Why Does Artificial Intelligence Create Economic Power?
By Why of Money
Published: October 13, 2026
Last updated: October 13, 2026
Introduction
Artificial intelligence can help businesses reduce costs, develop products, and serve customers more efficiently. Yet the economic benefits of a technology do not necessarily flow equally to everyone who uses it. Some companies may gain much more influence because they control the resources, infrastructure, or platforms on which other businesses depend.
The distinction is between using a technology and controlling the conditions under which others can use it. A small company might use an existing AI model to compete with larger firms, while the companies supplying computing infrastructure or controlling essential distribution channels occupy a different position: they can influence the costs and options available to many businesses at once.
This does not mean AI inevitably produces monopolies. Falling model prices, new competitors, and open-source alternatives can weaken established advantages. But the structure of AI markets creates several mechanisms through which an early lead can become lasting economic power.
Understanding these mechanisms helps explain not just who develops AI, but who can capture the economic value it creates.
——-
How Economies of Scale Can Create an Advantage
Developing advanced AI systems can require substantial upfront investment in research, specialized equipment, computing capacity, and technical talent. These costs create an advantage for companies that can spread them across a large number of customers.
This is known as economies of scale: as production expands, the average cost of providing a service may fall because fixed costs are distributed across more units of output.
Consider a hypothetical example. Two companies develop AI services with similar initial research and development costs. One serves a relatively small customer base, while the other serves millions of users. If both can deliver additional requests without increasing their costs proportionally, the larger company may spread its initial investment more efficiently.
Scale can also support economies of scope. A company that already operates cloud services or business software may be able to incorporate AI into existing products, share technical resources, and reach established customers without building an entirely new business from scratch.
However, size alone does not guarantee lower costs or stronger competition. Large systems can be expensive to operate, and smaller firms can compete through specialized products, more efficient models, or different business strategies.
The important economic distinction is that the cost of using AI can fall even when the cost of developing and operating the most advanced systems remains a major barrier to entry.
This is one reason AI can simultaneously make certain business activities more accessible and strengthen the position of companies that can afford to build or operate the underlying technology.
——-
Why Control of Essential Infrastructure Matters
Economies of scale explain why large companies may have cost advantages. Control of essential infrastructure explains how those advantages can translate into influence over other businesses.
AI depends on interconnected layers: specialized chips, data centers, cloud computing, models, and applications. A company may compete successfully at one layer while depending on suppliers at another.
For example, an AI startup might develop a useful application but rely on a cloud provider for computing capacity and an external model provider for core functionality. Its ability to compete then depends partly on the price, availability, and technical conditions of those inputs.
When several suppliers offer credible alternatives, the startup has greater bargaining power. When a critical input has few substitutes, the supplier may have more influence over the businesses that depend on it.
The same issue applies to infrastructure that is difficult to reproduce quickly. Building data centers, obtaining advanced chips, and establishing the necessary technical capabilities can require substantial capital and time. Even a promising competitor may struggle to expand if access to these resources is constrained.
The OECD’s 2026 report, Artificial Intelligence Markets: Recent Developments and Competition Issues, identifies concentrated computing resources, high fixed costs, and vertical integration as potential risks to competition across the AI value chain. It also recognizes that specialized model providers continue to challenge established technology companies in some market segments.
Infrastructure control does not automatically constitute abusive behavior. A supplier may invest heavily, provide a valuable service, and compete fairly. The economic concern arises when control over an essential input allows a company to restrict rivals’ options or extend its influence into adjacent markets.
——-
How Switching Costs Can Lock Businesses Into a Provider
A company can gain economic power not only by offering a better product, but also when customers face substantial costs to leave it.
Switching costs are the financial, technical, and organizational burdens involved in changing providers. They can include transferring data, rebuilding software integrations, retraining employees, and testing replacement systems.
Suppose a company builds its customer support and internal workflows around one AI service. Over time, employees learn its tools, developers connect it to other systems, and managers adapt business processes to its capabilities. Even if another provider offers a lower price, switching may require significant time and investment.
This can weaken competition. A customer who is already committed to one provider may be less responsive to a competitor’s lower prices or better features because the cost of changing reduces the potential savings.
