“AI shouldn’t be compared to the internet or mobile. It’s much more akin to the discovery of electricity or fire.”1
— Sir Demis Hassabis
There are certain moments in history when a single sentence quietly changes how you see the world.
This was one of them.
When Sir Demis Hassabis, CEO of Google DeepMind, said that artificial intelligence should be compared not to the internet or smartphones, but to electricity or fire, it sounded provocative.
Most people heard another bold prediction from one of the world’s leading AI researchers.
I heard something else.
I heard a framework.
Because if he’s right, then almost every conversation businesses are having about AI is asking the wrong question.
Coincidentally, this week Europe announced it will invest €10 billion in building seven AI Gigafactories, with the ambition of attracting a further €20 billion of private investment. Each facility is expected to house around 100,000 advanced AI chips, creating computing infrastructure on a scale the continent has never seen before.2
Pause for a moment.
Governments don’t build gigafactories because a new app has become popular.
They build them because something has begun to resemble national infrastructure.
And that changes everything.
We keep comparing AI to the internet.
I think that’s a mistake.
The internet connected people.
Smartphones connected people to the internet.
Social media connected people to each other.
Each represented a remarkable technological leap.
But fundamentally they remained products.
Electricity wasn’t a product.
Electricity became the foundation upon which countless other products could exist.
That distinction matters more than it first appears.
When electricity arrived, nobody asked whether restaurants should become “electric-first”.
There were no conferences asking whether accounting firms should “embrace electricity.”
There were no consultants explaining the “electricity strategy.”
Instead, electricity quietly infiltrated everything.
Factories changed.
Medicine changed.
Agriculture changed.
Architecture changed.
Entertainment changed.
Education changed.
The winners weren’t necessarily the companies that generated electricity.
They were the organisations that redesigned themselves because electricity existed.
History rarely rewards those who merely adopt technology.
It rewards those who reorganise around it.
We’re beginning to see exactly the same pattern emerging with AI.
Not in ChatGPT.
Not in image generation.
Not in coding assistants.
Those are simply the visible edge of something much larger.
The real story is happening underneath.
Consider another statistic.
Gartner now forecasts that global data centre electricity consumption will rise 26% during 2026, reaching 565 terawatt-hours, with AI-optimised servers expected to consume almost one-third of all data centre electricity next year. By 2027, AI servers will consume more power than conventional servers.3
Read that again.
One of the biggest conversations surrounding artificial intelligence is no longer algorithms.
It’s electricity.
Think about how extraordinary that is.
Within only a few years, one of humanity’s most exciting software revolutions has become constrained not by software developers, but by power grids.
Perhaps that’s exactly what Hassabis meant.
Infrastructure technologies eventually stop competing with one another.
Instead, society begins reorganising itself around them.
Railways changed where cities were built.
Motorways changed retail.
Electricity changed manufacturing.
The internet changed communication.
Artificial intelligence may change how knowledge itself is produced.
This explains something that has puzzled me for months.
Every company now claims to have an AI strategy.
Almost none can explain what that actually means.
Most strategies amount to buying software licences.
Some encourage employees to use ChatGPT.
Others automate customer service.
These are useful improvements.
But they feel remarkably similar to replacing candles with electric light bulbs.
Necessary.
Sensible.
Entirely missing the bigger transformation.
The organisations that benefited most from electrification weren’t those that illuminated old factories.
They were those that redesigned factories altogether.
Henry Ford didn’t become revolutionary because electricity existed.
He became revolutionary because he redesigned production around its possibilities.
Technology was the enabler.
Organisation was the innovation.
That lesson feels remarkably contemporary.
Today’s executive meetings are filled with questions like:
“Which AI platform should we use?”
“Should we build or buy?”
“Can we automate customer support?”
Those questions aren’t wrong.
They’re simply too small.
The better question is:
What happens when intelligence becomes abundant?
Every major technological revolution dramatically reduces the cost of something.
Steam reduced the cost of mechanical power.
Electricity reduced the cost of energy distribution.
The internet reduced the cost of communication.
Artificial intelligence appears to be reducing the cost of cognition.
Research.
Translation.
Analysis.
Drafting.
Pattern recognition.
Decision support.
Knowledge itself is becoming cheaper.
If that’s true, then the organisations designed for expensive knowledge work may no longer be optimally designed.
That isn’t a software problem.
It’s an organisational one.
There’s another clue hidden in this week’s headlines.
As AI infrastructure expands, governments are no longer debating chatbots.
They’re debating electricity markets, grid capacity, water use and national competitiveness. The United Nations has even called for data centres to run on renewable energy by 2030, acknowledging that AI’s rapidly growing electricity demand has become a global infrastructure issue.4
Notice how quickly the conversation has shifted.
Only a few years ago we wondered whether AI would write poems.
Now governments are asking whether there will be enough electricity to power it.
That isn’t the trajectory of another software trend.
It’s the trajectory of infrastructure.
This changes how I think businesses should respond.
The question is no longer:
“How do we use AI?”
Eventually everyone will.
That’s like asking whether businesses should use electricity.
The competitive advantage disappears once adoption becomes universal.
Instead, leaders should ask something much more uncomfortable:
If intelligence becomes effectively abundant, which assumptions about our business stop making sense?
Which meetings disappear?
Which hierarchies flatten?
Which skills become more valuable?
Which become less scarce?
Which decisions move closer to customers?
Which bottlenecks vanish entirely?
Those questions are harder.
They’re also far more important.
I suspect that twenty years from now we won’t describe companies as “AI companies.”
Just as we don’t describe successful organisations today as “electricity companies.”
The technology will simply become part of the operating environment.
Invisible.
Expected.
Essential.
The organisations that thrive won’t be remembered because they adopted AI first.
They’ll be remembered because they understood what abundant intelligence allowed them to become.
Perhaps that’s the real lesson behind Hassabis’ observation.
Every generation experiences technologies that feel revolutionary.
Only a handful become civilisation’s operating system.
Fire did.
Electricity did.
The internet arguably did.
Artificial intelligence may become the next.
If so, history won’t remember this period as the beginning of the AI industry.
It will remember it as the moment intelligence itself became infrastructure.
And that is a very different story.