The Machine Underneath: AI, Data Centers and Trust

AI runs on buildings most people never see. Nabeel Mahmood on why data centers lost public trust, the nutrition label fix, and why AI needs plumbers.

By Mykel Salomon Β·

Every prompt runs on a building. Who gets a say in it, and who is going to build it?


πŸ”“ Opening Reflection

Every time you ask a chatbot a question, somewhere a building the size of several football fields draws power and water to answer you.

We talk endlessly about the model. Which one is smarter, which one writes better, which one won the benchmark this week. We almost never talk about the machine underneath it.

I spent more than a decade in IT operations before I moved into global support. Change advisory boards. Incident command. The calls at two in the morning when something critical goes down and an entire business is waiting on a handful of people to bring it back. That world taught me something most people never have reason to learn: the cloud is not a cloud. It is concrete, copper, cooling, power lines, water pipes, and people. Always people. And when it works, nobody knows they exist.

My guest this week has spent more than twenty years building that invisible layer. Nabeel Mahmood is a technologist, futurist, and co-founder of Nomad Futurist, a nonprofit working to demystify technology and bring new people into the industry. He calls this era the fifth compute: after fire, steam, electricity, and the digital age, an epoch driven by computation at a scale and speed we have never seen.

He also carries this conversation in his body. Nabeel has survived three heart attacks and has been, in his words, legally dead a couple of times. The ring and watch he wears, measuring data around the clock, are part of why he is still here. When he says technology extends and improves life, he is not speaking theoretically.

So I came into this conversation expecting to talk about infrastructure. I left it thinking about trust.

Here is where I land.

AI's two biggest bottlenecks are not chips or power. They are human: trust and talent. The industry is losing communities because it never explained itself, and it cannot fill the jobs it needs without the very people it has failed to convince. Both problems close the same way. Tell people the truth in plain language, and invite them in.


πŸ”¦ Signal of the Week

The machine is getting much bigger, fast.

The International Energy Agency projects that global data center electricity use will more than double, from about 415 terawatt-hours in 2024 to roughly 945 by 2030, with AI as the main driver. The United States accounts for the largest increase, up to 240 TWh of growth, a jump of about 130%.

And communities are pushing back harder than ever.

Data Center Watch, which tracks local opposition, counted at least 75 US data center projects worth about $130 billion blocked or delayed in the first three months of 2026 alone. That one quarter roughly matched all of 2025.

Organized opposition groups grew from 396 at the end of 2025 to 833 by March 2026, across 49 states. More than 300 state-level data center bills were filed in the first six weeks of the year, and statewide moratorium proposals were introduced in 14 states. The concerns are concrete: electricity demand and power bills, water use, noise, and whether these facilities give enough back to the places that host them.

Meanwhile, the people to build and run all of it are running short.

The United States needs more than 300,000 new electricians over the next decade, with around 20,000 retiring every year, and nearly 30% of union electricians are already between 50 and 70. Electricians top the list of hardest-to-fill trades in data center construction.

Zoom out further and Korn Ferry projected a global shortage of 85.2 million skilled workers by 2030, putting $8.5 trillion in annual revenue at risk. That projection dates from 2018 and the shape of the gap has shifted since, but the direction has not.

Put those three signals side by side and you see one story. The builders need communities to say yes. Communities do not trust the builders. And the people who would actually do the work are often living in the very communities saying no.


🌍 In the Wild

1. The industry built prisons, then wondered why nobody wanted to live next door

What surprised me most was how quickly Nabeel took responsibility on behalf of his own industry.

"We have not done a good job. We've actually built ugly buildings, buildings that look like prisons, buildings that are not welcoming." Nabeel Mahmood

There was a reason for it. For decades, secrecy was security. If nobody knew where a data center was or what it held, it was harder to attack. So the industry built blank, windowless boxes and stayed quiet.

Nabeel used a scene from Office Space to describe what happened next. Technology people used to be like the guy pushed down to the basement, the best-kept secret in the building. Then the pandemic hit, the whole world ran on data overnight, and suddenly that same person was in the corner office, at the board table, invited to every meeting.

But he never learned how to explain himself.