The effect becomes stronger when switching costs interact with infrastructure dependence. A business may rely on a cloud provider for computing, storage, security, and other services, while also using that provider’s tools to operate its AI applications. Moving one component may require changing several connected systems.
Switching costs are not always harmful. Integrations can improve reliability and make services more useful. The concern is whether customers can realistically leave and whether providers compete on the merits of their products rather than benefiting from avoidable barriers to switching.
A documented example: Competition in cloud computing
The UK’s Competition and Markets Authority investigated the public cloud infrastructure market and published its final decision in 2025. The investigation examined competition involving Amazon Web Services and Microsoft, including concerns about data egress fees, barriers to interoperability, and software licensing practices that could make switching providers or using multiple cloud services more difficult.
In 2026, the CMA reported that the investigation had identified significant market power held by Amazon and Microsoft and limits to customer choice associated with these barriers. It also described subsequent steps by the companies to lower egress fees and improve interoperability.
This is a useful example of the mechanism in practice. A business choosing a cloud provider is not evaluating only the initial price or technical quality. It must also consider how difficult and expensive it may be to move its data and applications later. Those future costs can influence competition today, even before the customer decides to switch.
The case concerns cloud computing rather than AI models alone. Its relevance to AI is that cloud infrastructure supports the development and operation of many AI services. When AI businesses depend on cloud providers, barriers to switching in the underlying infrastructure market can affect their costs and competitive options.
——-
How Network Effects and Feedback Loops Reinforce an Advantage
Some digital services become more useful as more people use them. Economists call this a network effect. For example, a communication service may become more valuable when more of a user’s contacts join it.
AI markets can also develop feedback loops, although the mechanism is not always a direct network effect. A larger customer base can generate more revenue, attract developers, encourage integrations, and create additional opportunities to understand customer needs. Under the right conditions, these advantages can improve a service and attract even more customers.
But more users do not automatically mean a better AI model. User data may be irrelevant, difficult to use, or unsuitable for training. Improvements depend on data quality, technical methods, privacy constraints, and the type of service.
The important mechanism is cumulative advantage: an existing lead can generate resources and relationships that help a company maintain or expand that lead.
For instance, a widely adopted AI platform may attract more developers because it already has a large customer base. More applications can then make the platform more attractive to customers. A rival may need to build not only a competitive product but also a comparable ecosystem of integrations and support.
This advantage is not permanent. Open-source tools, lower-cost models, new technical approaches, and changing customer preferences can disrupt established platforms. Feedback loops matter most when they create benefits that competitors cannot easily reproduce.
——-
Why Vertical Integration Can Strengthen Market Power
A company can also gain influence by controlling multiple stages of the same value chain. This is called vertical integration.
In AI, a company might operate cloud infrastructure, develop models, and distribute applications through an established software platform. These businesses can reinforce one another: infrastructure supports model development, models improve applications, and applications help attract customers to the company’s wider ecosystem.
Integration can benefit customers. It may improve compatibility, reduce coordination costs, and make services easier to deploy. A business may prefer a combined solution because it works reliably without requiring multiple suppliers to coordinate.
However, integration can also create competitive risks. A company controlling an essential platform may have incentives or opportunities to favor its own AI products, bundle them with existing services, or make it harder for rival providers to reach customers.
For example, a business that already relies on a particular software ecosystem may find its provider’s AI tool easier to adopt than a competing product. That preference can be entirely reasonable if the integrated tool offers better functionality or lower costs. The competitive concern arises when rivals are disadvantaged by restrictions or practices unrelated to the quality of their products.
The OECD’s 2026 analysis identifies vertical integration as one of the potential risks to competition in AI markets, particularly when firms operate across infrastructure, models, and downstream services.
The key question is not whether a company operates at multiple levels. It is whether its position at one level allows it to restrict competition at another.
——-
Why Creating Value Does Not Guarantee Capturing It
A business can create economic value without capturing all of that value for itself. This distinction helps explain why AI’s benefits may be distributed unevenly.
