Distribute the Capability

For the better part of four decades—since deregulation, the rise of shareholder primacy, and the leveraged buyout boom reshaped corporate incentives in the early 1980s—the dominant strategy in the American economy has been consolidation: aggregate demand, centralize operations, boost margin, repeat. The financial results have been spectacular—U.S. GDP has grown nearly tenfold in nominal terms.¹ But the gains have not been shared with the people who produce them. Real GDP per capita has more than doubled since 1980, while real median household income has grown just 15%.² Middle-wage workers saw their hourly wages rise less than 0.2% per year over the 34 years from 1979 to 2013; since 1978, productivity has grown 80.5% while typical worker compensation has grown just 26%.³ The income share captured by the top 1% more than doubled, from roughly 10% to nearly 24% during the same period.⁴

The consequences extend beyond income. Consolidated and globalized systems degrade quality to service acquisition debt. They destroy institutional knowledge through rolling layoffs. They replace curation with algorithmic promotion. The pattern repeats across every sector consolidation touches. Every category PE has rolled up—veterinary care, dental practices, local news, casual dining, emergency medicine—delivers measurably worse outcomes than it did a decade ago. The damage is not confined to any one industry. It is cumulative, compounding, and increasingly visible to the people living inside it.

The cumulative effect is a trust deficit that extends far beyond any single industry. This is not a sentiment problem. It is a market inefficiency—a widening gap between demand for trustworthy commerce and the supply of it. The economy produces more but delivers less to the people who participate in it.⁵

The divergence described above is structural, not perceptual: a K-shaped bifurcation in which asset owners and credentialed professionals in a handful of metropolitan economies accelerate upward while the broad middle slides backward—earning less in real terms, owning less, and facing higher costs for the essentials that define economic security. The question is what mechanism produces rising aggregate wealth and declining lived experience simultaneously. Three concepts structure the analysis below: productive capacity, extraction, and capability. The relationships are a conceptual framework for strategic reasoning, not empirically calibrated equations—but as directional tools for evaluating whether economic activity is building or degrading the productive base, they are precise enough to be useful.

Definitions

The Extraction Ratio

For any economic actor—a company, a platform, a fund—define:

When E approximates the competitive equilibrium level (E*), the actor captures returns proportional to the value it creates. This is healthy commerce—profit earned through competitive merit.

When E persistently exceeds the competitive benchmark, the actor is capturing economic rent—returns sustained by market power rather than by the value it delivers. This is extraction: a quantifiable measure of the damage that anti-competitive consolidation inflicts on productive market economies.

The trajectory of E over time (dE/dt) is the critical directional indicator. A competitive market pushes E toward E*—competition forces actors to create more value relative to what they capture. A consolidated market allows E to drift upward—reduced competition means actors can capture more while delivering less.

The Capability Curve

Historically, capability has been strongly correlated with organizational scale: Kᵢ ≈ f(scaleᵢ). A five-person company had dramatically less operational capability than a five-hundred-person company. This is why consolidation wins—scale confers capability, capability confers competitive advantage, and the cycle reinforces itself.

AI changes the function. With distributed AI tooling: Kᵢ ≈ f(toolsᵢ). Capability decouples from scale. A five-person business with the right tools approaches the operational capability of a fifty-person business. The structural advantage of consolidation collapses.

The critical question is who builds those tools and for whom. If AI tooling is built exclusively for large enterprises—which is the current default—the capability curve shifts up for incumbents while small operators remain locked out. The gap widens. If AI tooling is built for the long tail, the curve flattens. More participants can compete. Markets become more competitive. Extraction ratios fall. Productive capacity expands.

Productive Capacity

At the macro level, productive capacity is the aggregate of individual actors’ capability, weighted by how much of that capability translates into real value rather than being captured as rent:

P ∝ Σᵢ (Kᵢ × Aᵢ)

where Aᵢ = 1 – normalized extractionᵢ

For simplicity, assuming representative averages across actors:

P ≈ N × K̄ × (1 – Ē)

With the formal model established, we can now examine the specific mechanisms through which extraction has increased and productive capacity—along with consumer trust, institutional quality, and the broader social fabric that sustains a functioning market economy—has declined.

