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

Capitalism’s next legitimacy crisis may not come from stagnation. It may come from abundance that is too narrowly distributed.

Artificial intelligence promises a new era of productivity, growth, scientific discovery, and operational intelligence. It can improve healthcare, accelerate software development, personalize education, optimize logistics, strengthen research, and help companies make better decisions. If deployed well, AI could expand the productive capacity of the global economy. But the central question is not only how much value AI creates. It is who captures that value.

This is where the future of capitalism becomes unstable. If AI’s gains flow mainly to a small group of technology firms, capital owners, highly skilled workers, and regions with compute infrastructure, while millions of workers experience job insecurity, wage pressure, and declining mobility, the market system will face a deeper political problem. People may not reject innovation. But they will reject an economic order that asks them to absorb disruption while others capture the upside.

The AI-driven wealth divide is not only a labor-market issue. It is a legitimacy issue. It touches ownership, bargaining power, regional inequality, firm concentration, education access, data control, infrastructure, and public trust. It asks whether capitalism can adapt its institutions quickly enough to prevent a new productivity revolution from becoming a new era of resentment.

The answer will not come from slowing AI alone. It will come from redesigning the economic systems around AI so that productivity becomes more broadly useful, more broadly accessible, and more broadly trusted.

The New Inequality Mechanism

Previous waves of technology often increased inequality by favoring higher-skilled workers over lower-skilled workers. AI is more complicated. It can assist less experienced workers, compress skill gaps in some tasks, and make expert-level tools more widely available. But it can also concentrate value in the hands of those who control data, models, chips, distribution, cloud platforms, capital, and customer relationships.

This creates a paradox. AI may democratize certain capabilities while concentrating economic rewards.

A junior employee can use AI to draft stronger analysis. A small business can generate marketing content that once required an agency. A local manufacturer can improve forecasting. A student can access tutoring. A doctor can receive diagnostic support. These are real democratizing effects.

But the largest economic gains may still flow to firms that own the infrastructure, platforms, models, and capital required to scale AI. The more AI becomes embedded in enterprise systems, the more advantage may accrue to organizations that already have data, distribution, talent, capital, and compute. This is how a productivity tool can also become a concentration engine.

The risk is not simply that AI replaces jobs. It is that AI reorganizes bargaining power. Workers may become more productive, but if they do not share in the gains, wages may not rise proportionally. Smaller firms may gain access to AI tools, but if dominant platforms control the terms of access, market power may deepen. Regions may adopt AI applications, but if data centers, capital investment, and talent cluster elsewhere, regional divides may widen.

Capitalism can survive inequality. It has done so repeatedly. But it struggles when people believe inequality reflects structural exclusion rather than contribution.

Labor Market Polarization

The labor-market effects of AI are still emerging, and early evidence remains mixed. Some research suggests that AI can improve productivity, especially for less experienced workers in certain tasks. Other evidence points to pressure on entry-level hiring, task de-skilling, and growing anxiety among younger workers. The danger is not that every job disappears. The danger is that career ladders become less reliable.

Entry-level work has always played a developmental role. Junior analysts learn by preparing drafts. Young lawyers learn through document review. New coders learn by writing and debugging routine code. Customer-service workers learn products and customers through repeated interaction. If AI absorbs too much of this early work without new apprenticeship models, companies may weaken the pipeline through which judgment is built.

This creates a subtle form of inequality. Senior workers and capital owners may benefit from AI leverage, while younger workers face fewer pathways into skilled careers. Workers with strong networks, elite credentials, and access to AI-intensive firms may advance faster. Workers without those advantages may find that the first rung of the ladder has moved.

AI may also polarize work inside companies. High-trust, high-skill employees may use AI to expand their influence. Lower-autonomy employees may experience AI as monitoring, speed-up, or replacement pressure. The same technology can feel empowering at the top and disciplinary at the bottom.

Leaders should not assume that productivity gains automatically translate into worker benefit. Productivity is a measure of output. Fairness depends on distribution, mobility, dignity, and future opportunity.

The Ownership Question

The most important inequality question may be ownership. Who owns the assets that AI makes more valuable?

AI rewards ownership of complementary assets: compute infrastructure, proprietary data, cloud platforms, advanced chips, customer networks, enterprise software ecosystems, distribution channels, and intellectual property. Workers may use AI, but owners of these assets often capture the residual gains. That is why AI could widen wealth inequality even if it improves productivity across many occupations.

This is not a new problem. Capitalism has always rewarded ownership. But AI intensifies the issue because the technology can scale globally, substitute for some labor tasks, and strengthen network effects. Once a firm builds a dominant AI-enabled platform, marginal costs may fall while returns concentrate.

A capitalism that concentrates AI wealth too narrowly will face backlash. This backlash may appear as antitrust pressure, labor activism, tax reform, data-rights debates, restrictions on automation, local resistance to data centers, or broader distrust of technology companies. The public may not object to AI itself. It may object to an economy in which AI’s benefits are privatized while its disruptions are socialized.

