- Forward Future by Matthew Berman
- Posts
- 👾 The Hidden AI Boom: Data Centers Are Reshaping the Global Economy
👾 The Hidden AI Boom: Data Centers Are Reshaping the Global Economy
From Texas wind farms to Indian tech hubs, a trillion-dollar infrastructure revolution is quietly powering the future of artificial intelligence

Estimated Read Time: 11 minutes
The artificial intelligence revolution isn't just happening in Silicon Valley boardrooms or university labs. It's unfolding in concrete and steel across the globe, from the wheat fields of North Dakota to the industrial zones of central India.
While the world fixates on ChatGPT updates and AI breakthroughs, a quieter but equally transformative story is taking shape: the largest infrastructure buildout since the Interstate Highway System. Companies you've never heard of are erecting "AI factories" that consume more electricity than entire cities. Utilities are scrambling to build nuclear reactors and upgrade power grids to meet unprecedented demand. And communities from rural Texas to urban Spain are betting their economic futures on becoming the next great AI hub.
This is the physical backbone of the AI age—and it's reshaping the global economy in ways most people haven't even begun to grasp.
The New Industrial Revolution Takes Shape
In the small town of Ellendale, North Dakota, something extraordinary is happening. Applied Digital, a company that started in cryptocurrency mining, is building what amounts to a digital factory spanning 400 megawatts of power capacity. That's enough electricity to power roughly 300,000 homes, all dedicated to training AI models and running machine learning algorithms.
But here's the thing—this isn't even their biggest project. The company plans to expand the Ellendale site to 2 gigawatts and is developing three additional campuses across North America, targeting a total of 1.4 gigawatts of capacity. To put that in perspective, a single gigawatt facility would consume as much power as a major metropolitan area.
Applied Digital chose North Dakota for a reason that reveals the new economics of AI: "stranded power." The state produces abundant wind energy that often can't be efficiently transmitted to distant markets. By building directly where the power is generated, these AI facilities can access cheap, renewable electricity while solving a grid efficiency problem.
It's a strategy being replicated worldwide, and it's creating some unlikely industrial powerhouses.
Welcome to the AI Factory Era
Chase Lochmiller has a bold vision for what he's building in Abilene, Texas. As CEO of Crusoe Energy Systems, he's overseeing construction of what he calls an "AI factory"—a 980,000-square-foot data center campus
that will eventually consume 1.2 gigawatts of power and house up to 50,000 high-end AI chips.
"The sheer scale of compute power concentrated here is remarkable, defining an entirely new category for digital infrastructure," Lochmiller explains. At full capacity, the facility will enable "intelligence to be manufactured with unprecedented speed and scale."
The construction site tells its own story about the economic impact. Some 2,000 workers are already on-site daily, with numbers expected to swell to 5,000 as construction peaks. The city of Abilene estimates the initial phase alone will inject $1 billion into the local economy over 20 years.
This isn't just an American phenomenon. In Indore, India, data center firm RackBank is building the country's first purpose-built AI facility—an 80-megawatt campus designed to house 60,000 GPU processors. "With India's massive digital user base and exponential data growth, this facility is uniquely positioned to become a global AI training powerhouse," says CEO Narendra Sen.
Even in Spain, investment giant Blackstone is funding a €7.5 billion campus in Aragón that will supply 300 megawatts of AI computing capacity. The message is clear: the race to build AI infrastructure is global, and the winners will be determined by who can most efficiently convert electricity into computing power.
The Energy Challenge That's Keeping Utilities Up at Night
Here's where the story gets really interesting… and complicated. All this AI infrastructure needs power. Lots of it. And that's creating an energy crisis that's forcing utilities to completely rethink their business models.
In Texas, utility company Oncor has received inquiries to connect 119 gigawatts of new load, almost four times the peak usage of its entire existing system. In the Northeast, PPL Corporation is dealing with over 50 gigawatts of data center requests, including 9 gigawatts in advanced development. That's more than the company's entire current generation capacity of 7.2 gigawatts.
"What we're seeing is this huge proposed influx of these abstract projects that nobody knows anything about," observes Jon Gordon, a director at Advanced Energy United, describing the secretive nature of AI companies' power negotiations.
The scale is staggering. Even mid-sized utilities are seeing proposals that would double their entire systems. Evergy, which serves Kansas and Missouri, has a pipeline of 11 gigawatts in data center demand—roughly equal to its whole system's peak load.
But here's the catch: renewable energy alone can't solve this problem. AI models need to run 24/7, and the wind doesn't always blow and the sun doesn't always shine. The International Energy Agency projects that electricity use by data centers will roughly double by 2030, and warns that baseload power sources will be needed to fill gaps when renewables falter.