Enshittification

Cory Doctorow’s term describes the lifecycle of dominant platforms: attract users with genuine value (low E), shift extraction toward business customers once users are locked in (rising E), then extract from both sides to maximize platform returns (E >> 1).⁶ The pattern now extends well beyond digital platforms. PE applies the same logic to physical businesses—acquire a trusted brand, load it with debt, cut costs until the product degrades, extract fees and dividends before the brand equity is depleted. In formal terms, the PE playbook is to acquire a business with low E, raise E as rapidly as the market will tolerate, and exit before the consequences fully materialize.

In physical goods, the equivalent is planned obsolescence—the deliberate engineering of product failure to force repurchase cycles. Appliances that once lasted decades now barely survive their warranties. Electronics are designed to be irreparable. Materials are downgraded while packaging is upgraded. The extraction ratio is encoded into the product itself: the manufacturer captures the same or higher price (C) while delivering progressively less durable value (V declining over the product’s engineered lifespan). This is not a market outcome. It is a design decision made possible by consolidated markets where consumers lack viable alternatives.

The Disappearance of Healthy Competition

The extraction trajectories above are not inevitable features of successful businesses. They are symptoms of markets in which competition has been structurally removed. In a functioning competitive market, rising E is self-correcting: an actor that captures more than it creates loses customers to one that doesn’t. Extraction invites entry. Entry restores equilibrium. That is the mechanism through which free markets are supposed to work.

What has occurred over the past two decades is not that certain companies won the competition. It is that the competition itself was eliminated. PE does not roll up 200 dental practices to compete more effectively in dentistry. It rolls them up to remove competition from the market—to create pricing power, reduce consumer choice, and extract margin that a competitive landscape would not permit. Platforms do not outcompete local merchants; they interpose themselves between merchant and customer, making the direct relationship impossible and then charging rent on the indirect one. This is the mechanism that allows E to rise persistently rather than reverting to equilibrium. Consolidation does not just shift market share—it dismantles the feedback loop. Multiply this across every sector PE has touched—dental, dermatology, emergency medicine, home services, local news—and the result is not a more efficient economy. It is an economy in which the corrective mechanism has been removed and extraction proceeds without constraint.

The AI investment thesis, as currently configured, accelerates this. If AI capability flows exclusively to enterprises and platforms, it does not merely create a capability gap—it makes competition structurally impossible. An independent operator cannot compete against an entity with AI-driven pricing, AI-optimized customer acquisition, and AI-automated operations while running on spreadsheets and phone calls. The gap becomes a wall. The market does not correct because there is no market—only incumbents whose AI-augmented scale makes entry unviable.

This is why distributing AI capability is not merely a commercial opportunity. It is the restoration of the competitive mechanism itself. Increasing K across the long tail does not just help small businesses—it rebuilds the market structure under which extraction is punished and value creation is rewarded. In formal terms: competition is the force that pushes E toward E*. Without it, E rises indefinitely. Restoring it requires restoring the capability conditions under which entry is viable.

From Manufacturing to Knowledge Work: The Completion of Offshoring

The offshoring of American manufacturing in the 1980s and 1990s was sold under the doctrine of comparative advantage: let cheaper labor markets handle production while Americans move up to higher-value design, engineering, and management. What actually happened was the hollowing out of entire regional economies and the destruction of the most reliable pathway to middle-class income. The communities that lost manufacturing largely never recovered. The “higher-value work” materialized for some, in some places, but nowhere near the scale needed to replace what was lost.

Technology and knowledge work are now following the identical script. First call centers and basic IT. Then software development. Now engineering, data science, AI research, legal analysis, financial modeling. The justification is always the same: labor arbitrage improves margins, shareholders benefit, and displaced workers will “upskill” into something else. But the logic collapses when applied to the work that was supposed to be the something else. There is no next rung on the ladder when you offshore the top of the ladder.

The Logical Endpoint

Follow these trends to their conclusion and the picture is sobering. The largest companies domiciled in the United States are American in a legal sense but operate as vehicles for global capital allocation. They manufacture where labor is cheapest, recognize revenue where taxes are lowest, and optimize employment for shareholder returns rather than domestic economic health. Their fiduciary duty runs to a globally dispersed investor base that is indifferent to local outcomes. They invoke their American identity selectively—when lobbying for subsidies, tax policy, IP protection, or military security—but disclaim any reciprocal obligation to American workers or communities. For those who are pro-free-market and pro-democratic institution, the gradual conversion of the world’s most productive economy into a financial intermediation layer—a holding structure for globally dispersed capital rather than a generator of broadly shared prosperity—represents a betrayal of the very principles that made the system worth defending.

