The Coming Reckoning for Megacap Dominance

Artificial intelligence may loosen the advantages of corporate scale, while leaving smaller enterprises dependent on a few immense suppliers.

For years, the dominance of the great corporations has seemed almost self-perpetuating. Companies with ample capital could attract exceptional talent, finance their work and carry its results into markets that smaller competitors struggled to reach. Each success supplied the means for another, until continued expansion came to look like a rule of law.

Artificial intelligence may unsettle that assumption by allowing a few people to undertake work that once required a substantial organization. As technical and administrative capabilities become more readily available, corporations may derive less advantage from their ability to assemble the personnel needed to conduct complex businesses. An entrepreneur who can obtain much of that assistance without building an extensive payroll has less reason to surrender an idea to an established employer.

Matthew Gallagher’s Medvi offers a striking, albeit imperfect, illustration. According to The New York Times, the telehealth business generated $401 million in sales in 2025. When the newspaper profiled it in April 2026, Gallagher and his brother constituted its entire payroll, and the company was projecting $1.8 billion in sales for the year. That projection should be distinguished from completed sales, just as revenue should be distinguished from valuation. Even with those qualifications, the business had achieved an extraordinary volume of commerce for so small a permanent staff. Times reporting, republished by GV Wire.

Much of the labour behind those sales took place elsewhere. Medical providers and pharmacies supplied essential services, while contractors, agencies and professional firms performed work that Gallagher could not manage himself. Medvi’s own terms describe a patient management platform working with independent practitioners. Its small payroll therefore tells us more about how the business organized its labour than about how much human labour it required. Medvi’s terms.

The company’s growth also brought questions about its conduct. In February 2026, the Food and Drug Administration issued a warning letter to MEDVi, LLC over website claims concerning compounded drugs that the agency considered false or misleading. Any account of its efficiency must take those concerns into consideration, since the cost of supervision cannot simply be omitted from the calculation of a successful business. FDA warning letter.

More measured evidence nevertheless suggests that AI can substantially increase what an individual worker accomplishes. A study involving more than 5,000 customer-support agents found that access to an AI assistant increased the issue resolution per hour by approximately 15 percent on average, there was considerable variance among each worker. The result gives some foundation to the expectation that smaller teams will be able to undertake larger responsibilities, while its undetermined how far such improvements can extend across all occupations. Brynjolfsson, Li and Raymond, Generative AI at Work.

For someone contemplating a new enterprise, even a modest reduction in the cost of skilled assistance may be decisive. When developing an idea requires fewer employees and less initial capital, a founder can test its commercial promise earlier and retain a greater share of the business. Financing and sound judgment remain indispensable, but the expense of discovering whether an idea deserves a company may become considerably less forbidding.

Established corporations should take this possibility seriously. A newcomer can threaten a profitable line of business without reproducing the incumbent’s entire organization. By concentrating on a particular service and avoiding the overhead accumulated, a small enterprise may find room to compete where the larger company has grown accustomed to comfortable margins.

Acquisitions may become an increasingly important response. Large companies could devote more resources to finding promising newcomers before those businesses develop into substantial rivals. Yet that response complicates any prediction of declining corporate concentration. If the most successful independent firms are regularly absorbed by the giants, entrepreneurial activity may strengthen the very companies it initially appeared to threaten.

Whether AI encourages smaller firms ultimately depends on how it changes the relative expense of doing business inside and outside an organization. Ronald Coase explained that firms bring activities under common management when they can coordinate them more economically than through market transactions. Finding suppliers, negotiating agreements and monitoring performance all impose costs that an established organization may be able to reduce. Coase’s Nobel lecture.

AI could make dealing with independent suppliers easier, giving companies greater reason to purchase services they once performed themselves. It could also improve internal administration and allow large organizations to operate with greater efficiency. The prospect of fragmentation rests on which of these improvements proves more substantial, rather than on technological progress alone.

Many semesters ago, my accounting professor proposed treating departments such as accounting and legal services as separate businesses, accountable for what they consumed and what they contributed. I found the suggestion persuasive because it promised to expose expenses that an organizational chart could otherwise conceal. A department required to justify its charges would have to examine its work more closely, while those purchasing its services would have reason to consider whether their demands warranted the expense. This belief was further affirmed when Jane Street described a related approach to allocating computing resources. Researchers bid for access to its computing cluster through a real-time auction, bringing the opportunity cost of their work into their daily decisions. The arrangement gives scarce computing time a price that reflects competing demands within the firm. Jane Street’s account.

Extending that discipline to other departments would require care. The value of preventing a lawsuit or maintaining reliable accounts is harder to measure than the use of computing time. Nevertheless, greater visibility into costs could encourage corporations to grant their operating units more independence, and perhaps to purchase more services from outside specialists. Some might gradually resemble federations of businesses held together by common ownership.

The physical foundations of industry will place limits on such changes. TSMC reported capital expenditure of $40.9 billion in 2025, while ASML reported €4.7 billion in research and development spending. Their positions rest on enormous investments and accumulated technical knowledge that access to AI alone cannot reproduce. TSMC’s management report, ASML’s annual report.

Ford and Procter & Gamble likewise possess production systems, supply relationships and distribution networks that would take a challenger considerable time and capital to establish. These advantages may prove more durable than those derived chiefly from the employment of large office staffs, though their durability offers no guarantee of perpetual leadership.

Indeed, an economy containing more small businesses could still depend on a few immense suppliers. Widely available AI models would leave those businesses reliant on chips, data centers and electricity, whose production may remain highly concentrated. The spread of entrepreneurial capability could therefore coincide with a further concentration of power among the companies providing its infrastructure.

There is reason, then, to question megacap dominance without announcing its imminent demise. Broad access to capable AI could permit more independent firms to reach a commercially meaningful scale, provided they can also reach customers on workable terms. Whether they remain independent will depend on their financing, their suppliers and their ability to withstand the responses of established competitors.

The giants have ample resources with which to defend their position. What has become less certain is the assumption that economic progress must continually enlarge their advantage. As smaller groups acquire the means to undertake more ambitious work, the great corporation will have to demonstrate more often that its size serves a useful purpose.