[External Email] The Data Center Bargain: What Every Town Should Demand The Registry May 13, 2026 Free power. Free heat. Clean water. The bargain every town should be demanding. By Billy the Broker (I just want to stay anonymous) By the time you finish reading this sentence, a data center somewhere in Virginia has consumed enough electricity to power your home for a week. By the time you finish this column, it will have dumped enough waste heat into the sky to warm a small village through January. We are, collectively, building the largest network of intentional energy waste in human history—and calling it progress. I have nothing against data centers. I’d be writing this on a stone tablet without them. But the way we’re siting, powering, and operating them is the kind of thinking that gave us strip malls with twelve-acre parking lots: technically functional, aesthetically punishing, and an embarrassment to anyone who’ll have to look at it in fifty years. We can do better. More to the point, we can demand better. Here’s the deal I’d propose to every municipality currently being courted by a hyperscaler: yes, you can build here—on three conditions. Condition one: bring your own power, and then some. A modern data center campus draws somewhere between 15 and 500 megawatts. The honest ones admit they’re straining grids that were designed when “the cloud” meant something you saw out an airplane window. The dishonest ones promise renewable energy certificates from a wind farm in Oklahoma and call it carbon-neutral. Enough. If you want to build a 300-megawatt facility in our town, you bring a 350-megawatt power source with you. A small modular reactor (SMR) on the back lot. A solar array on every available acre. A geothermal loop drilled into whatever’s underneath. I don’t particularly care which, and frankly the engineers will figure it out faster than the lawyers can object. The point is: produce what you consume, plus a surplus for the neighbors. Every household within a 25-mile radius gets free electricity. Not “discounted.” Not “rebated through a complicated utility scheme.” Free. This sounds radical until you do the arithmetic. A single 300-megawatt SMR generates roughly 2.4 billion kilowatt-hours per year. Your average data center campus might use a billion of those. The other 1.4 billion would power something like 130,000 American homes. That’s not a rounding error. That’s a city. A note on the numbers, before the engineers write in. The figures in this column are back-of-envelope estimates based on publicly available data, industry rules of thumb, and the kind of arithmetic you can do on a napkin. Data center power consumption varies wildly—an AI training facility might draw three times what a traditional cloud data center does in the same square footage. SMR economics are projected, not proven; no commercial SMR is yet operating in North America. Heat recovery percentages, district heating radii, and water purification yields all depend on technology choices, climate, and local infrastructure. Treat the specifics as directional, not definitive. The argument doesn’t rest on any single number being exact; it rests on the fact that even if you halve every figure here, the result still beats what we’re doing now. Condition two: the heat is not yours to throw away. Virtually every watt of electricity that enters a server rack leaves it as heat. Currently, we use additional electricity to pump that heat into the atmosphere via cooling towers—a process so wasteful that aliens observing us from a polite distance must assume we’re doing it on purpose as some kind of religious offering. Stockholm figured this out a decade ago. So did Helsinki, Dublin, Copenhagen, and Paris. Data centers in those cities pipe their waste heat into district heating networks that warm tens of thousands of homes. Microsoft’s Espoo campus, when complete, is projected to heat hundreds of thousands of Finns. The technology is mature. The economics work. The only thing missing in America is the will. So: condition two. Every home within practical service distance gets heating and hot water from data center waste heat, delivered through a district loop the developer pays to install. In summer, the same infrastructure runs in reverse for absorption cooling. The marginal cost to the data center operator is rounding error against their capital budget. The marginal benefit to a working family in winter is the difference between paying a heating bill and not. Condition three: clean the water while you’re at it. Data centers are thirsty. A large facility can evaporate millions of gallons annually for cooling, which has made them deeply unpopular in places like Arizona, where water is the local equivalent of cabernet. But here’s the thing: low-grade heat is exactly what you need to drive membrane distillation, multi-effect desalination, and other water purification processes that would otherwise require burning natural gas. A data center co-located with a water reclamation facility, using its own waste heat to purify municipal wastewater or brackish groundwater, isn’t a drain on local resources. It’s a net contributor. The exact throughput depends on the technology and the feedwater—but the principle is sound, and the byproduct is something every drought-stressed county in the Southwest is currently paying through the nose for. The grand bargain. What I’m describing isn’t science fiction. Every piece of it exists today, deployed somewhere in the world, usually somewhere Scandinavian. The barrier isn’t technology. It’s the assumption that data centers are something done to a community rather than for one. Flip that assumption. A data center built on these principles doesn’t just avoid being a burden—it becomes the most valuable piece of civic infrastructure a town has acquired since the post office. Free power. Free heat. Clean water. Property tax revenue. A few hundred well-paid jobs. And, almost incidentally, a place where the cat videos get sorted. The hyperscalers will tell you this is impossible, too expensive, beyond the scope of their business model. They’ll say it for about six months, until the first town actually demands it, at which point they’ll discover they can do it after all—the way they discovered they could do liquid cooling the moment Nvidia’s chips made air cooling impossible. We’ve spent twenty years letting these buildings appear on our landscape as windowless beige monoliths that consume everything and give nothing back. The next twenty don’t have to look like that. They could look like the best deal any small town ever signed. All we have to do is ask.