Donald Trump recently declared that anyone opposing the construction of massive new artificial intelligence datacenters wants society to "end up being backwards and poor." It is a vintage piece of political theater: loud, binary, and entirely divorced from the physical realities of modern power grids. The lazy consensus says we must pave every county in Virginia and Texas with server racks or risk losing the geopolitical race to Beijing.
That framing is a trap. Discover more on a similar topic: this related article.
The opposition to these facilities is not driven by Luddite nostalgia or a collective death wish for economic stagnation. It is driven by municipal planners, grid operators, and local utility customers staring down the barrel of structural energy starvation. When a single hyperscale facility demands as much electricity as a medium-sized city, the math stops working. Calling local resistance backward ignores the cold engineering truth that our current infrastructure cannot physically handle the load without breaking the power supply for everyone else.
I have spent the last decade watching corporate balance sheets collide with municipal substations. I have seen billion-dollar projects grind to a halt not because of environmental lawsuits, but because the local transformer literally melted under the surge. The debate is not about whether we want progress. It is about whether we are willing to bankrupt our electrical grids to feed silicon monsters that employ thirty people once construction wraps up. Further analysis by Gizmodo explores similar views on the subject.
Let us dismantle the core falsehoods driving the current datacenter frenzy.
The Job Creation Fallacy
Politicians love to stand in front of ribbon-cutting ceremonies for tech campuses and talk about high-paying jobs. The reality inside a modern hyperscale facility looks very different.
Once the concrete is poured and the liquid-cooling loops are sealed, a million-square-foot server farm requires a skeleton crew. You need security personnel, a handful of systems engineers, and facility maintenance technicians. The economic multiplier effect that politicians promise—the bustling downtowns, the roaring local service economies—simply does not materialize around these windowless fortresses. They are capital-intensive, not labor-intensive.
When a community trades its agricultural water rights and local grid capacity for a datacenter, it is making a terrible tax trade. They get a massive industrial asset that drives up real estate values and strains emergency services, paired with an anemic property tax yield relative to the footprint, and almost zero permanent employment upside. Opposing that trade is not being backwards. It is basic financial literacy.
The Grid Math Does Not Care About Your Prompts
The entire artificial intelligence boom rests on an unspoken vulnerability: the American electrical grid is old, fragile, and choking on deferred maintenance.
For two decades, electricity demand in the United States remained relatively flat thanks to efficiency gains in appliances, lighting, and manufacturing. Utilities grew comfortable with modest, predictable load growth of one percent or less per year. Then generative models arrived, and hyper-scalers started demanding blocks of power measured in gigawatts rather than megawatts.
A single rack of modern AI accelerators draws as much power as an entire residential suburban block. Multiply that by tens of thousands of racks per facility, and you create localized demand spikes that regional transmission organizations cannot absorb.
When utilities are forced to prioritize these sprawling compute hubs, residential ratepayers pay the price. Across major power pools from PJM to ERCOT, households are watching their monthly utility bills spike to subsidize the grid upgrades required by private tech conglomerates. If your local power bill doubles so a Silicon Valley firm can train a slightly better chatbot, you are not participating in an economic miracle. You are being taxed for a private corporation's capital expenditure.
The Nuclear Fantasy
Tech giants understand this bottleneck, which explains their sudden, desperate romance with nuclear energy. We are told that small modular reactors or restarted atomic plants will soon solve the power deficit.
Here is the dirty secret the press releases leave out: those reactors do not exist at scale today.
Restarting shuttered facilities like Three Mile Island takes years of regulatory clearance and billions of dollars in retrofits. Building unproven advanced reactors takes even longer. Meanwhile, datacenters are being built right now, drawing power from existing coal and natural gas plants that were scheduled for retirement.
Instead of decarbonizing our grid, the AI boom is actively extending the lifespan of fossil-fuel generation assets. Utilities are keeping coal plants online past their expiration dates simply to keep the servers humming. If your primary metric for progress is carbon reduction, the current trajectory is a catastrophe wrapped in a marketing brochure.
What Intelligent Growth Actually Looks Like
Refusing to rubber-stamp every proposed server warehouse is not anti-technology. It is a demand for structural accountability.
If tech companies want unlimited power, they need to build it themselves, off-grid, and pay the true marginal cost of their expansion without shifting the financial burden onto residential ratepayers. We need localized microgrids, behind-the-meter renewable generation paired with dedicated storage, and strict efficiency standards for model training architectures that currently waste staggering amounts of compute cycles on redundant calculations.
Stop treating grid stability as an enemy of innovation. Without a stable power supply, the entire digital economy collapses anyway. The choice is not between advanced artificial intelligence and living in the dark. The choice is between disciplined engineering and reckless, subsidized expansion that burns out the physical foundation of our economy.
