The Cloud Has an Address

Data centers, communities, and the questions worth asking — a pre-workshop handout for adult educators.

Prepared for the UFT-CWE Chapter · A companion to the workshop series

1 · Why this belongs in adult education

Adult learners use AI every day — for job applications, translation, health information — but most have never seen where that work happens. A data center is basically a warehouse full of computers running around the clock. It needs a great deal of electricity and water, and it gets sited somewhere: in a county, near a substation, upstream or downstream of somebody.

That connects to three themes adult educators already teach:

• Digital literacy: AI isn't "in the cloud." It's in a building, in a county, on a grid.
• Civics: who gets to say yes or no to that building, and who pays the utility bill.
• Work: what jobs actually show up, and for how long.

And it opens onto a fourth question this handout will return to at the end: not just who decides, but who owns.

2 · Three facts to get right — and how to say them simply

Fact 1 — The jobs are real, but building and running are different things.

A typical 100-megawatt hyperscale campus employs roughly 850 construction workers for about 18 months to two years, then settles to about 100 to 200 permanent operations staff. The most automated large campuses can run on far fewer — industry analyses put some at 20 to 30 permanent staff per 100 megawatts, and reporting has found major sites operating long-term with crews well under 150, in some cases just a few dozen.

The fuller picture depends on what kind of facility and how many. A May 2026 Brookings Institution study of roughly 770 U.S. data centers found that counties receiving their first large data center do see total private employment rise about 4–5% over five to six years, with wages up 3–4% — but the durable gains concentrate in hyperscale clusters. Single facilities produce modest effects, and colocation centers — landlord operations renting space and power to tenants — generate construction activity but little lasting local upside.

Teacher note: learners often hear "a thousand jobs are coming." Help them ask the follow-ups: for how many months? What skills? And what kind of facility — because the answer changes everything.

Fact 2 — The local impacts are water, air, and noise. Not abstractions.

A May 2026 study from Santa Clara University researchers (published by the think tank Next 10) mapped every known operating and planned data center in California and found that larger proposals are increasingly landing in places with higher social vulnerability and less reliable water — with Imperial County and Sacramento showing the starkest overlap of the two.

Diesel backup generators matter for neighbors: they release sulfur compounds, hydrocarbons, and other pollutants that degrade air quality at the neighborhood scale, and when many run at once they can spike nitrogen oxides and particulate matter. Noise from cooling systems and generators is one of the most frequent complaints in local hearings.

How to explain environmental justice: "It's about who carries the costs. If a town already has a wastewater plant and a freeway, adding a water-hungry data center means the same families absorb more strain."

Fact 3 — The policy deal is changing: from tax breaks to pay-your-own-way.

For a decade, states competed to attract data centers with incentives. In 2026, two very different states began moving the other direction at the same time — which is the clearest sign the ground is shifting.

3 · What's emerging right now (2026–2027)

New York

• On June 4, 2026, the legislature passed the Responsible Data Center Development Act. It would put a one-year pause on new permits for data centers with peak demand of 20 megawatts or more.
• It is not law yet — it awaits Governor Kathy Hochul's signature, and she has not announced a position.
• If signed, large data centers would have to fund benefits for their host communities and would be placed in a separate electric and water rate class, so the cost of new power lines and pipes can't be quietly passed to residential ratepayers. Facilities of 5 megawatts and up would face renewable-energy and labor standards, including prevailing wages and US-made iron and steel.

Texas

• In June 2026, Governor Greg Abbott directed state regulators to require data centers to "fully fund the costs of electric infrastructure needed to serve their operations, preventing those costs from being passed on to residential ratepayers."
• He also pledged to work with the legislature next session to make data centers add to grid capacity rather than only draw on it, and to require water-efficient cooling, annual reporting of electricity and water use, repeal of some tax exemptions, plus setbacks and noise controls.

Teacher note: a progressive legislature and a conservative governor are saying the same thing in different accents — "if you build here, you pay your own way, and you show us your water and power numbers." When both ends of the spectrum move at once, something real is moving underneath.

4 · Classroom-ready ways to teach it

A. Start with something tangible

• Bring a utility bill, or a photo of a local substation. Ask: "Where does the power for a chatbot come from on a hot afternoon?"
• Show a short video of a data center hall, then ask learners to list what it needs: power, water, cooling, land, workers.

B. A local map activity (20 minutes)

1. Pull up your county GIS map or an online map.
2. Have small groups mark: industrial zones, water sources, schools, lower-income neighborhoods.
3. Ask: "If a company proposed a 20-megawatt site here, what three questions would you ask at a town meeting?"

This makes "siting" concrete without any technical jargon.

C. A structured trade-off talk

Give two roles: a workforce board member who wants the construction jobs, and a neighbor worried about well water. Provide the job numbers above and the California water findings. Let learners practice asking for data, not just opinions: "How much water per day? Who pays for the new transformer?"

Then close with the question neither role has asked yet: "Is being paid the same as having a say? What would change if the community held a share of the facility itself?"

D. Connect to skills

• Numeracy: calculate gallons per megawatt from a fact sheet.
• Reading: compare a developer's one-page press release with a town's one-page moratorium notice.
• Writing: draft a three-sentence public comment using "I notice, I wonder, I request."

5 · Language that helps (and what to avoid)

Helpful frames for adult learners:

• "AI needs real-world resources" — rather than "AI is bad for the planet."
• "Who benefits, who pays, who decides — and who owns" as a four-part check.
• "Temporary construction, small permanent crew" — rather than "no jobs."

And model civic accuracy: don't present New York's bill as law, or the Texas rules as already in effect. Say "passed, awaiting signature" and "directed, to be codified." Getting the status right is part of the lesson.

6 · Take-home handout prompts

These fit on a half-sheet:

1. Data centers power the AI tools we use. They need electricity, water, and land — in a real place.
2. In 2026, California research shows new large sites trending toward communities with higher vulnerability and tighter water.
3. Jobs: big build, small long-term crew — roughly 100 to 200 for a large campus, and it depends on the kind of facility.
4. New York and Texas are both moving toward separate utility rates and community payments, not blanket tax breaks.
5. Questions to ask locally: How much water? Who pays for grid upgrades? What noise limits? What local hiring plan?
6. And the question behind the questions: the company pays the town — but is being paid the same as having a say?