“While technological successes are celebrated, the social fabric is slowly being eroded, as if by a silent virus.” — Pope Leo XIV
Announcements, events, and more from Tyee and select partners
Do you think there is still work for human journalists to do?
Only if the people who read it support it. And then comes AI.
Contest: Win tickets to the Vancouver Folk Music Festival
One lucky Tyee reader will win two weekend passes worth $500.
The artificial intelligence industry prefers to refer to large “hyperscale” data centers as “campuses.” That’s complete bullshit. There are no students, no laughter, no libraries in the data center. The poet William Blake would have given these ugly bunkers the apt name “Dark Devil’s Factory.”
Three trillion dollars of global capital is currently flowing into this vast vortex, much of it flowing from bond markets, private credit, and government programs that uncritically view dangerously flawed technologies as miracles, inevitabilities, and necessities. This means trillions of dollars are not being spent to combat the cost of living crisis and the climate storm.
The concentration of capital in AI is largely directed by and for secret cartels of billionaires, creating job deserts. According to one estimate, it takes $54 million in capital investment to create one permanent job in the data center industry, compared to $322,000 in capital investment to create one full-time job in other sectors. Not to mention the hundreds of millions of jobs that AI promises to make obsolete.
This circulation sucks not only capital, but also vast energy supplies. There are currently more than 11,000 data centers on the planet, half of which are in the United States. Installing one 100 megawatt hyperscale facility requires enough power to light 100,000 homes. A large-scale AI factory planned for New Zealand would consume 280 megawatts, or 6% of the country’s electricity needs. And the power consumption of the facility will be more than five times smaller than the giants planned to build around the world.
Already in Ireland, the data center industry consumes 22 per cent of the island’s electricity. This is more than all urban households in the country. It appears that the AI giant can create silence in civil servants. The Irish government’s recent report on the importance of data centers happened to omit any mention of the proliferation of AI and how it is driving up electricity prices, which are among the highest in Europe.
Now consider the energy appetite of Kevin O’Leary’s Wonder Valley Project in Alberta. O’Leary is proposing to funnel $70 billion of other people’s money into the “largest” data center on the planet. That would use more electricity than 8 million homes, or 15 cities the size of Calgary. Its voracious energy appetite links this robotic entity directly to the Montney formation in western Canada, and its crushed methane is used to ignite the turbines that power the colossus.
If history is any guide, these turbines will pollute the air by emitting lung-clogging particulate matter and harmful chemicals, shortening the lifespans of rural people. Alberta Premier Daniel Smith’s government has already abandoned any pretense of an environmental review.
hungry for land and water
This gyre not only swallows capital and energy, but also forests, grasslands, suburbs, and farms in its destructive spiral. Data center footprints continue to grow along with the demand for computing power, and hyperscale factories now typically span more than 500 acres, which is 3 1/2 times the size of Disneyland. But what happens if regulators stipulate that data centers “green” by requiring power to come from utility-scale solar or wind power? That saves methane, but also impacts more land.
There seem to be no limits. A project in Montana proposes building a building the size of 3,000 football fields in the middle of cattle country. The project will be 31 times the size of the nearest rural community, Broadview. O’Leary’s project would similarly fill about 8,000 acres of parkland south of Grand Prairie. It’s a swath of Alberta larger than Manhattan.
Amidst this land loss is a trove of metals and minerals that require large-scale mining and processing. Copper and aluminum for power and cooling systems. Backup battery is lithium or diesel. Gallium and germanium are semiconductors. According to the World Economic Forum, every megawatt a data center consumes requires 60 to 75 tons of materials, primarily for power and cooling systems. As data bunkers grow larger and consume more energy, this material strength grows like a city.
AI proponents will tell you to relax, because all this mind-numbing consumption will feed the stock market and boost that sacred metric: GDP. But you can’t devote most of your resources to endless stock plays. For example, clean fresh water has its limits. The AI cycle relies on large amounts of water and is already stressing drought-prone regions across North America. By 2028, data centers will consume enough fresh water to meet the indoor water needs of 18 million people in the United States.
The AI actually starts out as a water eater. Semiconductor chip manufacturing requires ultrapure water for production and uses 400 chemicals, including perfluoroalkyl and polyfluoroalkyl substances, which are virtually indestructible carcinogens. According to the researchers, manufacturing the complex chips releases toxic emissions of “many unknown and dangerous contaminants.”
When finished chips are assembled into data center racks, warehouse cooling systems require more water to keep the facility from overheating.
