DATA CENTERS WON’T KILL YOU

The Reality Behind AI Infrastructure, Water Usage & Power Consumption

By Dan Harley | Tech Talk

There’s suddenly a LOT of fear surrounding AI data centers.

If you spend even a few minutes on social media, you’ll quickly see posts claiming that data centers are going to:

• drain California’s water,

• collapse the power grid,

• destroy open land,

• increase wildfire risks,

• and somehow become the next industrial apocalypse.

And honestly?

I understand why people are concerned.

These are large infrastructure projects.

• They use power.
• They generate heat.
• They absolutely deserve intelligent public discussion.

But after digging deeper into the engineering side of modern data centers, I realized something important:

Most people are imagining older systems.

That’s really the key takeaway.

My Background

Now personally, I probably have kind of a unique background to discuss this topic.

I’ve spent decades around:

• networking,

• internet infrastructure,

• servers,

• cybersecurity,

• and technology systems.

But I also spent more than 25 years around the water and wastewater industry working with SCADA systems — the industrial automation systems that help operate pumps, reservoirs, treatment systems, and utility infrastructure.

So I’ve spent a LOT of time around both technology infrastructure and utility infrastructure.

That’s part of why many of the headlines around data centers started raising my eyebrows.

The Reality: Data Centers ARE Coming

Whether people like it or not, AI infrastructure is expanding rapidly.

The question really is not IF anymore.

The question is:

• where,

• how,

• and whether these facilities are designed intelligently.

Countries around the world are aggressively building AI infrastructure right now.

China.
India.
The Middle East.
The United States.

This is becoming part of the next major technological race.

The Water Discussion

Let’s start with the biggest concern people bring up:

Water.

Modern AI systems generate massive amounts of heat.

That heat has to be removed.

Historically, one of the cheapest ways to cool large facilities was evaporative cooling using water.

So yes — some large data centers can consume enormous amounts of water.

But here’s the part many discussions leave out:

The real engineering challenge is not “using up water.”

The real challenge is removing heat.

That distinction matters.

Most People Picture Drinking Water

When people hear “water usage,” they immediately imagine drinking water.

Their sinks.
Their showers.
Their lawns.

But modern infrastructure systems are far more complicated than that.

Wastewater treatment plants already process enormous amounts of reclaimable water every single day.

Much of that water can potentially be reused for industrial cooling.

Not drinking water.
Cooling water.

In regions like Sacramento, reclaimed water infrastructure already exists at large scale.

That dramatically changes the conversation.

Modern Cooling Changed Everything

This is probably the biggest misconception surrounding modern data centers.

Water may be the CHEAPEST cooling method.

But it is absolutely NOT the only cooling method.

Modern facilities increasingly use:

• closed-loop cooling,

• direct-to-chip cooling,

• liquid cooling,

• dry cooling systems,

• immersion cooling,

• and advanced heat exchanger systems.

Your gaming PC at home already uses similar concepts.

It doesn’t have a garden hose connected to it.

Many cooling systems work more like a car radiator:

The system moves heat instead of constantly consuming water.

That’s a huge difference.

And the technology already exists TODAY.

Power Usage Is Real — But So Are Modern Solutions

The second major fear surrounding data centers is power consumption.

And yes — AI data centers absolutely consume massive amounts of electricity.

California already has legitimate grid challenges.

Brownouts.
Transmission bottlenecks.
Infrastructure strain.

But here’s the important part many discussions ignore:

Large-scale data centers increasingly cannot rely entirely on the public grid.

Why?

Because downtime is EXTREMELY expensive.

A major outage for an AI facility could cost enormous amounts of money very quickly.

So what’s happening?

Modern data centers are increasingly becoming partially self-powered infrastructure ecosystems.

That includes:

• natural gas generation,

• battery systems,

• microgrids,

• solar,

• wind,

• and eventually modular nuclear technologies.

In simple terms:

Some data centers are increasingly becoming their own mini power plants.

Natural Gas Reality

This part makes some people uncomfortable, but engineering discussions should stay practical.

Natural gas remains:

• relatively cheap,

• reliable,

• scalable,

• and capable of providing stable baseline power.

For facilities operating 24 hours a day, reliability matters enormously.

That’s simply engineering reality.

The Biggest Issue May Actually Be Regulation

Ironically, the biggest challenge for California data centers may not be water or power.

It may be regulation.

California remains one of the most technologically advanced regions in the world.

But it’s also one of the hardest places to build major infrastructure projects.

Permitting.
Politics.
Environmental review.
Development costs.
Legal uncertainty.

All of these factors push companies to consider other states.

Places like North Dakota or South Dakota offer:

• cheaper land,

• abundant energy,

• easier expansion,

• and simpler development pathways.

And where infrastructure goes… jobs follow.

Why I’m Generally Pro-Data-Center

Now this may sound like I’m pro-data-center.

Well… maybe because I am.

Not because I think every project is perfect.
And not because I think environmental concerns should be ignored.

But because these facilities are going to power technologies that could significantly improve our future.

AI systems may dramatically accelerate:

• healthcare research,

• engineering,

• scientific discovery,

• logistics,

• energy management,

• and countless other industries.

Things that once took years or decades to discover may eventually happen in hours or days.

And honestly, we’re already beginning to see signs of that.

Final Thoughts

No — data centers are not going to magically solve every problem.

And no — communities should not stop asking questions.

Responsible planning matters.

Environmental concerns matter.

Smart infrastructure matters.

But fear and misinformation do not help anyone.

The reality is much more nuanced than the social media panic often suggests.

Modern engineering evolved.

Cooling technology evolved.

Power infrastructure evolved.

And a lot of the public conversation is still imagining older systems.

Data centers aren’t evil.

They’re infrastructure.

And infrastructure built the modern world.

🎥 Watch the full Tech Talk episode here:
https://youtu.be/Elkn0sDPi9s

#DataCenters #AI #Technology #Infrastructure #Engineering #California #ArtificialIntelligence #TechTalk