Data centers and pollution: what it means for your SaaS

Every month the invoice for your company's software shows up. You never think about the power plant that keeps it running. Fair enough. But that power plant just made the news. The EPA, the US environmental agency, issued guidance that makes life easier for anyone building power for data centers. Critics say that in practice it's a way around part of the pollution rules. It sounds like an American environmental debate. It isn't. If your business depends on SaaS, and almost every business does today, this reaches your bottom line.

I'll explain it without the legal jargon and show you what I'd do in your place.

What the EPA changed, in plain English

Data centers use a lot of energy. And AI made that consumption explode. According to the International Energy Agency, data centers used about 415 TWh in 2024. That number is projected to double by 2030.

The power grid can't keep up. So tech companies started making their own power: gas turbines and generators set up right next to the building. But each of those turbines needs an environmental permit. And permits take time.

The new EPA guidance shortens that path. It gives more flexibility so permits move faster and construction can start before everything is approved. The government calls it unlocking investment. Critics call it a side door around air quality rules.

I don't need to decide who's right to do the math that matters to you. Cheaper, faster power today, under a rule that could change tomorrow.

The cloud has a smokestack

We say "cloud" as if it were something light, floating around. But the cloud has an address, a power bill and, more and more, a smokestack.

When you use a CRM, an ERP, an AI chatbot or an email tool, you're renting a piece of a data center. The price you pay bundles three things: servers, people and power. With AI, power stopped being a detail. It became a big slice of the cost.

Every "smart" feature in your software has a hidden power bill inside it.

That's why this news matters. It changes the cost of the raw material behind everything that runs on the internet.

How data center power affects your SaaS pricing

Here's my strong opinion: a loosened rule isn't a gift. It's a loan. Someone pays later.

Think about three scenarios.

Short term: prices hold or even drop. With power easier to get, the big cloud providers get some breathing room. That keeps service prices steady for a while. For you, nothing changes on the invoice.

Medium term: the rule changes again. US environmental policy swings back and forth with every election. If the next administration tightens the rules, turbines permitted in a rush may need upgrades, filters or a shutdown. That cost doesn't disappear. It rolls down the chain until it reaches the "Pro" plan you pay for.

Long term: the bill shows up somewhere else. Lawsuits, neighbors complaining, retroactive fines. None of that appears on the vendor's website. All of it becomes operating cost.

Something concrete: I've seen a small company get a 30% price hike on an AI tool from one year to the next. The reason given was "infrastructure costs". There's no way to prove it was power. But that's exactly the kind of increase that gets more likely when the base cost is unstable.

"But I'm not in the US, this doesn't affect me"

It does, in two ways.

First, a lot of the SaaS you use runs on servers in the United States. Even when the company is local, it's often hosted on AWS, Google Cloud or Azure in some US region. Their cost becomes your cost. And it comes converted into dollars.

Second, your big customers are paying attention. Mid-size and large companies already send vendor questionnaires asking about emissions, privacy and where data lives. If you sell to multinationals, this is coming. And "I don't know where my software runs" is a bad answer to give.

I'm not saying you need a carbon report tomorrow. I'm saying that knowing where your infrastructure comes from is no longer just an IT topic.

What to do in practice, without becoming an energy expert

You don't control the EPA. But you control how much you depend on each vendor and how much you waste. That's the good side of the story: using less energy almost always means spending less money.

A simple checklist:

• List the SaaS you pay for. Spreadsheet, name, monthly cost, what it's for. Most companies find a couple of forgotten tools at this step.
• Flag the ones that use heavy AI. They're the most exposed to price hikes blamed on infrastructure costs.
• Ask where your data lives. One email to support does it. A serious vendor replies within a day.
• Check whether there's an alternative. Not to switch now. Just so you're not held hostage if the price goes up.

And then there's the part I like most: automation done right.

Efficient AI uses less power and less money

A lot of companies use AI like someone turning on every light in the house to find their keys. It works, but it's expensive.

An example I see all the time: customer service automation. The company plugs in the most powerful AI model on the market to read every incoming message. Including "ok", "thanks" and "👍". Each of those messages costs processing, power and money.

You can do it differently:

1. A simple filter sets aside the messages that don't need a reply. That's a rule, not AI.
2. A small, cheap model sorts the rest: question, complaint, request.
3. Only the tricky cases go to the big model.

In projects like this, it's common to cut more than half of the calls to the expensive model. The reply to the customer stays the same. The invoice doesn't. And as a bonus, you use less power from that gas-turbine data center. It won't save the planet on its own. But at least you won't be the one paying the neighbor's power bill.

Other savings that work:

• Cache repeated answers. If ten customers ask the same question, you don't need to generate the answer ten times.
• Process in batches. A report that can run overnight doesn't need to run right away.
• Cut what nobody uses. That automatic meeting summary nobody reads? Turn it off.

What this news teaches anyone running a business

The lesson isn't about environmentalism. It's about dependency.

Your business runs on top of a stack of things you don't see: software, servers, power, regulation. When a layer at the bottom changes, the effect slowly climbs up to your spreadsheet. The EPA decision is just a reminder that this stack exists and that it moves.

Companies that know their own stack react fast. The ones that don't find out from the invoice.

If you want to look at your company's tools and automations and find where you're paying more than you need to, I can help with that. I work with design, automation and AI for businesses, always starting with the simplest thing that works. Check out my work and let's talk.

The cloud that runs your software has a smokestack. And its bill will end up in your cash flow.

The EPA made it easier to permit gas turbines for data centers in the US. Faster power today, under a rule that could flip tomorrow.

A loosened rule isn't a gift. It's a loan. When the policy changes, the cost rolls down the chain to the "Pro" plan you pay for.

And if you're outside the US, this still hits you. A lot of the SaaS you use runs on American servers and bills you in dollars.

Here's what you control: list your tools, know where your data lives, and stop using the most expensive AI model to read "ok" and "thanks".

Companies that know their own infrastructure react fast. The ones that don't find out from the invoice.

I wrote the full article with a practical checklist. If you want to see where your company is paying more than it needs to, reach out.

#SaaS #ArtificialIntelligence #Automation #CostManagement #Technology