The ownership question therefore belongs in boardrooms, not only in policy debates. Companies should ask whether workers participate in AI-enabled productivity gains. They should consider profit-sharing, employee ownership, skills investment, internal mobility, and broader access to value creation. These are not charitable gestures. They are ways to strengthen the social license for technological transformation.

Firm Concentration and Market Power

AI may also widen inequality between firms. Large companies are better positioned to invest in data infrastructure, integrate AI into workflows, hire technical talent, negotiate cloud contracts, absorb experimentation costs, and spread fixed costs across many customers. Smaller firms may access AI tools, but they may remain dependent on platforms owned by larger players.

This creates a productivity dispersion problem. If top firms use AI to pull further ahead while smaller firms remain tool-users rather than system-builders, markets may become less competitive. The strongest companies will have better data, better models, better talent, better customer access, and better capital. Over time, this can turn AI advantage into market power.

There is also a risk that AI becomes a rent-extraction layer. If a few infrastructure providers control access to models, compute, or enterprise platforms, they may capture value from thousands of downstream firms. AI could then become less like a general-purpose technology available on open terms and more like a toll road controlled by the few.

This is why competition policy, interoperability, open-source ecosystems, data portability, and procurement discipline matter. Inclusive capitalism in the AI era depends not only on helping workers adapt, but on ensuring that firms can compete. If AI consolidates markets too aggressively, public trust in capitalism will weaken.

Executives should recognize that market legitimacy depends on contestability. A company can win by innovating. It loses legitimacy when it wins by making competition impossible.

Regional Inequality and the Geography of AI

AI is also reshaping the geography of opportunity. Data centers, AI labs, venture capital, cloud regions, chip supply chains, and advanced talent pipelines are not evenly distributed. They cluster in places with capital, energy, universities, infrastructure, and policy support. Regions without these assets may become consumers of AI rather than producers of AI-driven wealth.

This matters politically. The last era of globalization created places that felt left behind by trade, automation, and financial concentration. AI could repeat that pattern if investment clusters in a few metropolitan and infrastructure-rich regions while other communities see only labor disruption.

Data centers illustrate the tension. They can bring investment, tax revenue, and infrastructure demand. But they also require energy, water, land, and grid capacity. Communities may ask whether they receive durable jobs or merely host the physical infrastructure for wealth created elsewhere. If local costs are visible and local benefits are limited, resistance will rise.

Inclusive AI growth requires regional strategy. Governments can invest in broadband, energy infrastructure, technical education, research institutions, and regional innovation hubs. Companies can distribute operations, build local talent pipelines, support supplier ecosystems, and create AI-enabled opportunities beyond elite hubs. The goal is not to force every region to become a frontier AI center. It is to ensure that AI adoption creates productive opportunity across places, not only in already dominant regions.

Capitalism loses legitimacy when geography becomes destiny.

Policy Responses

Policy will matter, but policy must be more sophisticated than either techno-optimism or anti-automation resistance. The goal should be to spread AI’s gains while preserving the innovation that produces them.

The first policy priority is human capital. Workers need access to continuous learning, not occasional retraining slogans. Education systems must adapt to AI-enabled work. Mid-career workers need practical pathways into new roles. Apprenticeship models must be redesigned so that AI does not eliminate the early tasks through which workers learn. Public funding, employer investment, and credentialing systems must align around actual labor-market demand.

The second priority is competition. Policymakers must monitor whether AI deepens market concentration through data advantages, cloud dependency, platform lock-in, acquisitions, or exclusionary practices. Competition policy should preserve innovation while preventing infrastructure control from becoming economic gatekeeping.

The third priority is portable security. Workers navigating AI disruption need benefits, mobility, wage insurance, training support, and transition mechanisms that are not tied entirely to one employer or one job. A more dynamic economy requires stronger transition infrastructure.

The fourth priority is broader ownership. Policymakers can support employee ownership, profit-sharing, retirement access, sovereign wealth mechanisms, public investment vehicles, or other structures that allow more citizens to participate in AI-driven capital growth. If AI mainly rewards capital, then broader capital ownership becomes more important.

The fifth priority is regional investment. AI infrastructure should not deepen spatial inequality. Energy, education, digital infrastructure, and local innovation systems will determine whether regions can participate in the AI economy.

None of these policies is simple. But the alternative is to allow AI wealth to concentrate first and attempt to repair the political damage later. That was the mistake of earlier economic transitions.

The Corporate Role

Companies cannot wait for governments to solve the AI inequality problem. Business leaders are making the deployment decisions that will shape how AI affects work, wages, skills, and trust. They therefore have a role in mitigation.

The first responsibility is transparency with employees. Leaders should explain how AI will affect roles, what skills will matter, where automation is likely, and how the company will support transition. Vague reassurance is less credible than honest planning. Employees can adapt to change more effectively when they understand the direction of travel.

The second responsibility is redesigning work rather than simply cutting labor. AI should be used to eliminate waste, improve quality, expand capability, and create better jobs where possible. If every AI gain is translated only into headcount reduction, companies may weaken both culture and public trust.

