Data Center Secrecy Is a Builder's Problem Too

Data center secrecy just got a famous opponent. Erin Brockovich, the environmental campaigner whose Pacific Gas & Electric fight became a 2000 Julia Roberts film, has launched a public map of US data centers, and the headline finding surprised me. The loudest complaint from communities wasn't water or noise. It was simply not being told.

I read the TechCrunch report and wanted to think through what it means for those of us building software far from Memphis or Silicon Valley. Because the buildings Brockovich is mapping are where our code actually runs.

---

🔍 The complaint isn't water, it's silence

Brockovich's site collected nearly 4,000 submissions in April 2026 alone. She says one word kept showing up in submission after submission: transparency. Not utility bills, not aquifers, not the hum of cooling fans.

The pattern her map documents looks like this:

• Projects announced only after permits are already secured.
• Developers who don't return calls from residents.
• Local officials who signed NDAs before their own neighbors knew a project was even on the table.

Key takeaway: The objection here is to process, not technology. Brockovich is clear that she opposes neither data centers nor AI, only the secrecy around how they get built.

That distinction matters. It's easy to read a story like this as "activist versus AI" and tune out. It isn't. It's about who gets to know what, and when.

---

🌐 Why a Sri Lankan engineer should care

You might think this is a US zoning fight with no bearing on Colombo or Galle. I'd push back on that.

First, the compute we use every day lives in buildings exactly like the ones on that map. When you call an LLM API, fine-tune a model, or host a side project, a physical facility somewhere is drawing power and water to serve your request. The supply chain behind "the cloud" is genuinely opaque, and that opacity is now a documented complaint, not a vibe.

Second, the region around us is courting this exact kind of investment. Data centers are pitched everywhere as clean, quiet job creators. The Brockovich submissions are a useful reminder to ask the boring questions early:

Question to ask	Why it matters
Where does the power come from?	Decides the real carbon and grid load
How much water for cooling?	Competes with local supply in dry seasons
Who saw the plan first?	Tests whether the process was open
What was promised to the public?	Holds developers to their word later

If a project can't answer those without an NDA, that's the story.

---

⚡ The hidden footprint of your own AI apps

Here's the part I find useful as a builder. You can't audit xAI's Memphis turbines, but you can measure the footprint of what you ship. Most of us have no idea how much compute our AI features actually burn, which is its own small version of the transparency gap.

A few habits that have helped me:

1. Estimate before you scale. Know roughly what a feature costs in tokens or image generations before you wire it to a million users.
2. Cache aggressively. A repeated prompt that hits a cache draws no new compute. This is the cheapest green decision you'll make.
3. Right-size the model. A smaller model that passes your evals beats a frontier model you reach for out of habit.

This is partly why I built free AI cost and usage tools on this site, including a token counter and an image-generation cost calculator. They won't fix a power grid, but they make your own consumption legible instead of invisible. Transparency starts at home.

Bottom line: You can't demand openness from a hyperscaler while having no clue what your own app consumes. Measure first.

---

💡 Transparency is a default you can choose

The thing I keep coming back to is that secrecy is a choice in Brockovich's findings. Permits could be public early. Calls could be returned. NDAs are signed by people, not forced by physics.

We make the same choice at our own scale. As a small-team builder, you decide whether your privacy policy is honest, whether your changelog is real, whether you tell users what data leaves their browser. None of that is regulated into existence. It's a habit.

For tools on this site, the default is that file conversions and calculations run client-side in your browser wherever possible, so nothing is silently shipped to a server. That's the same principle Brockovich is asking of data center developers, just pointed at our own products: tell people what's actually happening.

---

What this means for you

The Brockovich data center map is a US story on its face, but the lesson travels. Infrastructure built in secret breeds distrust, and AI's physical footprint is exactly the kind of infrastructure that's easy to build quietly.

If you're shipping software from Sri Lanka:

• Treat "the cloud" as a place, not a metaphor. It has power bills and water lines and neighbors.
• Ask the boring questions early when local data center investment shows up, because they get harder to ask after the permits are signed.
• Make your own stack legible. Measure your AI spend, cache what you can, and tell users what your app does with their data.

You don't need to be Erin Brockovich to practice transparency. You just have to stop treating it as optional.