I recognize this completely. In IT operations, invisibility is the goal. If you are doing your job well, nobody notices you. The only time anyone hears from you is when something breaks. We trained a whole profession to be silent and competent, and it worked, right up until we needed the public's permission.

Here is my take. Secrecy made sense when data centers were background infrastructure. It cannot survive the scale we are talking about now. When a single AI venture announces commitments of hundreds of billions of dollars, people do not hear "infrastructure." They hear a world being built by billionaires, for billionaires, somewhere they will never be asked about. I said that on the show, and I believe it is how a lot of people feel.

You cannot ask a community to host the physical heart of the AI economy while treating it like a classified facility. Silence used to protect these buildings. Now it is what puts them at risk.

2. A nutrition label for compute

Nabeel's solution is simple enough that I am surprised it is not standard practice already. He calls it nutritional labeling for compute.

Before a data center goes into a community, publish its footprint in plain language, the way a food label or a car's miles-per-gallon sticker does. How many jobs it will create. How much tax revenue it will bring. How much energy it will use, and where that energy comes from. How efficiently it will run. Something an ordinary person can read in two minutes and understand.

He was even blunter than that.

"And if it is bad, they need to stop us. They need to question us." Nabeel Mahmood

I have heard a lot of industry leaders talk about transparency. I have rarely heard one invite the community to say no.

And it is already happening. In the European Union, the Energy Efficiency Directive already requires data centers above 500 kilowatts of IT power to report their energy performance into a European database, and the EU is building a common rating scheme on top of it. Measuring and disclosing this is clearly possible, which makes it a choice.

So I love the idea. But I want to push it further in two ways.

First, the label has to show the costs, not just the benefits. A nutrition label does not only list the protein. It lists the sugar and the sodium too. A compute label that lists jobs and tax revenue but not peak power draw, water consumed, noise levels, and the tax incentives the community gave away to get the project is a brochure with a label's font. People can tell the difference, and the moment they suspect a brochure, trust is gone.

Second, timing matters as much as content. If the label shows up after the protests start, it reads like damage control. Publish it before the first permit hearing, before the land deal is final, while the community still feels it has a real say. Transparency offered early is respect. Transparency offered late is spin.

3. Don't take the emotion out. It is data.

This is where Nabeel and I see things differently, and I want to be honest about it.

Nabeel loves machines partly because they are logical. Ones and zeros. Something works or it does not. He believes conversations go better when we put emotion aside.

"You cannot debate logic with emotion." Nabeel Mahmood

I understand where he is coming from. I work with developers every day, and there is something clean about code. It runs or it fails. No feelings involved.

But I have spent my career in customer support, and that taught me the opposite lesson. Emotion is often the most valuable data you will ever get. A frustrated customer is telling you exactly where your product hurts. If you dismiss the frustration because it is not "logical," you miss the signal and keep shipping the problem.

The people in those 833 opposition groups are worried about their power bills, their water, the noise outside their windows, and whether anyone will ever ask what they think. That fear is the community's error log. Engineers would never ignore an error log. They read it, trace it, and fix the root cause.

So I would reframe Nabeel's point rather than reject it. He is right that fear without information goes nowhere. That is why he keeps telling people to get educated, ask hard questions, and find people who can explain the technology. But asking people to leave their emotions at the door gets the industry nowhere. The job is to treat those emotions as information and answer them.

If you want communities to be logical about data centers, start by being honest about what they are afraid of.

4. The most important person in the AI economy might be a plumber

This was my favorite part of the conversation, because it flips the usual story about AI and jobs.

When people imagine the AI workforce, Nabeel said, they picture people in glasses staring at screens all day. Some of those jobs exist. But someone has to build the machine. That means architects, electrical, mechanical, and civil engineers, builders and tradespeople, metal fabricators, cybersecurity specialists, finance and marketing teams. And plumbers. Liquid cooling for high-density AI hardware runs on pipes, pumps, and people who know how to keep them from failing.

"That plumber is the most important asset in compute all of a sudden." Nabeel Mahmood

There is something here I think almost everyone misses. We spend so much energy asking which jobs AI will replace. The answer to which jobs are hardest to replace might be hiding in plain sight: the physical jobs that AI itself depends on. No model can run without someone keeping its building powered and cooled. This is the same pattern we have seen across the last several editions. The jobs nobody counts, AI evaluators, governance professionals, and now the trades, are the ones the whole system needs most.