The third responsibility is sharing productivity gains. This can take many forms: bonuses, profit-sharing, wage progression, employee ownership, training investment, career mobility, shorter workweeks in some contexts, or reinvestment in human capability. The specific model will vary, but the principle is clear. Workers should see some benefit from the technologies they help implement.

The fourth responsibility is preserving learning pathways. Companies must create new apprenticeship models for an AI-enabled workplace. Junior employees should learn through AI review, supervised judgment, exception handling, simulation, and mentorship. If firms reduce entry-level hiring too sharply, they may create a future leadership and expertise gap.

The fifth responsibility is supporting competitive ecosystems. Large firms should consider how their AI strategies affect suppliers, small partners, developers, and customers. A stronger ecosystem can create more durable value than a closed system that extracts from dependent participants.

Corporate leaders should understand that inequality is not external to strategy. It affects talent, regulation, demand, legitimacy, and social stability.

Inclusive Growth as Innovation Strategy

The best response to AI inequality is not defensive redistribution alone. It is inclusive growth. This means using AI to expand access, lower barriers, improve productivity for smaller firms, support workers, and create new markets.

AI can help small businesses manage finance, marketing, logistics, customer service, and compliance. It can help local governments improve service delivery. It can support healthcare in underserved areas. It can personalize education and training. It can help workers move into better jobs by identifying skill pathways. It can reduce friction for entrepreneurs who lack elite networks or expensive professional support.

But these outcomes require design. Without intentionality, AI may simply strengthen those already advantaged. Inclusive growth asks: who is this technology empowering? Who is being displaced? Who owns the gains? Who has access? Who is accountable? Which communities benefit? Which firms become dependent?

Companies that answer these questions well may discover new markets. The next phase of AI competition will not be only about serving large enterprises. It may also involve building tools for overlooked segments: small businesses, frontline workers, regional employers, public institutions, independent professionals, and emerging-market entrepreneurs.

Inclusive AI can be a growth strategy. It can expand demand, reduce inequality pressure, and strengthen capitalism’s social foundation.

A Framework for Leaders

Executives should assess AI inequality through five lenses.

The first is distribution of gains. Who captures the productivity upside: shareholders, executives, workers, customers, suppliers, or communities? The answer will shape trust.

The second is labor mobility. Does AI create pathways into better work, or does it remove entry-level opportunities and trap workers in declining roles? Companies should measure not only productivity, but mobility.

The third is market concentration. Does the company’s AI strategy strengthen competition or increase dependency and lock-in? Firms should be cautious about models that create short-term advantage by undermining market legitimacy.

The fourth is regional impact. Where are AI investments located, and who benefits from them? Leaders should consider energy, infrastructure, jobs, local suppliers, and community trust.

The fifth is institutional resilience. Does AI adoption strengthen the company’s long-term capability, or does it mask dependence on human expertise while eroding the workforce pipeline? Short-term efficiency can create long-term fragility if leaders fail to invest in human capital.

This framework does not require companies to solve inequality alone. It requires them to understand how their choices affect the economic system in which they operate.

The Social License for AI Capitalism

Every economic system depends on a social license. People must believe that the system is not only productive, but legitimate enough to support. Capitalism’s social license has always rested on an implicit promise: if markets grow, opportunity expands; if innovation succeeds, society benefits; if companies prosper, workers and communities have pathways to prosperity as well.

AI tests that promise. If the technology produces extraordinary wealth while narrowing opportunity, the backlash will be severe. If it creates productivity without mobility, growth without fairness, and intelligence without shared prosperity, capitalism will become harder to defend.

The companies that recognize this early will be better positioned. They will build AI strategies that improve performance while protecting trust. They will invest in people as well as platforms. They will share gains in ways that make transformation credible. They will support competition rather than rely only on lock-in. They will help build markets that more people experience as fair.

The companies that ignore the inequality question may still grow quickly. But they will grow in a system whose legitimacy is weakening around them.

The Real Reform

The AI-driven wealth divide is not inevitable. It is a design challenge. Institutions, policies, business models, ownership structures, labor systems, education pathways, and corporate choices will determine whether AI deepens inequality or supports a more inclusive capitalism.

The mistake would be to wait until resentment hardens. By then, the policy response may be blunt, reactive, and hostile to innovation. The better path is to build distributional fairness into the AI economy while the system is still taking shape.

Capitalism has renewed itself before by expanding participation: through mass employment, public education, homeownership, retirement systems, competition policy, entrepreneurship, and access to capital. The AI era will require its own renewal. It will need new ways for people to learn, own, contribute, and share in the value created by intelligent systems.

The future of capitalism will not be determined only by how powerful AI becomes. It will be determined by whether AI makes the economy feel more open or more closed.

That is the executive and policy challenge of 2026: to ensure that intelligence does not become another wall between those who own the future and those expected to adapt to it.

AI can make capitalism more productive. Whether it makes capitalism more legitimate will depend on what leaders do next.