AI Is Changing the World — But Who Is Paying for It?

Artificial intelligence is often described as a digital revolution.

That description is incomplete.

AI may appear on a screen, but the system powering it is profoundly physical. Behind every chatbot, image generator and increasingly sophisticated AI model are semiconductor factories, vast data centres, power plants, transmission lines, cooling systems and an unprecedented volume of capital.

The world is not simply building artificial intelligence.

It is building the industrial infrastructure required to sustain it.

And the question is becoming impossible to ignore:

Who is going to pay for it?

The AI gold rush

The scale of investment is extraordinary.

The largest technology companies are committing hundreds of billions of dollars to AI infrastructure. Reuters reports that spending by the major hyperscale providers is expected to reach around $725 billion in 2026, while one major Ohio project involving OpenAI, Nvidia and SB Energy is planned to reach an eventual 8 gigawatts of capacity.

These are no longer ordinary technology investments.

They are infrastructure projects on the scale of industrial systems.

And that distinction matters.

The early internet revolution was largely associated with software, websites and devices. The AI revolution requires all of those things—but it also requires an enormous and continuous supply of electricity.

The real bottleneck is increasingly not computing power.

It is the infrastructure needed to power the computers.

The hidden bill

According to the International Energy Agency, global electricity consumption from data centres is expected to rise from roughly 485 terawatt-hours in 2025 to around 950 terawatt-hours by 2030. AI is a major driver of that expansion.

The numbers reveal a simple reality:

AI is not weightless.

Every expansion in AI capability ultimately produces a demand somewhere else—for more power generation, more grid capacity, more land, more cooling and more capital.

That pressure is already changing where data centres are built.

Developers in Europe are increasingly moving projects away from major cities in search of cheaper energy, available land and faster grid connections. New hyperscale facilities are being planned much farther from traditional urban hubs than they were only a few years ago.

The location of the next great AI hub may therefore be determined less by where the best programmers live than by where the electricity is available.

The debt question

The financial implications are just as significant.

The AI buildout is increasingly being financed through a combination of corporate investment, long-term contracts, private capital and debt.

Through mid-August 2026, U.S. corporate bond issuance had reached $1.68 trillion, with technology and AI infrastructure contributing to the surge in long-term borrowing. Reuters notes that the expansion has intensified concerns about the sheer volume of capital being committed to AI infrastructure.

This does not mean an AI financial crisis is inevitable.

But it does create a difficult question.

What happens if infrastructure investment continues to grow faster than the revenue generated by the technology it supports?

The companies building the AI future are making a long-term bet.

They are betting that demand for artificial intelligence will justify billions of dollars in chips, data centres, power contracts and network infrastructure.

Perhaps they will be right.

But infrastructure cannot be built on optimism alone. At some point, the revenue must justify the investment.

The AI race is therefore also a financial race: who can finance the infrastructure long enough to survive the gap between investment and returns?

Kenya and the African AI paradox

This global question has a particularly powerful example in Africa.

Kenya had been expected to host a landmark $1 billion data-centre project involving Microsoft and UAE-based technology company G42. The facility was planned for Olkaria, drawing on Kenya's geothermal resources and expanding cloud capacity in East Africa.

But the project ran into a basic problem: power at the scale required.

Reuters reported that negotiations over the project encountered difficulties, including disputes involving guaranteed capacity payments and the enormous power requirements of the proposed facility. Kenya later clarified that the project had not been formally cancelled, but its structure and scale remained under discussion.

The controversy became a symbol of something larger.

The issue was not that Kenya lacked ambition.

Kenya has ambition, digital talent, geothermal resources and a clear interest in becoming a regional technology hub.

The problem was that an AI infrastructure project can demand electricity at a scale that changes the national conversation.

That is the African AI paradox.

The continent is being told that AI represents an opportunity to leapfrog development.

But the machines that power the leapfrog must first be powered.

The new AI divide

The world has spent years talking about a digital divide.

Who has internet access?

Who has smartphones?

Who has access to advanced technology?

The AI era is creating a different divide.

Who has enough electricity?

The countries that dominate AI may not simply be those with the most advanced software.

They may be those capable of rapidly building:

- new generation capacity - stronger electricity grids - renewable and nuclear infrastructure - transmission networks - water and cooling systems - financial systems capable of funding massive capital projects

This is why AI policy can no longer be separated from energy policy.

A country cannot simply announce an AI strategy, invite a hyperscale investor and assume the rest will take care of itself.

The data centre must connect to something.

That something is the physical economy.

The danger for Africa

Africa should not respond to the AI boom by giving up.

But neither should governments confuse a data-centre announcement with technological transformation.

The greatest risk would be to subsidise AI infrastructure without building the energy systems needed to sustain it.

That could create an uncomfortable competition between digital infrastructure and public development.

The question would no longer be simply whether a country can attract an AI investor.

It would be:

Can the country provide the electricity without weakening power availability for homes, hospitals, businesses and industry?

That is why the Kenya case matters.

It is not an argument against data centres.

It is a warning against treating AI infrastructure as if it exists independently of national infrastructure.

AI must be built with energy

The solution is not to stop the AI revolution.

The solution is to build the physical infrastructure alongside it.

For Africa, that means governments and investors should pursue AI infrastructure together with:

- dedicated new generation capacity - long-term power-purchase agreements - geothermal, solar, hydro and other locally appropriate energy sources - stronger national and regional grids - phased data-centre development rather than oversized projects disconnected from current capacity - cross-border power markets - clear rules preventing critical infrastructure from undermining public electricity access

The most successful AI economies may not be those that build the biggest data centres first.

They may be those that first build the power systems capable of supporting them.

The bottom line

Artificial intelligence is changing the world.

But the world is now discovering that intelligence has an infrastructure bill.

It has to be financed.

It has to be generated.

It has to be transmitted.

And, ultimately, it has to be paid for.

The AI race is therefore not just a race for better algorithms, better chips or better chatbots.

It is a race for electricity.

And for Africa, the lesson from Kenya is not that the continent should abandon the ambition to become part of the AI revolution.

It is that Africa must stop treating AI infrastructure and energy infrastructure as separate conversations.

The countries that build both together will be able to participate in the next technological revolution.

Those that do not may discover that they have the ambition to host the future—but not yet the power to turn it on.

By Alain Nzeyimana Founder & CEO | Verifyr