Two protests. Same week. Same grievance — almost.

In Visakhapatnam, Andhra Pradesh, activists held banners reading "We cannot drink DATA" outside Google's $15 billion data center site, worried about a facility drawing millions of gallons of water daily from a city already 70 million liters short of its daily need. Across the Pacific, activists in 42 American states held 142 separate protests that same month — and in many cases, their governments agreed with them.

Here is where the stories diverge completely.

In Visakhapatnam, India's state government listened, issued clarifications, and kept building. In Franklin Township, Indianapolis, Google withdrew its $1 billion data center proposal minutes before a city-county council vote that would have rejected it. Same company. Same infrastructure. Opposite outcomes.

🇺🇸 UNITED STATES

The Country That Says No

530+

counties/localities with bans or restrictions

1,800

projects halted in Texas alone

75%

of U.S. voters oppose nearby data centers

🇮🇳 INDIA

The Country That Says Build

$20–25B

projected data center investment by 2030

4×

capacity growth in six years (to 1,263 MW)

188,000

jobs from Google's Visakhapatnam project

🇺🇸  UNITED STATES

The Country That Says No

530+

counties with data center bans or restrictions

75%

of U.S. voters oppose nearby data centers

12%

of U.S. electricity consumed by data centers by 2028

WHY AMERICA TURNED AGAINST DATA CENTERS

American opposition to data centers is one of the stranger political phenomena of the decade. In a country divided on almost everything, the backlash has achieved rare bipartisan consensus — rural conservatives in Texas and urban progressives in New York have arrived at the same position: not in our backyard.

Water: A large facility consumes 2–4 million gallons per day, competing with agriculture and already-stressed municipal systems.   Power: Data centers will consume 12% of total U.S. electricity by 2028, up from 4.4% in 2023 — directly linked to rising utility bills.   Jobs: A multi-billion dollar project might employ just 50–200 people permanently.   Noise: Industrial cooling systems run 24/7 at decibels that make nearby residential land unliveable.

UNITED STATES — STATE-BY-STATE BACKLASH

STATE / REGION

ACTION TAKEN

SCALE

Texas

Governor's moratorium; ERCOT grid reliability test failures

~1,800 halted

New York

State law blocking projects exceeding 50 megawatts

12+ blocked

Pennsylvania

Exec. order: permits + electricity cost-sharing required

Statewide

Maine

6-month moratorium; extended to 2 years by legislature

Statewide

Franklin Township, IN

Google withdrew $1B project before council vote

$1B cancelled

Nationwide

County-level bans or construction restrictions

530+ localities

Even President Trump publicly called the rejections "a mistake." It made no difference. Local democracy, it turns out, is impervious to federal pressure when residents believe an industry is taking something from them without giving enough back.

"These facilities promise thousands of jobs and deliver hundreds — while taking our water, overloading our grid, and running noise around the clock."

— Franklin Township council member, Indiana

KEY CONCERN

Water consumption, grid overload, noise, and minimal permanent employment

POLITICAL SPLIT

Bipartisan opposition — conservatives (grid/costs) + progressives (environment)

FEDERAL STANCE

Federal government supports buildout; local councils hold override power

TRAJECTORY

Accelerating resistance — 530+ localities in 2025, expected to exceed 700 by end of 2026

CXO RISK

Pipeline capacity delays; 50%+ of planned U.S. AI data center capacity affected

🇮🇳  INDIA

The Country That Says Build

$15B

Google's Visakhapatnam data center commitment

4×

capacity growth in 6 years — to 1,263 MW

188,000

jobs projected from Andhra Pradesh project

INDIA'S CALCULATION: DEVELOPMENT FIRST

India is running the opposite equation — and doing so with full awareness of the trade-offs. The country's data center capacity has grown fourfold in six years, reaching 1,263 megawatts by early 2025. The government's target is 4,500+ megawatts by 2030, requiring $20–25 billion in new investment.

Google's $15 billion project in Andhra Pradesh — built in partnership with Adani Group — is the centrepiece of that ambition. The Visakhapatnam campus alone is designed to create up to 188,000 jobs across direct employment, construction, supply chain, and downstream economic activity. For a city that has been trying to diversify beyond its naval and port economy for two decades, the number is transformational.

