How AI data centers can drain a community — and what to do
Hakeem Anwar explains why thousands of hyperscale sites threaten water, power and health — and how neighbors are fighting back.
“These AI data centers have the capacity to actually make our wells run dry.” — Hakeem Anwar
Hakeem opened with that line, and it’s not hyperbole: the conversation that follows lays out a chain reaction many people haven’t connected. AI hyperscale sites are not just server rooms. They are massive construction projects with enormous power draws, cooling needs that chew through water, and zoning tricks that often land them in poorer, rural places.
This episode runs through three things that matter to anyone with a homestead, a well, or a vote: the scale of the buildout, the human costs stacking up around the sites, and the practical moves communities are using to stop—or at least slow—new projects.
How big is “big”? the numbers that should wake a town hall
There are already roughly 4,000 data centers in the U.S.; the worry is the new class of AI hyperscalers, which plan to add hundreds more.
If planned projects continue, AI data centers could consume over 40% of today’s U.S. power capacity by 2030, according to the map and methodology Hakeem describes.
Hakeem points to specific megaprojects: a reported 10-gigawatt facility and other “gigascaler” sites that use power equivalent to millions of homes.
Those figures have two immediate consequences: higher electricity bills (Bloomberg’s work shows neighborhoods near clusters already pay roughly three times more), and enormous water demand because cutting-edge AI processors require continuous, water-based cooling at scale.
Not just spreadsheets: health, land and local economies
Hakeem and TD walk through the collateral damage beyond the bills. In Virginia’s concentrated cluster, eight years of monitoring around backup diesel generators shows doubled or tripled pollution levels in some measures. Hakeem reminded the room why that matters in plain terms: those emissions are linked to strokes, asthma, reduced fertility and other serious conditions.
Then there’s rezoning and property value. Once a site is rezoned industrial, nearby property markets change fast — buyers disappear, long-held family farms lose practical value, and eminent domain pressures follow. TD described a local reservoir fight where 20–60,000 acres are slated for water storage while a larger 200,000-acre footprint becomes effectively unsellable, trapping families and shrinking options.
A political argument you’ll hear often — “we must build to beat China” — gets pushed back in the episode. Hakeem and TD call it a diversionary narrative, noting that China’s footprint of data centers is much smaller and that the U.S. buildout looks more like a speculative bubble inflated by tax breaks and accelerated depreciation.
Where pushback works and what anyone can do
Hakeem’s report and the open-source AI datacenter map are halfway between journalism and civic tool: they list planned sites, estimate power draw and water use, and show how to spot projects early. That’s exactly the leverage communities used in cases like San Marcos, Texas, where a large turnout at zoning hearings forced elected officials to reckon with public opposition.
Concrete actions the episode makes realistic:
Check the AI datacenter map for planned projects in your county and note the proposed power figures.
Monitor county commissioner and city council agendas—data centers show up in rezoning or conditional-use items.
Mobilize neighbors early: San Marcos brought 100 people to a hearing and talked until the council listened.
Opt out where you can: Hakeem recommends private devices and disabling built-in AI features as one individual-level hedge.
TD and Hakeem also highlight legal and fiscal levers: contesting tax incentives, exposing federal bonus depreciation benefits, and using public health data to force stricter review. These aren’t instant wins, but they make projects harder and more costly for builders—and those extra costs change the economics that created the rush.
“If Texans deplete the aquifer, that's gonna impact Louisiana, Oklahoma, states in all directions.” — Hakeem Anwar
That quote nails the scale: aquifers are regional commons, not backyard pools. A mine of detail in Hakeem’s free 250-page report (he emphasizes he wrote it without AI) walks through how data centers pick sites, where to look in local records, and how to calculate projected water use.
Final thought
This isn’t a single-issue protest. The debate about AI data centers ties into energy policy, land rights, public health, and the kind of local democracy that notices a rezoning application before it’s a done deal. If you care about farming, clean wells, or preserving property value, this episode argues one clear point: show up early, get the data, and organize with neighbors.
If you want the numbers and the map Hakeem references, the report and the AI datacenter map are linked in the show notes for anyone who wants to dig in further.