Nabeel told me about people he has hired from outside tech entirely. A freight hauler who had spent years driving eighteen-wheelers across the continent. What did he know? Logistics. Criticality. Uptime. Getting valuable equipment to the right place, on time, undamaged. Instead of watermelons, he now moves servers, racks, and generators between sites. He has also hired military veterans who had never thought of themselves as technologists and who now write the sequences of operations that keep data centers running.

My own career says the same thing. The habits I built in IT operations, change control, incident response, staying calm when something critical breaks, turned out to transfer directly into leading global support. The title changed. The muscle did not.

Nabeel also told a story about a job that disappeared: the town crier, the person who walked through a neighborhood calling out the news. That job vanished a long time ago. Today, he points out, everyone is a town crier. We all share news and opinions on social platforms, and an entire industry of infrastructure and people exists to carry those posts. One job eliminated, thousands created.

That is true across a century. I said on the show, and I still believe, that it can be brutal across a single career. You cannot flip a switch and turn a forty-year-old freight hauler into a data center logistics lead overnight. There is a gap between losing the old job and being ready for the new one, and that gap is where people get hurt.

Nabeel described learning through change as building a parachute on the way down, one that has a few holes in it. I think that is honest, and I think we owe people better parachutes. Free training like the Nomad Futurist Academy, skill mapping that shows people what they already know, and mentorship that pairs them with someone who has made the jump. That is how the transition happens without leaving people behind.

5. Fear people, not machines

Toward the end, I asked Nabeel whether he really believes AI will never be smarter than humans. His answer was more interesting than a yes or no.

He does not even like the term artificial intelligence. Intelligence, he argues, cannot be artificial. He prefers actionable intelligence. Machines are getting extraordinarily powerful at processing information, but the human is still in control. And that is exactly what concerns him.

"So I would not necessarily be afraid of technology. What I'm afraid of is people." Nabeel Mahmood

His point is that AI only becomes "smarter than us" in a practical sense if we stop educating ourselves and leave the decisions to a small group of people who did not stop. Someone is always controlling the system. The question is who.

He also had the most memorable advice of the hour about how to use it. Treat AI like a meal, not a buffet. Take what you need, in moderation. And do not ask it for relationship advice, he joked, because the only relationship it has ever had is with whoever charges it.

I am less certain than Nabeel about how far machine intelligence will go. But I think his deeper point matters more than the prediction: what we should fear is concentration of control, not computation itself.

And this is where the whole conversation comes full circle for me. Most of the decisions about AI feel like they happen far away, in boardrooms and labs and capitals. But data centers have addresses. They need land, water, power, permits, and zoning approval. That means one of the few places where ordinary people still have a direct, formal say over the future of AI is a local planning meeting.

We talk a lot about AI governance at the global level. Some of the most important AI governance in the country is happening right now, in county meeting rooms, between residents and builders who have never been taught to understand each other.

That is why the nutrition label matters. And why the plumber matters. And why showing up matters.


πŸ’¬ The Big Question

Nabeel asked himself a question that I think every one of us should answer.

Would you live next to a data center?

And if your answer is "it depends," the more useful question is:

What would they have to tell you, honestly and in plain language, before you would say yes?


🧠 A Small Exercise

1. Find the machine near you.

Search for data centers operating or proposed in your county or state. Most people have no idea there is one twenty minutes from their home. Knowing where it is changes how you think about every prompt you send.

2. Ask for the label.

If a project is proposed near you, ask your local officials for the plain-language version: power, water, noise, jobs, tax revenue, and the incentives offered. If nobody can answer, that is information too.

3. Map your transferable skills.

Write down three things you are already good at that do not depend on your current title. Logistics, crisis response, coordinating people, explaining complex things simply. Then ask where those skills are needed in the AI economy. You may be closer than you think.

4. Take one free module.

Nabeel's Nomad Futurist Academy is free and open globally, with around 25 modules covering the core of how compute infrastructure works. Start with one. Understanding the machine is the first step to having a say in it.

