Will AI Be Dominated by a Few Firms?

The forces that could concentrate AI in a handful of large companies, the forces pushing toward competition, and how regulators are responding.

A key economic question about AI is whether it will be dominated by a few giant firms or remain competitive.

Forces toward concentration

Huge fixed costs: training frontier models requires enormous computing power, data and talent, favouring deep-pocketed firms.
Control of key inputs: a few companies control advanced chips, cloud computing and large user bases.
Vertical integration: big technology firms own several layers of the AI stack, from chips and cloud to models and consumer products.
Partnerships and investments: large firms have invested heavily in AI start-ups, such as Microsoft in OpenAI and Amazon and Google in Anthropic, raising questions about influence over potential competitors.
Data and distribution: companies with billions of users can gather feedback and distribute AI products widely.

Forces toward competition

Falling costs: the price of using AI models has dropped sharply as efficiency improved.
Open models that anyone can build on.
New entrants: companies like DeepSeek showed that competitive models could be built at lower cost than expected.
Many applications: specialised AI products for particular industries and languages leave room for many firms.

Regulators respond

Competition authorities in the United States, the United Kingdom, the European Union and elsewhere have studied AI markets and partnerships between large technology firms and AI start-ups. The UK’s Competition and Markets Authority published reports on foundation models warning about risks of concentration. Regulators have examined whether big firms’ investments in AI start-ups amount to mergers that should be reviewed.

The partnership question

A large technology company invests billions of dollars in an AI start-up and provides it with cloud computing. The start-up's models run on the company's cloud and are integrated into its products. Is this a partnership that helps a new competitor grow, or a way for the large company to control a rival? Regulators have grappled with exactly such arrangements.

Thinking the market structure is already settled

AI markets are changing rapidly, with new entrants, falling costs and shifting partnerships. Whether AI becomes concentrated or competitive depends partly on technology and partly on policy choices yet to be made.

Key takeaways

High fixed costs, control of chips and cloud, and vertical integration push AI toward concentration.
Falling costs, open models and new entrants push toward competition.
Big tech investments in AI start-ups raise questions about influence over rivals.
Competition authorities are studying AI markets and partnerships.