Panic over data centers is wildly exaggerated — they use less water than golf courses and less energy than the USA’s fridges

NY Post ^ | June 16, 2026, | Panic over data centers is wildly exaggerated — they use less water than golf courses and less energ

[DeplorablePaul]

Data centers have become the chic new enemy among activists.

Critics claim the centers are using inordinate amounts of electricity and water to power artificial intelligence, inspiring protesters to take to the streets and Democratic lawmakers to head to Albany to stymie their development.

However, some experts say the anti-data center push is more of a moral panic than an empirical one, often based on speculative and sometimes bunk projections.

---

TOPICS: News/Current Events
KEYWORDS: ai; computers; datacenter

[DeplorablePaul]

The most ridiculous example of the data center panic was when AOC held up a vial of dirty water that was supposedly polluted by a data center. Complete bs. Computers have cooled by water since the beginning of the mainframes. Water is pumped in through pipes, carry off heat and is pumped out. If the water is polluted it started out that way.

[Repeal The 17th]

From yesterday:
https://freerepublic.com/focus/f-chat/4384071/posts
Panic over data centers is wildly exaggerated — they use less water than golf courses and less energy than the USA’s fridges (only 4.63 years left)
NY Post ^ | 6/16/26 | Rikki Schlott
Posted on 6/16/2026, 8:20:58 PM by Libloather
38 comments

[tired&retired]

What data will they be storing?

The AI summaries of your emails and phone calls? All your financial transactions, including credit card purchases.

It is already far worse than Snowden warned us about.

Big brother is tightening the noose of control. China is already doing this on their citizens.

[DeplorablePaul]

Missed it. I usually scroll through them all.

[Hot Tabasco]

they use less water than golf courses

The issue isn't the volume of water usage, but rather the fact that they will be using that water and increasing the demand on the source that provides it.

[maddog55]

Quick search shows differently:

Typical hyperscale facility power draw: ~100 MW is common; some campuses/planned projects are on the GW scale.
Global data‑center electricity consumption (mid‑2020s): ~415–460 TWh (≈1.5–2% of global electricity in 2024), with projections to double or reach 650–1,050 TWh by late‑2020s depending on scenario.

U.S. data‑center electricity (recent): ~176 TWh in 2023 (several percent of U.S. electricity), with rapid growth projected. iaeimagazine.org

Direct water use per large data center: up to ~5 million gallons per day (many hundreds of millions to >1 billion gallons per year for the largest sites); city‑scale aggregated estimates for U.S. data centers reach tens to hundreds of billions of gallons annually when including indirect (power‑generation and supply chain) water.

Indirect water footprint (via electricity generation and manufacturing) can exceed on‑site cooling use; estimates show hundreds of billions of gallons annually in the U.S. when combined.

[Doctor Congo]

I’m assuming they mean ALL the fridges in the US draw more energy than a single data center. That’s a ridiculous comparison. It should be compared to the energy needed for a medium sized city.

[Dalberg-Acton]

RIOT has one 4 miles from my home. Several 18 wheeler water trucks run day and night to supplement the water available to them from the water company.

[DeplorablePaul]

I think power and water usage concerns can be offset by making the data centers self sufficient. That said I’m with Musk and his plan to build data centers in space.

[central_va]

Amazing the USA can build hundreds of data centers so quickly . When elites want something they get it.

[Renkluaf]

The anti-DC crowd is doing the PRC’s work for them.

[rodguy911]

Water and electric useage is wildly overstated below some AI answers to some of the questions.

Even if large data center have to get small nuclear powered plants to produce their electric it will wind up being a benefit to the local community, since they may produce more that needed. On the water cooling sooner or later desalinization plants will become realities and will solve much of our fresh water needs. Also im surprised that AI has not come up with cooling methods for computers,that has to be in the works here's a self cooling method being designed that you just told me about:

Those points have merit, but each has limits and trade-offs. Brief responses to each claim:

Small nuclear plants for local power
Benefits: reliable low‑carbon baseload, potential local jobs, possible excess power for community use.
Limits: high capital cost, long permitting and construction timelines, waste handling/security, grid integration and financing challenges, and potential public opposition. Siting many small reactors near population centers raises regulatory and logistical hurdles.
Desalination for cooling water
Benefits: provides drought‑resilient source where seawater is available.
Limits: desalination is energy‑intensive and often carbon‑intensive unless paired with low‑carbon power; it produces brine discharge that harms marine ecosystems; inland data centers far from coasts can't use it without costly pipelines; scaling to meet massive cooling demands would transfer burdens (energy, brine disposal) rather than eliminate them.
New cooling technologies (including “self‑cooling”)
Progress exists: liquid immersion cooling, direct-to-chip liquid cooling, heat-pipe/phase-change systems, and waste-heat capture are being deployed and improved.
Limits: higher upfront cost, maintenance and reliability concerns, retrofit complexity for legacy hardware, supply-chain and standardization gaps, and heat rejection still required—if heat is dumped to air or water, local thermal/power impacts remain. Capturing waste heat for district heating works well in some climates but is location‑dependent.
Net: nuclear, desalination, and advanced cooling can substantially reduce local impacts, but each introduces costs, trade‑offs, and new environmental or logistical challenges. Solving the scale problem will require a combination: more efficient chips and software, higher PUE standards, broader deployment of low‑carbon power, smarter siting (avoiding water‑stressed areas), waste‑heat reuse where viable, and regulatory frameworks to manage externalities.

Yes, self-cooling methods are being developed for computers, particularly for high-performance chips like those used in AI, which generate significant heat. Innovations such as microfluidics allow cooling liquids to flow directly onto the chip, improving heat removal efficiency significantly.

Correct — microfluidic and related direct-liquid cooling methods are actively developed and deployed for high-performance AI chips. Key points:

Microfluidic/direct-to-chip cooling: coolant flows through microchannels or cold plates attached to die or package, giving much higher heat-transfer coefficients than air and lowering junction temperatures. Benefits: higher power density support, smaller thermal gradients, and improved energy efficiency.
Immersion cooling: entire servers or components submerged in dielectric fluids removes heat uniformly and simplifies plumbing; used by hyperscalers and HPC sites.
Two‑phase and heat‑pipe approaches: use phase change (boiling/condensation) at the chip or package level for very effective heat removal with passive or compact active loops.
Waste‑heat reuse & heat pumps: combining direct cooling with heat-exchange and heat-pump systems can raise rejected heat to useful temperatures for district heating or desalination feedwater preheating.
Limits/challenges: reliability and serviceability (liquid compatibility, leak management), retrofit complexity for existing datacenters, coolant cost and handling, standards and supply chain maturity, and the need to reject heat ultimately (so local thermal/environmental impacts remain).
Net effect: these techs can greatly reduce water and air-cooling needs and improve PUE, enabling denser AI racks and lowering energy per compute, but they don't eliminate the need for power or the ultimate disposal/reuse of rejected heat.
If you want, I can summarize specific microfluidic/immersion vendors, performance numbers, or a short comparison of cooling methods.

[ClearCase_guy]

When a data center uses water as part of their cooling system, where does the water go? Of course there would be some incidental loss, and perhaps some evaporation, but I would expect something of a closed system. Water goes here and collects heat, then water goes there and loses heat, then the water goes back and collects more heat before it loses the heat yet again. It doesn’t seem like this would be an endless drain on the region’s water supply.

[rodguy911]

Ai is invasive as hell but that will come to a stop real quick. They will still do it we just won’t see it printed in our emails.
We are entering a brave new world lots of challenges ahead. Hard to say how it all winds up.