More World, Fewer Things
AI uses electricity, water and materials. It can also help us need fewer cars, fewer new buildings and less expansion. What comes into view when we look beyond the data centre?
When an airport runs out of space, it is natural to picture a building site. More square metres, larger facilities, another extension. The passengers have to be somewhere, and their suitcases have to pass through the system.
On 28 October, Ingeniøren described another part of the answer at Copenhagen Airport. The airport has built a data platform intended to host AI solutions for, among other things, making better use of baggage belts and understanding passenger patterns. Passenger numbers are growing faster than the physical facilities can keep up. That means using the existing infrastructure more intelligently.
A suitcase is still a suitcase. Artificial intelligence makes it neither smaller nor weightless. But how much space a facility needs also depends on how the suitcases move through it, and where they pile up.
I think that idea holds a much larger possibility. We talk a great deal about the resources AI requires. The electricity, the water for cooling, the materials for chips and data centres. When the technology is called a waste of resources, its own consumption can easily become the whole story. Everything it can help us get more out of elsewhere disappears from view.
A data centre can be photographed. The office building we never need to construct cannot.
The abundance standing idle
A car can be expensive, heavy and full of sophisticated technology and still spend much of its life without moving. It sits outside the home while its owner sleeps, and at the workplace while its owner works. Meanwhile, someone else may need a way to get somewhere.
That is one of the big ideas behind shared self-driving cars. If a car can pick up the next passenger while the first is at work, the same car can meet several people’s transport needs. With a fleet organised for that purpose, there is an opportunity to use fewer cars and less space for parking.
What matters is the whole arrangement the car can be part of. It needs to know who needs a ride, where they are going, and when the car needs to be available again. The ability to drive is one part of the opportunity. Coordination is another.
A city contains many similar situations. A classroom may stand empty while an association looks for somewhere to meet. A machine in a workshop may be waiting for its next task while someone elsewhere considers buying one like it. A city may have plenty of something that is inaccessible to the person who needs it.
Imagine a choir trying to find somewhere to rehearse. A list of available rooms is a start. But the choir needs a room with space for people and sound, access at the right times, and an arrangement that will still hold next week. A basement room with free hours is not worth much if half the singers cannot get down the stairs.
An AI could help understand the needs, find suitable options and put arrangements in place. It could learn from which rooms actually get used, and what makes an arrangement work. Sharing would then become easier both for those lending a space and for those coming through the door.
In Pattern Sense, I wrote about all the information we have collected without getting very much out of it. Here, that learned capability can have a physical effect. Patterns in the use of a room can become more life within it.
The empty room has already been built. The concrete has been poured and the roof put on. Every time a new activity can find space there, more becomes possible with what we have already paid for in money, materials and energy.
We do not all need to arrive at once, either
Sometimes there is no spare capacity at the very moment we need it. The train is full at eight. Traffic has ground to a halt. Everyone needs to arrive at roughly the same time, and afterwards the pressure eases again.
When infrastructure has to cope with peak demand, a few busy hours can determine how much we build and keep ready for the rest of the day. So there is also an opportunity in changing when the need arises.
AI could make more kinds of work less dependent on everyone being present at the same time. Some of this already exists in a simple form. Microsoft Teams can recap meetings with AI-generated notes and suggested follow-up tasks. An employee can return to what matters without first asking a colleague to recount the whole thing.
If an AI can also keep track of a task, find the necessary information and prepare the work for the next person, fewer handovers may require a shared moment. One colleague could then start early and another later, while the work still holds together.
It is worth following that possibility all the way out onto the road. A changed workflow inside a business can become a different departure time from home. More such choices can spread the pressure on the train, the road and the office. Freedom at work can have an effect in the physical world outside.
That requires flexibility to become real control over one’s day. If AI merely makes it possible to send more tasks at all hours, we have acquired a longer working day. The opportunity here is to arrive later, pick up a child or do some of the work somewhere else without collaboration falling apart.
We can therefore get more out of the same infrastructure both by sharing it and by spreading demand across more times of day. A road can have more room because fewer people need to be on it at eight.
The accounting beyond the data centre
In its 2025 report Energy and AI, the International Energy Agency was already examining how sensors and AI-based control can release more capacity in existing power lines. With widespread adoption, the agency estimated potential for up to 175 gigawatts of additional transmission capacity without new lines. This concerns how much power the grid can carry, and making better use of facilities that are already there.
It is the same movement as at the airport. A physical limit can sometimes be shifted by understanding and managing what already exists more effectively.
In Knowledge Explosion, I explored how knowledge can change the constraints we work within. What looks like a fixed shortage of capacity can also be a lack of knowledge about how that capacity can be used. That possibility extends far beyond the computer itself.
If we want to understand what AI means for our overall resource consumption, we therefore need to follow the technology out into the systems it changes. A computation uses electricity. It may also make it possible to postpone an expansion, use less energy for a task or meet more people’s needs with fewer physical things.
That is a different question from whether the next chip becomes more efficient. It concerns the whole world around the chip. How much do we need to produce, transport and build to have the everyday life we want?
The building that becomes unnecessary gets no spectacular opening ceremony. No press photographer turns up to photograph the materials that do not need to be extracted. A need has been met in another way, and the building site remains something else.
When the debate follows only what AI itself consumes, it becomes blind to that kind of progress.
Room for a spontaneous idea
There is something valuable about a car waiting just for you. It gives you the freedom to change plans. An unbooked room may be the place where something unexpected is allowed to begin.
That freedom needs to be part of our idea of better utilisation. A society where every chair is reserved and every machine is in use all the time struggles to accommodate a spontaneous idea. When everything runs right at its limit, even the smallest delay can spread to what comes next.
A good shared solution therefore needs breathing room. It can keep some capacity free, find another room when the first is needed for something else, and make it easy to change an arrangement. AI’s ability to coordinate can be used to preserve the freedom that otherwise makes us want to own one of everything ourselves.
Likewise, fewer cars are interesting if people can still get where they need to go. A freed-up parking space can become trees, a playground or somewhere to sit. The benefit takes on meaning through the life it makes room for.
In The Revolution Beneath the Surface, I wrote about technological changes that first alter the connections beneath the familiar world. Here, it may be the connection between a person’s need and a place, a car or a piece of equipment that already exists.
We are used to imagining progress as something new being added. A bigger road. Another building. More things within reach.
There is also a future in which more people gain access even as we build less. Where work gives us the freedom to take a different train, and where the choir’s rehearsal evening begins with someone opening the door to a room that has been there all along.