5. If you lead a team, write your own nutrition label.

Pick one AI initiative inside your organization and describe it in plain language for the people it affects: what it uses, what it costs, what it changes, who it helps, and what risks you are watching. If you would be uncomfortable sharing that label, that is worth noticing.


πŸŽ™οΈ This Week on The Human Protocol Podcast

The Machine Underneath AI: Data Centers, Trust, and the Jobs Nobody Sees

Guest: Nabeel Mahmood, technologist, futurist, keynote speaker, and co-founder of Nomad Futurist

Nabeel has spent more than twenty years leading global organizations through growth, acquisition, and technology reinvention, bridging IT and business strategy. Through Nomad Futurist, he is working to demystify technology and open the industry to people who never imagined they belonged in it.

In this episode, you'll learn:

🎧 Available now on Spotify, Apple Podcasts, and YouTube.

You can find Nabeel on LinkedIn and through Nomad Futurist.


🧭 Thought to End the Week

Somewhere in the middle of a conversation about megawatts, cooling systems, and zoning fights, Nabeel said something that had nothing to do with infrastructure.

"None of this that we are doing is going to go with us. We're all going to leave this behind." Nabeel Mahmood

He has come close enough to the end to mean it. And he told me what matters to him instead of the size of the technology stack he builds or the famous company names he could put on his gravestone. Three questions. Did I contribute? Did I make a difference? Was I helpful?

I keep thinking about how differently the AI buildout would look if more of the people making these decisions asked themselves those questions.

The cloud has an address. It has a power meter, a water line, and a plumber on call. It sits on land that belonged to a community before it belonged to a company. And every decision about how it gets built is, in the end, a decision made by people, about people.

We can build the machine underneath AI in secret, and keep fighting the communities that have to live beside it. Or we can build it in the open, label it honestly, train the people who will run it, and give the neighbors a real voice in what comes next.

The first path might be faster for a while. Only the second one lasts.

The code may change. And it will.

But our humanity is the constant.

β€” Mykel


πŸ“š Sources

  1. Data Center Dynamics, "IEA: Data center energy consumption set to double by 2030 to 945TWh" (415 TWh in 2024; US growth up to 240 TWh, about 130%) https://www.datacenterdynamics.com/en/news/iea-data-center-energy-consumption-set-to-double-by-2030-to-945twh/

  2. International Energy Agency, Energy and AI https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai

  3. Newsweek, "Data Center Projects Face 'Unprecedented Surge' in Blocks and Delays: Study," July 2026, reporting Data Center Watch's Q1 2026 findings (at least 75 projects worth about $130 billion; opposition groups from 396 to 833; 300+ state bills; moratorium proposals in 14 states) https://www.newsweek.com/data-center-projects-face-unprecedented-surge-in-blocks-and-delays-study-12240534

  4. Data Center Watch, Q1 2026 report https://www.datacenterwatch.org/q1-2026

  5. Classet, "Data Center Labor Shortage 2026: The Numbers by Trade," summarizing Fortune (March 2026) on the need for 300,000+ new electricians over the next decade and the age profile of union electricians https://www.classet.ai/blog/data-center-labor-shortage-2026

  6. Korn Ferry, "Study Reveals Global Talent Shortage Could Threaten Business Growth Around the World," May 2018 (85.2 million workers; $8.5 trillion in unrealized annual revenue by 2030) https://ir.kornferry.com/news-events/press-releases/detail/214/korn-ferry-study-reveals-global-talent-shortagecould-threaten-business-growth-around-the-world

  7. White & Case, "Data centres and energy consumption: evolving EU regulatory landscape and outlook for 2026" (Energy Efficiency Directive reporting for data centres of 500 kW+, European database, rating scheme) https://www.whitecase.com/insight-alert/data-centres-and-energy-consumption-evolving-eu-regulatory-landscape-and-outlook-2026

  8. Nomad Futurist Foundation, launch of the free Nomad Futurist Academy https://media.nomadfuturist.org/nomad-futurist-foundation-partners-with-chekhub-to-launch-the-nomad-futurist-academy/

A note on voices: This newsletter reflects my own perspective. Quotes and ideas from podcast guests are shared with attribution and remain the views and intellectual property of those guests. Featuring a guest's perspective does not mean I endorse every view they hold, and their appearance does not imply they endorse mine.