI age, corporate responsibility will be judged less by statements and more by distributional choices.

Innovation With Shared Returns

AI may become one of the most important engines of economic growth in the coming decade. But growth alone will not settle the inequality question. The distribution of that growth will determine whether AI strengthens or weakens public confidence in markets.

The risk is clear. AI could accelerate wealth concentration through capital ownership, data monopolies, platform power, skill-biased gains, and uneven access to infrastructure. It could widen the gap between firms that own AI systems and workers who are managed by them. It could deepen the divide between regions with technology ecosystems and regions without them.

The opportunity is also clear. AI could expand productivity, improve services, support small businesses, personalize education, enhance healthcare, reduce administrative burden, and create new forms of work. But that outcome will require deliberate design.

For leaders, the imperative is to move beyond the question of adoption. The more strategic question is participation.

Who benefits from AI? Who is displaced? Who is trained? Who owns the upside? Who governs the risks? Who gets access to the tools? Who is left behind?

The companies and economies that answer these questions well will be better positioned to sustain both innovation and legitimacy.

AI will test whether modern capitalism can still convert technological progress into broad-based opportunity. That test has already begun.