Value creation occurs when a technology improves a product, reduces the resources needed to produce something, or provides a service customers value. Value capture refers to the portion of the resulting economic benefit retained by businesses, workers, suppliers, and other participants.
Consider a hypothetical company that uses an AI service to develop products faster. The company may increase its sales or reduce its operating costs, but it must also pay for computing, model access, and software integrations.
If several providers compete to offer comparable services, the company may benefit from lower prices and retain more of its gains. If it depends on a supplier with few alternatives, that supplier may have greater bargaining power and capture a larger share of the economic value through its pricing.
The same principle applies to workers. If AI complements employees’ skills and increases demand for their work, they may benefit from higher wages or better opportunities. If their tasks become easier to replace and their bargaining power is weak, the business may capture more of the savings instead.
Market power can affect this distribution. A company facing strong competition may need to pass cost savings on to customers through lower prices or better products. A company with limited competitive pressure may be able to retain more of those gains.
This does not mean that every powerful company charges excessive prices or that every supplier captures an unfair share. The outcome depends on the alternatives available, the value each participant contributes, and the terms on which businesses and workers negotiate.
AI can therefore expand the total economic value produced while leaving the distribution of that value unequal.
——-
Does AI Always Increase Market Concentration?
No. The same technology that strengthens established companies can also help smaller businesses compete.
A small firm can use existing AI services without developing its own advanced model. It may automate administrative work, create software prototypes, analyze customer feedback, or offer specialized services with fewer resources than previously required.
These possibilities can lower barriers to entry. New firms may be able to test ideas, reach customers, and compete in markets that once required larger teams or greater investment.
Competition can also intensify when models improve, prices fall, open-source alternatives become more capable, or customers can move between providers more easily.
The evidence varies across technologies and markets. In its 2026 report on AI competition, the OECD describes a mixed picture: specialized providers have challenged established companies in some model markets, while structural risks remain in areas such as computing infrastructure and data. The report does not suggest that concentration is inevitable in every AI market.
The distinction is between making a technology easier to use and making a market easier to compete in. AI may lower the cost of launching a business while essential infrastructure, distribution channels, or customer relationships remain difficult to access.
Whether AI increases or reduces concentration depends on the balance between these forces. The technology can make entrepreneurship more accessible while simultaneously strengthening the companies that control critical parts of the market.
——-
What Determines Who Gains Economic Power From AI?
The distribution of AI’s economic benefits depends on whether companies can turn their advantages into positions that competitors struggle to challenge.
Several conditions are especially important:
* Access to essential resources: Can new firms obtain computing capacity, chips, data, and technical expertise on reasonable terms?
* Customer mobility: Can businesses change providers without disproportionate financial or operational costs?
* Interoperability: Can different AI services work together, allowing customers to combine products from competing suppliers?
* Access to customers: Can new providers reach users directly, or must they depend on platforms controlled by established competitors?
* Opportunities for entry: Can smaller firms challenge incumbents through better products, lower prices, or more specialized services?
These conditions determine how effectively competition can limit the influence of established companies. They also help distinguish a legitimate competitive advantage—such as a more efficient model or a better product—from market power sustained by barriers that prevent customers and rivals from choosing alternatives.
For businesses adopting AI, this means considering not only what a service can do today, but also the long-term implications of depending on it. For policymakers, the challenge is to preserve incentives for investment and innovation while ensuring that control over essential resources does not unnecessarily restrict competition.
——-
Conclusion
Artificial intelligence can create economic power through several connected mechanisms. Economies of scale can favor companies that spread large investments across many customers. Control over essential infrastructure can give suppliers leverage over businesses that depend on their resources. Switching costs can make customers reluctant to leave, while network effects and vertical integration can reinforce an established position.
These mechanisms influence who can compete, how much businesses pay to operate, and who ultimately captures the value AI creates.
Yet none guarantees permanent dominance. Lower model prices, open-source alternatives, new entrants, and effective competition can weaken established advantages. The outcome depends on the structure of each market and the ability of rivals and customers to challenge existing firms.
The central economic question is therefore not simply who has the most advanced AI. It is who controls the resources and relationships that make AI commercially valuable—and whether others can compete without depending on that control.
Sources
