“They call it the cloud because ‘industrial-scale electricity and water consumption, fossil-fuel emissions and taxpayer-subsidized infrastructure’ doesn’t fit on a marketing brochure. If AI can turn trillions of dollars into private wealth, it can damn well pay the environmental bill for creating it. Don’t let the richest industry on Earth privatize the profits and socialize the pollution—tax the machines, price the water, clean the grid, and put the money where humanity actually needs it: curing disease, feeding people, educating children and defending a planet that doesn’t have a delete button.”
-adaptation-guide
THE AI DATA-CENTRE NIGHTMARE: TAX THE MACHINES BEFORE THEY TAX THE PLANET
The tech industry wants us to believe artificial intelligence is weightless, frictionless and almost magical. It isn't. Behind every “simple” AI prompt sits a physical industrial machine consuming electricity, water, land, steel, concrete and—too often—fossil fuels. And someone is going to pay for that. The only question is whether it will be Big Tech or everyone else.
There is a dirty little secret hiding behind the shiny promise of artificial intelligence:
AI is not in the cloud.
The “cloud” is a gigantic industrial facility full of computers, transformers, cooling equipment, backup generators, transmission lines and warehouses of hardware.
It consumes electricity.
It consumes water.
It consumes land.
It produces heat.
And increasingly, it produces carbon.
And while the corporations building this infrastructure are valued in the trillions of dollars, much of the environmental bill is quietly being pushed toward the public.
That isn't innovation.
That's privatized profit with socialized consequences.
So here's an idea that Silicon Valley will absolutely hate:
Tax AI.
Not because AI is inherently evil.
Not because technology should be stopped.
Not because we should smash the servers and return to typewriters.
Tax it because nothing this enormous should be allowed to operate as though its environmental footprint is somebody else's problem.
THE NUMBERS ARE GETTING HARD TO IGNORE
Let's start with the number that should make every electricity regulator, politician and homeowner sit up straight.
Lawrence Berkeley National Laboratory's June 2026 update estimates that U.S. data centres could consume 11.8% of all U.S. electricity by 2030.
And that's the middle scenario.
The modelling range runs from 9.5% to 15.3%. The reference case translates into roughly 649 terawatt-hours of electricity in 2030.
Read that again.
Up to roughly one-sixth of America's electricity could be consumed by data centres.
This isn't a website.
This isn't an app.
This isn't something floating around in cyberspace.
This is an industrial electricity load approaching the scale of an entire national economy.
And the International Energy Agency expects global data-centre electricity consumption to roughly double from about 485 TWh in 2025 to around 950 TWh in 2030. AI-focused data centres are projected to triple their electricity consumption over that period.
The AI revolution has discovered an inconvenient truth:
Algorithms need atoms.
Lots of them.
AND THEN THERE IS THE WATER
Electricity gets most of the attention.
Water doesn't.
It should.
Servers generate enormous amounts of heat. Cooling them can require substantial quantities of water, depending on the technology, climate and facility.
Berkeley Lab's earlier U.S. modelling estimated direct data-centre water consumption could reach approximately 145–275 billion litres annually by 2028 under its low/high scenarios.
And that doesn't tell the whole story.
There is also indirect water consumption associated with generating the electricity that powers the centres.
In other words:
The water footprint doesn't stop at the data-centre fence.
Globally, research cited by Berkeley Lab estimates annual water consumption associated with data centres and AI could reach 4.2–6.6 billion cubic metres by 2027—roughly four to six times Denmark's annual water demand.
At precisely the moment humanity is discovering that freshwater is one of its most strategically important resources, we are building an industry that can require vast quantities of it to manufacture something called the “cloud.”
The irony would be hilarious if it weren't so serious.
THE CARBON BOMB UNDER CONSTRUCTION
Now comes the part that should make every climate pledge from Big Tech uncomfortable.
The Financial Times analysed 60 of the largest U.S. data centres being developed by Amazon, Microsoft, Google and Meta.
If they operate using the current U.S. power mix, the analysis estimates they could collectively generate approximately 101.5 million tonnes of CO₂ every year.
That's equivalent to:
27 coal-fired power plants
or
24 million gasoline-powered cars.
The estimated emissions equal roughly 7% of U.S. power-sector emissions in 2025.
Stop for a second.
Sixty data centres.
Not every data centre in America.
Not every AI facility on Earth.
Just 60 of the biggest planned U.S. projects.
And the potential annual carbon footprint is measured in more than 100 million tonnes.
That isn't a rounding error.
That's an industrial emissions story.
“BUT WE'LL USE RENEWABLES!”
Excellent.
Do it.
Seriously.
Nobody should object to enormous investments in solar, wind, storage, transmission, geothermal and nuclear power.
But here's the problem:
The servers are arriving faster than the clean electricity infrastructure can be built.
The IEA says natural gas currently supplies more than 40% of electricity consumed by U.S. data centres, compared with about 24% renewables, 20% nuclear and 15% coal.
And the IEA expects natural gas to be the largest source of additional U.S. data-centre electricity through 2030, adding more than 130 TWh of annual generation.
Why?
Because gas plants can be built comparatively quickly.
The AI industry doesn't want to wait for the energy system.
It wants the energy system to wait for it.
That's the problem.
THE AI INDUSTRIAL COMPLEX HAS DISCOVERED THE MAGIC WORD: “URGENT”
Everything is urgent when Big Tech wants something.
We need the data centres.
We need the GPUs.
We need the transformers.
We need the transmission lines.
We need the gas turbines.
We need the land.
We need the water.
We need the subsidies.
We need the tax breaks.
We need expedited permitting.
We need government support.
We need it now.
But apparently the climate crisis can wait.
The water crisis can wait.
The grid upgrades can wait.
The communities can wait.
The homeowners facing higher electricity bills can wait.
Everybody can wait except the AI industry.
That's not a sustainable industrial policy.
That's corporate impatience masquerading as technological destiny.
AND PEOPLE ARE STARTING TO SAY: ENOUGH.
This isn't merely environmental activism anymore.
It's becoming mainstream public opinion.
A 2026 Gallup survey found that 71% of Americans oppose constructing AI data centres in their local area, including 48% who strongly oppose them.
Only about a quarter favour them.
And 70% say they worry either a great deal or a fair amount about the environmental impact of AI data centres.
That number should terrify the industry.
Because the AI industry's greatest infrastructure problem may not ultimately be GPUs.
It may be public consent.
HERE'S THE QUESTION NOBODY WANTS TO ASK
If AI corporations are collectively worth trillions of dollars...
If investors expect enormous returns...
If AI is supposedly going to revolutionize everything...
why shouldn't the industry pay the full environmental cost of building its machines?
Why should ordinary households subsidize new substations?
Why should taxpayers subsidize infrastructure?
Why should communities surrender water?
Why should electricity customers absorb higher rates?
Why should society inherit the carbon emissions?
Why should future generations inherit the consequences?
And why, exactly, should the world's most profitable corporations receive a special environmental discount?
No comments:
Post a Comment