Extend this pattern and the United States begins to function less as a productive economy and more as a financial operating system—providing capital markets infrastructure, contract enforcement, IP protection, and a military umbrella while the actual value creation happens elsewhere and the returns flow to institutional investors in Riyadh, Oslo, and Greenwich alike. CEO-to-worker compensation ratios have grown from 21:1 in 1965 to 281:1 in 2024. CEO realized compensation has grown 1,094% since 1978; typical worker pay has grown 26%.⁷

This is not a partisan argument. It is a description of where the prevailing incentive structures lead. Public companies are legally bound to maximize shareholder returns; their executives are compensated on equity appreciation and quarterly performance. These incentives reward offshoring, headcount reduction, and margin expansion—regardless of the downstream effects on domestic employment, product quality, or community economic health. The incentives do not reward building durable productive capacity; they reward extracting value from it. Changing the outcomes requires changing the incentive architecture—and that is the context in which the opportunity we are describing becomes not just commercially attractive but structurally important.

A common narrative around AI is one of abundance and optimism. Productivity gains will make goods and services cheaper. Robotics will reduce manufacturing costs. Automation will eliminate inefficiency. Healthcare, education, housing, energy—the sectors where costs have risen fastest and most painfully—will become radically more affordable. Living standards will rise for everyone. This is the future being sold by technology leadership, venture capital, and policy advocates alike.

The promise is not necessarily wrong. The technology is genuinely capable of delivering abundance. AI and robotics have the potential to reduce the cost of producing almost anything. Supply constraints in housing, energy, infrastructure, and healthcare are primary drivers of the affordability crisis, and removing those constraints through technology and policy reform is among the most important economic projects of our time.

But abundance is not an automatic consequence of productivity gains regardless of the mechanism that produces them. It is a function of where those gains flow. If AI makes it 50% cheaper to produce a good, that cost reduction can manifest in three ways: the product gets 50% cheaper for the consumer (abundance), the producer pockets the savings as higher margin—legitimate profit if earned through competitive merit—or an intermediary captures the difference through platform fees, financial engineering, or market power (extraction, as defined above: returns sustained by market position rather than incremental value created). The technology enables all three outcomes. The ownership structure, the incentive design, and the market concentration determine which one actually occurs.

Abundance also requires a population capable of participating in it. A country without a functional middle class—in which a small cohort of asset owners trades an ever-larger share of GDP among themselves while the majority faces stagnant wages, rising costs, and diminishing economic mobility—is not a picture of abundance. It is a picture of a society consuming its own foundations.

To date, the pattern is clear. Productivity gains from technology over the past four decades have overwhelmingly been captured as extraction by capital owners rather than shared with consumers or workers. This is not speculation—it is the documented divergence between GDP growth and median income growth, between productivity and wages, between corporate profits and consumer prices. There is no empirical basis for assuming that AI-driven productivity gains will distribute differently under the same ownership and incentive structures that have captured every prior wave of technological improvement.

The gap between the promise of abundance and the reality of extraction is not a failure of technology. It is a failure of structure. In the language of the formal model: abundance occurs when rising K translates to falling prices and improving quality for end users, which requires low E. Under the current structure—enterprise AI focused on headcount reduction, consolidated markets preventing competitive pressure on pricing—AI produces margin expansion for capital owners, not abundance for communities. The technology promises abundance. The incentive structures guarantee extraction. That gap is the central economic question of the next decade, and it is the gap this thesis exists to close.

AI is reducing the cost of operational competence faster than any technology since cloud computing. Functions that required dedicated departments—marketing, bookkeeping, customer operations, design, procurement—can now be handled by a small team with the right tools.

But the current investment wave is aimed almost entirely at enabling large enterprises to cut headcount. The tools are priced for enterprise budgets, sold through enterprise sales motions, designed for enterprise workflows. The long tail of small businesses—which collectively employ more Americans than all Fortune 500 companies combined—is being ignored.

This is the gap. The same capabilities that let a Fortune 500 company eliminate 30% of its support staff can let a ten-person business operate with the sophistication of a company fifty times its size. Same technology. Different target. Different outcome.

In formal terms: increase N (more viable competitors), increase K̄ (through distributed AI tools and shared infrastructure), and decrease Ē (through business models aligned with value creation). The result is expanded productive capacity: P ≈ N × K̄ × (1 – Ē), with all three terms moving in the right direction simultaneously.