The protests in Visakhapatnam are real. Former bureaucrat E.A.S. Sarma has publicly questioned how Andhra Pradesh sustains a 1-gigawatt facility requiring 2.7 to 4.1 million gallons of water daily in a water-stressed region. The Kambalakonda Wildlife Sanctuary — home to leopards and pangolins — sits 860 metres from the construction site. India's National Green Tribunal is hearing challenges.

The difference is not that India has no opposition. The difference is that India's government has decided the development argument outweighs the opposition argument — and has the institutional authority to act on that decision without being vetoed at county level.

INDIA — DATA CENTER INVESTMENT & CAPACITY

CITY / REGION

CAPACITY / SHARE

KEY PLAYERS

Mumbai

41% national share

$4.2B committed

Chennai

23% national share

$3.1B committed

Delhi NCR

14% national share

$2.8B committed

Visakhapatnam

Greenfield — 1 GW target

Google + Adani $15B

Pune / Hyderabad

Growing fast

Microsoft, AWS expanding

"Development should not be at the cost of livelihood."

— Green Visakha activist, at protests against Google's Andhra Pradesh data center

GOVERNMENT STANCE

State governments competing to attract hyperscaler investment; regulatory floor protects projects from local veto

KEY CONCERN

Water consumption (Visakhapatnam short 70M liters/day), wildlife proximity, agricultural displacement

LEGAL CHALLENGE

National Green Tribunal and local courts hearing petitions — proceedings active as of August 2026

TRAJECTORY

Accelerating investment — on track for $20–25B by 2030; opposition maturing but not yet veto-capable

CXO OPPORTUNITY

Latency now viable for most enterprise AI workloads; colocation pricing 30–40% below equivalent U.S. capacity

TWO DEMOCRACIES, TWO ECONOMIC MOMENTS

Here is what I keep coming back to from three conversations with founders this quarter who are making real infrastructure allocation decisions.

The United States built its infrastructure decades ago. Communities blocking data centers have existing employment bases, functioning utilities, and property values they don't want disrupted. They are protecting what they have. 200 permanent jobs is not compelling relative to the water consumption, grid risk, and 24-hour noise they would absorb.

India is in a different moment. State governments are competing with each other for hyperscaler investment. Andhra Pradesh's ability to attract Google and retain the project through legal challenges is a competitive signal to every other hyperscaler deciding where to place their next gigawatt. The 188,000 jobs projection — even if optimistic — represents a transformational number for Visakhapatnam.

THE GOVERNANCE GAP

In the U.S., a single county council can kill a billion-dollar project. In India, state government support creates a regulatory floor that local opposition cannot easily breach. Neither model is simply right — but for infrastructure decisions measured in decades, executives need to know which produces predictable outcomes.

WHAT CXOS MUST DO WITH THIS

The executives I'm watching most carefully are not the hyperscalers — they've placed their bets. They are enterprise CXOs at $500M-plus companies making five-year decisions about where their AI workloads will actually run.

01.  Map your AI infrastructure exposure by jurisdiction — now

If your critical AI workloads run in Northern Virginia, Phoenix, or Dallas, know exactly which county your data center sits in, whether moratorium discussions are active, and what migration options exist. This is a board-level risk disclosure conversation, not an IT one.

02.  Take India's data center build-out seriously as a strategic option

Indian colocation capacity is expanding rapidly, regulatory certainty at state level is real, and latency now works for most enterprise AI applications. For companies with significant India operations or NRI investor bases, this is a natural conversation — for those without, it is worth starting.

03.  Read the U.S. pipeline delay signal carefully

The 530-county figure is alarming in aggregate but most existing capacity is unaffected. The real risk is in the pipeline — planned capacity that may not arrive on schedule. The AI buildout's 50% delay rate is substantially driven by these regulatory friction points.

04.  Build sustainability clauses into any India commitments

India's local opposition is raising exactly the water and energy arguments that reshaped U.S. sentiment. What looks like regulatory certainty in Andhra Pradesh today may look more complex in 2028. Lock in Indian commitments with sustainability clauses — don't assume current openness is permanent.

The geography of AI is being shaped right now by local planning boards in Nebraska, state courts in Andhra Pradesh, grid operators in Texas, and wildlife sanctuary buffers in Visakhapatnam. Executives who understand this — who see infrastructure as a geopolitical and governance question, not just a capex decision — will have more options when the next capacity crunch arrives.

The country that says no is not the enemy of AI progress. The country that says build is not without its tensions. What both are telling us is that the social licence for AI infrastructure is not automatic. In the United States, the industry is losing that argument in 530 counties and counting.