Who Really Wins the AI Era?

AI may democratise capability without democratising power.

Yesterday I spoke about the infrastructure underneath AI: data centres, chips, energy, cloud platforms and capital required to build them.

But infrastructure does more than enable technology.

It concentrates power around those who own and control it.

The original promise of the internet was that technology would distribute power outward.

It lowered the cost of publishing, communication and access to information, allowing individuals and smaller organisations to participate.

AI appears to extend that promise.

A single person can now access capabilities that once required teams of researchers, developers, designers, analysts or advisers.

That is a remarkable democratisation of capability.

But underneath it sits an increasingly concentrated economic system.

Advanced AI depends on computing capacity, AI chips, energy, cloud infrastructure, proprietary models, capital and global distribution.

Much of that is controlled by a small group of technology companies.

The technology barons of the AI era may become its greatest commercial winners.

That does not mean the population will be its greatest beneficiaries.

Their interests do not have to be malicious for this to matter.

These companies have shareholders, investors, competitive pressure and strategic interests. Those interests will not always align with the wider public interest.

We have seen this before.

Railways, oil, telecommunications and finance created enormous public benefit while also concentrating economic and political power among those who controlled the infrastructure.

That is why the antitrust and competition scrutiny now being applied to the technology sector matters so much to AI.

Governments and competition authorities are examining relationships between major cloud providers, AI developers and the critical infrastructure on which the industry depends.

This is among the strongest arguments for thoughtful regulation of AI.

Not regulation to slow innovation, but to prevent a transformative technology becoming so concentrated that a small number of existing technology powers determine who can participate, on what terms, and where the economic value accumulates.

Once that concentration becomes deeply embedded, correcting it becomes much harder.

AI may be developing into another foundational infrastructure on the scale of industries that came before it.

So perhaps the question is not simply whether AI is democratising technology.

It is whether we are democratising capability on top of an increasingly concentrated system of economic power.

Those are not the same thing.

Does commercial success in AI necessarily mean success for society?

If AI becomes one of the foundations of economic and social life, its success should be judged not only by the wealth it creates for those who build it, but by how widely its benefits, opportunities and power are distributed.