We Are Building the Body of the Machine
AI is usually talked about as software.
It is not.
It has a body.
Data centres are that body.
Power plants feed it. Transmission lines connect it. Chips process through it. Water cools it. Steel, concrete, generators, substations and fibre hold it together.
And that body is being built quickly.
Governments want AI investment. Utilities are planning around massive new loads. Municipalities are being approached with projects measured in hundreds of megawatts. New generation is being proposed to serve them. Rural lands, industrial lands and Indigenous territories are being examined for places to put them.
The conversation is moving fast because the money is moving fast.
But infrastructure lasts much longer than a technology cycle.
That should change the way we think about what is happening.
We are not simply adopting a new tool.
We are building the physical foundation for AI to become a permanent layer beneath modern life.
That deserves more attention than it is getting.
The Cloud Is Not in the Sky
I have already written about the physical demands of the AI data-centre rush.
The power.
The water.
The land.
The noise.
The transmission.
The emissions.
The public infrastructure.
A one-gigawatt data centre is not a warehouse with a lot of computers inside it.
It is industrial infrastructure operating at the scale of major power generation.
That matters because the language surrounding AI still makes it sound strangely weightless.
Cloud computing.
Digital infrastructure.
Artificial intelligence.
Compute.
None of those words sound like land.
But AI has a physical footprint.
The cloud is not floating somewhere above us.
It is on someone's land.
It draws from someone's electrical system.
It connects to someone's watershed.
Someone hears the generators.
Someone sees the transmission lines.
Someone lives beside the industrial development.
Someone assumes the long-term risk if the economics change.
That is why the current data-centre boom should be understood as more than another technology investment cycle.
We are giving the machine territory.
Not metaphorically.
Physically.
What Are We Building It For?
This is where the conversation becomes larger than water and electricity.
Most public discussion asks whether we have enough power to support AI.
That is the wrong place to stop.
The more important question is:
What are we creating enough power for AI to do?
These facilities will not exist only so people can generate pictures, summarize documents or talk to chatbots.
AI is moving into finance.
Insurance.
Employment.
Education.
Health care.
Government.
Security.
Surveillance.
Identity verification.
Content moderation.
Public administration.
Law enforcement.
Infrastructure management.
Every major institution is exploring how AI can make decisions faster, process more information and reduce human labour.
That sounds efficient.
Sometimes it will be.
But the physical infrastructure being built today creates the capacity for something much larger tomorrow.
It creates the capacity for AI to move from tool to system.
And eventually from system to authority.
That is the part we should be discussing before the concrete is poured everywhere.
Intelligence Is Not the Only Question
Most of the public debate asks how intelligent AI will become.
I am increasingly less interested in that question.
A highly intelligent tool can be useful.
A highly intelligent authority is something else.
The real issue is not whether AI can think faster than us.
The issue is what happens when human institutions begin saying:
The system says.
The model says.
The score says.
The algorithm flagged it.
The automated assessment rejected it.
The data places you in this category.
Those statements create a strange form of authority because responsibility starts disappearing.
Nobody rejected you.
The system did.
Nobody decided you were risky.
The model calculated it.
Nobody chose to reduce your visibility.
The algorithm did.
Nobody made the judgment.
The data produced the result.
That is not science fiction.
Pieces of that world already exist.
AI gives us the ability to expand it enormously.
And data centres give that expansion a body.
The Part We Are Overlooking
The danger is not that AI suddenly becomes evil.
It does not need to.
AI can amplify distortion without possessing intention of its own.
It needs an operator.
A dataset.
A goal.
A definition of success.
A set of assumptions.
Then it can become extraordinarily effective at producing results inside those assumptions.
If the assumptions are good, that can be powerful.
If the assumptions are distorted, intelligence can scale the distortion.
A foolish system fails visibly.
A sophisticated system built on a false premise can produce beautifully organized wrongness.
That should concern us because human institutions already carry distortion.
Governments do.
Corporations do.
Religious institutions do.
Financial systems do.
Political movements do.
Families do.
Every institution contains human priorities, blind spots, incentives and struggles for power.
AI does not remove those things.
It can encode them.
Accelerate them.
Automate them.
Then place distance between the decision and the person responsible for it.
That is why the data-centre rush cannot be treated only as an engineering question.
We are building infrastructure for decision-making power.
An Older Way of Seeing the Problem
The Ancient Way begins with a very different understanding of human life.
Human beings were not originally experienced as isolated categories.
Body.
Creator.
Land.
Community.
Breath.
Food.
Water.
Work.
Rest.
Season.
Prayer.
These belonged to one living relationship. Modern systems separated them into departments because separated things are easier to administer.
That matters here.
AI works through categorization.
It needs measurable information.
That is one of its strengths.
But a human being is always more than what can be measured.
Land is more than acreage.
Water is more than volume.
A meal is more than calories.
Community is more than demographics.
A person's history is more than a dataset.
A human life is relational.
Once the measurable part is allowed to stand in for the whole, something important has already been lost.
That is not an argument against measurement.
It is an argument against confusing the map with the land.
Data Centres Should Trigger a Bigger Public Test
My earlier concern about AI infrastructure remains.
AI companies are moving at market speed.
Electrical systems move at infrastructure speed.
Water moves at watershed speed.
Municipal planning moves at public-process speed.
Communities move at trust speed.
Those speeds do not match.
That creates obvious environmental and economic risks.
But now another question belongs beside them:
What kind of society is this infrastructure being built to support?
Before communities approve massive AI campuses, the public test should extend beyond megawatts and construction jobs.
We should be asking:
Who owns the computational capacity?
Who receives priority access?
What public functions will depend on it?
What decisions will eventually be delegated to AI?
What personal data will those systems require?
What happens when an automated decision is wrong?
Who remains accountable?
Can a human overturn it?
Can a person understand why a decision was made?
Can someone refuse participation without becoming unable to function in ordinary society?
Those are infrastructure questions too.
Because once society becomes dependent on the system, governance becomes much harder.
Build Slowly Enough to Decide
I am not arguing that AI should disappear.
I use it.
It can search enormous bodies of information.
It can identify patterns humans miss.
It can assist science.
Medicine.
Research.
Writing.
Planning.
Accessibility.
There is real value here.
But usefulness does not justify unlimited authority.
And economic opportunity does not justify building first and asking philosophical questions later.
That is the pattern we need to resist.
Build the infrastructure.
Normalize the dependency.
Integrate the systems.
Then debate the limits.
By then the limits are much harder to establish.
We should reverse that sequence.
Decide what AI should never control.
Decide where human judgment must remain.
Decide what rights people retain when an algorithm judges them incorrectly.
Decide what information should never become a condition of participation.
Decide how communities hosting the infrastructure benefit from it.
Decide where land, water, energy and human dignity place boundaries around growth.
Then build inside those limits.
The Machine Is Getting a Body
The AI debate often feels distant because software is invisible.
Data centres change that.
They make the future physical.
Concrete is being poured.
Power is being contracted.
Gas plants are being proposed.
Transmission is being planned.
Land is being committed.
Governments are writing regulations.
Companies are investing billions.
The machine is getting a body.
That does not mean the machine is evil.
It means our decisions are becoming harder to reverse.
The question is no longer whether AI is coming.
It is already here.
The question is what place we are preparing to give it.
Tool.
Advisor.
Infrastructure.
Gatekeeper.
Authority.
Those are not the same thing.
And before we build enough physical capacity to make AI unavoidable in every major system of human life, we should decide where the machine ends and the human being begins.
The body is being built now.
The boundaries should be built first.