Each element of the thesis maps directly against the extraction cycle. Where enshittification and planned obsolescence degrade quality (rising E), the opportunity is in tools the operator owns and products built to last. Where offshoring decouples profit from place (reducing N in domestic markets), the opportunity is in businesses rooted in communities. Where AI deployed for headcount reduction concentrates gains (increasing E while reducing N), the alternative is AI that makes independent operators more capable—distributing gains across the broadest possible base. This is not a corrective to capitalism. It is capitalism applied to the problem that consolidation has created.

The defensibility compounds. Extractive platforms lose their best suppliers and most discerning customers as alternatives emerge. Businesses built on genuine value creation exhibit the opposite: lower churn, stronger referral economics, and rising lifetime value per customer. In the language of the model, businesses with structurally low E build durable competitive advantage precisely because they do not extract from the participants who sustain them.

The opportunities are concentrated in sectors where consolidation has visibly degraded quality (high and rising E), independent operators have latent advantages they cannot currently exploit (high potential K constrained by lack of tools), and technology can close the gap without requiring the operator to become a technology company. Four categories stand out. First, AI-native small business operations—purpose-built tools for businesses with one to twenty-five employees that increase K across the long tail. Second, curated vertical commerce—category-specific marketplaces organized around trust and quality rather than algorithmic promotion, directly addressing the market gap created by enshittification. Third, shared infrastructure for independent operators—commercial kitchens, procurement cooperatives, fulfillment networks—delivering the cost structure of scale without surrendering ownership or margin. Fourth, regional supply chains that serve growing consumer demand for quality and provenance while reducing the fragility that consolidation has introduced. Each category increases N, raises K, or lowers E—and in most cases, all three.

36.2 million small businesses.⁸ 62.3 million employees. 45.9% of the private workforce.⁹ 43.5% of GDP.¹⁰ Nine out of ten net new jobs.¹¹ The vast majority have no access to the tools, infrastructure, or market channels their consolidated competitors take for granted. This is not a niche. It is the largest underserved segment in the American economy.

On the consumer side, willingness to pay for quality, transparency, and local provenance is a mass-market shift—not a coastal luxury. Surveys indicate 91% of consumers prefer buying from small businesses when convenient, and 77% will pay more for better service and local impact.¹² The preference is currently undersupplied because independent operators lack the infrastructure to serve it at scale. Resolving that supply constraint is the core of the thesis.

The globalization and consolidation thesis has run its course. Its returns depend increasingly on financial engineering and the degradation of product and service quality rather than value creation. The trust it has destroyed and the devastation it has inflicted on American cities and regional economies are opening durable opportunities for a fundamentally different approach to incentive structures, company building, and economic design. AI provides the mechanism to exploit those opportunities at a cost structure that was not previously possible.

The broader stakes are worth stating plainly. A country that no longer makes things, no longer builds things, and increasingly no longer thinks things—while its largest corporations optimize for the returns of a global investor class—is a country with a shrinking economic foundation and a growing legitimacy crisis. History is unambiguous on this point: populations that cannot find productive work do not indefinitely continue to participate as willing economic and political subjects. The populism already visible across Western democracies is the mild version of what happens when that social contract breaks.

When extraction ratios have climbed this steeply in the past, collective structures have always emerged in response—guilds in the mercantilist era, labor unions during industrialization, cooperatives during agrarian consolidation. Each represented an attempt to rebalance bargaining power against concentrated capital. Today the tension is more visible than ever: global firms leverage labor arbitrage to compress costs abroad while seeking land rights, tax incentives, and public infrastructure at home—asking local communities to subsidize facilities whose jobs are either imported specialists or temporary construction, while the community bears the externalities and gets almost nothing back. At some point, people stop consenting to that arrangement. The distributed capability model offers a market-native alternative: rather than organizing against concentrated capital, it equips independent operators with the tools to compete on their own terms. The leverage comes not from collective bargaining, but from closing the capability gap itself.

The consequences of ignoring it are far more expensive than the cost of building an alternative. The small business economy is not a sentimental cause. It is the productive base on which everything else depends. Strengthening it is not charity. It is maintenance of the system that makes the rest of the economy possible.

One question sits at the center of everything we build: if AI makes a small operator ten times more capable, does the value accrue to the operator or to the platform? In formal terms: does the increase in Kᵢ flow to the operator or get captured as higher E by an intermediary? Every product decision, every architectural choice, and every business model we design points toward the operator. That is both the more durable, more ethical, and the more profitable answer.