PART TWO......
WELCOME TO THE AI TAX
I'm not talking about banning AI.
I'm talking about something much more radical:
Make AI pay its own way.
An AI/data-centre tax could be structured around several components:
1. Electricity consumption
The more electricity a facility consumes, the more it pays.
2. Carbon intensity
If your electricity comes from gas or coal, your tax goes up.
If you genuinely add new zero-carbon generation, your tax goes down.
3. Water consumption
If your facility consumes scarce freshwater, you pay for the resource.
4. Grid infrastructure
If taxpayers have to finance substations, transmission lines or grid upgrades specifically because of your enormous electricity demand, you pay the incremental cost.
5. Local environmental impact
Communities should receive compensation for land, water, noise, heat, infrastructure and other externalities.
6. Windfall/excise taxation
If AI companies are extracting extraordinary profits from a resource-intensive business, a modest revenue or profit-based levy can capture some of that economic rent.
That's not anti-business.
That's capitalism with the bill attached.
VIRGINIA JUST SHOWED THAT IT CAN BE DONE
Virginia has already moved in this direction.
Beginning July 1, 2026, the state imposed a $0.011-per-kWh electricity consumption tax on certain data-centre operators.
The law caps the amount deposited into the state's general fund at $600 million per fiscal year, with additional revenues subject to specific provisions in the legislation.
Is that enough?
Probably not.
But it destroys one of the industry's favourite arguments:
“It can't be done.”
Apparently it can.
The question is whether governments have the political courage to do it at the scale required.
AND WHERE SHOULD THE MONEY GO?
Now we're getting interesting.
Because an AI tax shouldn't simply disappear into some government's general operating budget.
Put it to work.
Imagine billions flowing annually into:
Cancer research.
New treatments.
Clinical trials.
Early detection.
Precision medicine.
Food security.
Regenerative agriculture.
Drought-resistant crops.
Irrigation efficiency.
Food-waste reduction.
Clean energy.
Grid modernization.
Storage.
Transmission.
Community solar.
Nuclear.
Geothermal.
Climate adaptation.
Flood protection.
Wildfire prevention.
Heat shelters.
Cooling centres.
Water infrastructure.
Early-warning systems.
Education.
Teachers.
Schools.
Scholarships.
Digital literacy.
Scientific research.
Vocational training.
Workers displaced by automation.
Retraining.
Income support.
New industries.
Apprenticeships.
Transition programmes.
THINK ABOUT THAT TRADE-OFF
We are currently engaged in an extraordinary experiment.
We are taking enormous quantities of electricity, water, capital and human talent and pouring them into computational infrastructure.
Some of that will produce extraordinary breakthroughs.
Absolutely.
AI could accelerate drug discovery.
It could improve weather forecasting.
It could optimize electrical grids.
It could help farmers.
It could help scientists.
It could save lives.
That's precisely why we shouldn't allow the industry's physical costs to become an afterthought.
The question isn't:
“AI or no AI?”
That's a childish question.
The real question is:
“What kind of AI economy do we want?”
One where the winners privatize the gains while taxpayers absorb the externalities?
Or one where the industry pays for the resources it consumes and society decides how the proceeds are invested?
THE “BUT AI WILL SAVE THE PLANET” DEFENCE
Maybe it will.
Maybe AI will discover technologies that transform energy, medicine, agriculture and climate adaptation.
Fantastic.
Then prove it.
Because promising future benefits does not give an industry a licence to dump present-day costs onto everyone else.
The fossil-fuel industry made promises too.
The chemical industry made promises.
The plastics industry made promises.
The automobile industry made promises.
Industrial civilization has a spectacular history of discovering that “the technology will solve it later” is a very convenient sentence when someone else is paying the bill today.
AND HERE'S THE MOST UNCOMFORTABLE PART
The environmental cost isn't the only externality.
There is also the economic question.
What happens if communities spend billions preparing infrastructure for an AI boom...
and the AI boom doesn't deliver the expected returns?
The Financial Times reports that global data-centre investment could reach $7 trillion by 2030, while investors and insurers are increasingly scrutinizing financing, demand, technology-obsolescence and stranded-asset risks.
Imagine a community builds the roads.
The utility builds the transmission.
The taxpayers provide incentives.
The electricity grid expands.
The water system expands.
Then the technology changes.
Or AI demand disappoints.
Or efficiency improves dramatically.
Or another computing architecture replaces today's infrastructure.
Who owns the stranded assets?
And who pays?
THE WALLET IS YOUR WEAPON
Here's where you come in.
You don't have to be a computer scientist.
You don't have to understand transformer architecture.
You don't have to know what a GPU does.
You don't even have to hate AI.
You simply have to ask one question:
“Who pays?”
When a company builds an enormous data centre:
Who pays for the electricity?
Who pays for the water?
Who pays for the grid?