According to JPMorgan, a large data center can consume as much water as a chipmaker, about 19 million liters per day. This is the same amount of water used by a town of up to 50,000 people. Every AI-powered internet search is soaked in a bottle’s worth of fresh water, the world’s most important resource.
Forced into a “heat island”
Heat is expelled as the circulation consumes and digests it. Enough to pose a threat to the cutting edge of man and nature.
It’s no secret that modern megacities have created what scientists call “urban heat islands” by replacing soil and trees with concrete and asphalt. When temperatures reach 29 degrees Celsius in Sydney, Australia, the city center is actually 4 degrees warmer. In Paris, the urban heat island effect could increase average summer temperatures by 2 to 3 degrees Celsius.
What does Canada’s AI strategy mean for jobs and safety?
read more
This heat capture has the perverse effect of burning more fossil fuels in power plants to keep overheated people cool with air conditioning, resulting in more unstable weather, greenhouse gas emissions, and air pollution.
There is now a new cause of the urban heat island phenomenon in the world. Data centers raise temperatures in nearby areas, posing additional threats to local health.
A recent study by scientists from the UK, Singapore, France and Hong Kong found that data centers dissipate large amounts of heat when performing AI calculations, effectively creating their own microclimate zones. Researchers called this the “data heat island effect.”
Researchers used remote sensing data from satellites to locate 6,000 data centers built in rural areas over the past decade and measure temperature changes before and after construction. We found that surface temperatures everywhere have risen by an average of 2 degrees Celsius. In extreme cases, temperatures rose by as much as 9 degrees Celsius.
The researchers found that the warmer temperatures extended up to 10 kilometers away from the AI hyperscale facility. Many of the places they tracked are already quite warm. For example, in Brazil’s northeastern region, where daily highs hover in the low 30s during the hottest months, the data center has increased temperatures in the region by nearly 3 degrees.
Scientists pulled back the lens to consider how many people on Earth live within a 10-kilometre radius of a large AI data center. By their calculations, it’s about 343 million people. Their conclusion is that, “particularly in the context of global warming and climate change,” the data heat island effect “could have dramatic impacts on welfare, health care, and energy systems.”
When the biggest AI data center comes to your city
read more
mounting resistance
Regardless, the vortex rotates with increasing speed and force. Unless power consumption is curbed in some way, data centers will consume more power than all the world’s manufacturing industries in the next five years. According to the United Nations, data center water demand could exceed 9 trillion liters by 2030. This is enough water to quench the thirst of all 1.3 billion people in sub-Saharan Africa.
No wonder Canadians feel as if this vortex of destruction is being sold to them at hastily clipped clips by an incomprehensible Albertan auctioneer. To date, nearly 100 very large data centers have been proposed in Canada. About 90 per cent of these projects will be built in rural Alberta, where the government has relentlessly courted the industry without democratic mandate. Or transparency about environmental costs.
A recent paper from the University of York rightly points out that the rapid expansion of ultra-large data factories represents “tectonic change rather than incremental growth.” Alberta’s data center ambitions could challenge the reliability of the power grid, increase methane pollution and completely destabilize rural areas, he warned. “As a result, planned installations are concentrated in states with moderate to high water risk, which rely heavily on natural gas, and whose grid emission intensity is nearly five times the national average.”
So the AI vortex will eat up capital, land, and energy. Sip the scarce water. Extraction of metals and minerals accelerates as resource scarcity exacerbates geopolitical turmoil. It concentrates economic power in a few hands, most of it in foreign hands. And it accelerates the automation of everything, enabling a totalitarian infrastructure designed to gradually erode human fabric.
Don’t worry. The federal government wants to provide “AI literacy training” to people who aren’t yet fully hooked on AI.
But resistance increases. When residents of Olds, Alberta, learned that Canada’s largest AI data center was slated to be built in a small farming community, they organized their own educational session and sounded the alarm. State regulators halted the project, but backers reapplied.
“These projects are being thrust upon us with no accountability, and communities will be left bearing the consequences,” said Jesse Cardinal, executive director of the Indigenous-led group Keepers of the Water. He could have spoken for all those everywhere who woke up to the threat being ruthlessly peddled.
As Google’s own AI says, “Public resistance to data centers is rapidly growing, with recent polls showing a majority, even working-class neighborhoods, strongly oppose construction in their neighborhoods.” According to Data Center Watch, local opposition has blocked or delayed $64 billion worth of U.S. data center projects.
These citizens are on the front lines of a larger battle pitting human dignity and reason against the swirling void that is AI obsession. 