Who pays for the pollution?
Who pays for the carbon?
Who pays when the local electricity bill rises?
Who pays when a drought hits?
Who pays when the river runs low?
Who pays when the cooling system needs another billion litres of water?
Who pays when taxpayers subsidize the infrastructure?
And finally:
Who gets the profits?
If the answer is:
“The corporation gets the profits and the public gets the environmental bill,”
then we don't have an AI revolution.
We have an old-fashioned corporate extraction machine wearing a futuristic costume.
THE BILLION-DOLLAR QUESTION
There is nothing inherently wrong with making money.
There is nothing inherently wrong with AI.
There is nothing inherently wrong with building data centres.
There is everything wrong with pretending that industrial-scale infrastructure has no industrial-scale consequences.
Amazon itself says its absolute carbon emissions increased 6% in 2024, even while its carbon intensity fell.
Microsoft says its total emissions were 23.4% above its 2020 baseline, citing growth factors including AI and cloud expansion.
Those aren't climate activists making those admissions.
Those are the companies themselves.
And the climate doesn't care whether an emission was generated by a coal plant, a server farm or a beautifully designed corporate sustainability report.
A tonne of CO₂ is a tonne of CO₂.
THE AI DATA-CENTRE NIGHTMARE ISN'T HERE TOMORROW
It's being built right now.
The cranes are already there.
The transformers are already being ordered.
The gas turbines are already being planned.
The water systems are already being designed.
The electricity demand is already appearing in utility forecasts.
And the financial commitments are already reaching staggering proportions.
The technology may be revolutionary.
But the infrastructure underneath it is brutally physical.
Concrete. Steel. Copper. Water. Electricity. Land. Fossil fuels.
That's the real AI.
SO HERE'S THE DEAL
Don't ban it.
Tax it.
Don't demonize innovation.
Price its externalities.
Don't prevent companies from becoming extraordinarily wealthy.
Make extraordinarily wealthy companies extraordinarily responsible for the damage associated with their business model.
Don't stop building clean energy.
Make AI finance the clean energy it requires.
Don't tell families to conserve water while corporations get industrial exemptions.
Put every major water user under the same principle: scarcity has a price.
Don't tell homeowners to turn down their thermostats while data centres consume an ever-larger share of the grid.
Make the largest new electricity loads pay their fair share of grid expansion.
And don't let politicians hand out billions in subsidies while telling taxpayers there isn't enough money for schools, hospitals, climate adaptation and scientific research.
BECAUSE HERE'S THE GREAT IRONY
The AI industry tells us it can optimize everything.
Fine.
Let's optimize the priorities.
Would you rather spend the next billion dollars:
building another server farm?
Or:
developing a cancer treatment?
Feeding children?
Training teachers?
Building wildfire defences?
Protecting drinking water?
Hardening the electricity grid?
Preparing cities for deadly heat?
Developing drought-resistant agriculture?
Educating the next generation?
Perhaps the answer isn't always either/or.
But that's precisely why we need the debate.
Because capital is finite.
Electricity is finite.
Fresh water is finite.
Land is finite.
Public patience is finite.
And the planet is definitely finite.
THE FINAL PROMPT
Big Tech loves asking us to interact with AI.
So here's one prompt I'd like every politician, regulator and citizen to enter:
“Calculate the full environmental and social cost of this data centre. Then send me the bill.”
And if the answer comes back:
“The public pays.”
Hit REGENERATE.
YOUR WALLET. YOUR PURSE. YOUR POWER.
You don't have to wait for Silicon Valley to discover ethics.
You don't have to wait for politicians to grow a backbone.
You don't have to become an AI expert.
You have something every corporation ultimately understands: purchasing power.
Ask questions.
Challenge subsidies.
Support politicians who demand full-cost accounting.
Support companies that build genuinely additional clean energy.
Question businesses that hide behind green accounting while their absolute emissions rise.
And when you spend your money, remember:
Every dollar is a vote.
Every purchase tells a corporation what you tolerate.
Every subscription tells a company what business model you will finance.
Every investment tells markets what future you believe in.
Every election tells governments whose interests matter.
So use the weapon sitting in your hand.
YOUR WALLET. YOUR PURSE.
Use it.
Because the most dangerous thing about the AI data-centre boom isn't that machines are getting smarter.
It's that we may be allowing the machines to become richer while the planet becomes poorer.
And if the AI industry really believes its technology is going to build a better future—
then it should have no problem paying for one.
Tax the electricity.
Tax the carbon.
Price the water.
Make the grid upgrades pay-as-you-grow.
End sweetheart subsidies that socialize infrastructure costs.
Invest the proceeds in humanity.
Because if AI can generate trillions in private wealth, surely we can demand that it generate something else:
A public dividend.
Not AI at any cost.
AI worth the cost.
yours truly,
Adaptation-Guide