In practice, that means natural gas and even coal plants may carry much of the load for new data centers through this decade. It's a reality that's forcing even the most environmentally conscious tech companies to reconsider their energy strategies.
The Nuclear Renaissance Nobody Saw Coming
The solution, increasingly, is nuclear power. In the past year, we've witnessed a remarkable shift in how tech companies think about atomic energy.
Amazon funded four small modular reactors (SMRs) in partnership with Energy Northwest to supply its West Coast data centers, with the first phase expected to generate 320 megawatts. Google signed an agreement with reactor startup Kairos Power to develop advanced reactors by 2030. Microsoft went so far as to sign a deal aimed at restarting the dormant Three Mile Island Unit 1 reactor in Pennsylvania specifically for data center energy.
"There's a strong appetite for nuclear energy to meet the incredible demand for power globally," says a representative from engineering firm Fluor Corporation, which sees SMRs as a key growth area. According to Gordon Dolven, director of data center research at CBRE, these reactors offer "scalable and flexible solutions to support future energy needs."
The appeal is obvious: nuclear provides steady, carbon-free power that can run around the clock. For AI workloads that never sleep, it's becoming the preferred solution.
Creative Solutions in the Energy Transition
Not every answer involves high-tech nuclear reactors. Some of the most innovative approaches are happening right now, using existing renewable infrastructure in creative ways.
In South Texas, Soluna Holdings is building "Project Hedy," a 120-megawatt data center directly co-located with a 200-megawatt wind farm. The facility acts as what CEO John Belizaire calls a "digital battery," capturing curtailed wind energy that would otherwise be wasted and converting it into computing power.
By absorbing excess output when the grid can't handle it, the data center reduces renewable curtailment by nearly 80% at that site. It's a elegant solution that turns an energy challenge into an advantage: the AI workload benefits from cheap, green electricity, and the wind farm gains a steady buyer for its off-peak power.
Meanwhile, global investor KKR has launched a $50 billion partnership with Energy Capital Partners aimed at expanding "data center power and transmission infrastructure" for the AI era. The money is flowing into everything from new substations to transmission lines, all designed to handle the unprecedented power demands of AI computing.
A Truly Global Phenomenon
What's striking about this infrastructure boom is how geographically dispersed it is. This isn't just about traditional tech centers anymore, it's a global transformation reaching places that might surprise you.
The United States remains the epicenter, but growth is exploding in unexpected markets. Austin, Atlanta, Salt Lake City, and Las Vegas all saw more than 100% growth in data center capacity in just one year. Vacancy rates have hit a record low of 3%, and essentially every major project is leased before construction even begins.
"There appears to be no ceiling for how high this data center demand is going," says Andy Cvengros, a managing director at JLL who tracks the sector.
Europe is seeing its own surge beyond the usual hubs of Frankfurt and London. Construction recently began on a novel 5-megawatt data center inside an active Alpine mine in Italy, leveraging the naturally cool environment. In Germany, cryptocurrency mining company Northern Data is pivoting entirely to AI, considering selling its Bitcoin operations to fund AI computing centers.
Asia-Pacific is equally crucial. Google is investing $1 billion in new facilities in Thailand, while Oracle is putting $6.5 billion into Malaysian cloud infrastructure. Vietnam's military-run telecom Viettel plans 24 new data centers by 2030, and Australia is seeing one of the world's largest projects—a 504-megawatt campus near Sydney.
In Africa, the continent's data center capacity is expected to grow 50% by 2026, with countries like South Africa building "AI-ready" facilities powered by on-site solar and battery backups to ensure reliability despite power grid challenges.
The Economic Ripple Effects
The human impact of this infrastructure boom extends far beyond the tech industry. Communities are seeing genuine economic transformation.
In Abilene, local officials highlight how tax revenue from the Crusoe facility will fund schools and emergency services while creating high-quality jobs to anchor the region's future. In India, state authorities in Madhya Pradesh provided land and policy support for RackBank's AI data center, viewing it as a cornerstone of a emerging tech cluster.
These projects represent billions in investment flowing into places that were previously peripheral to the digital economy. The construction phase alone creates thousands of jobs, and the ongoing operations require skilled technicians, engineers, and support staff.
But the transformation isn't without challenges. In Texas, data centers already consume nearly 9% of the state's electricity production, and that share is rising. Power grids that have struggled during extreme weather events are now facing unprecedented new demands.
Balancing Ambition with Responsibility
The industry is acutely aware of these tensions. Nearly every new data center project emphasizes sustainability features—whether it's ultra-efficient cooling systems, on-site renewable generation, or waste heat recovery for nearby buildings.
Companies like RackBank are deploying immersion cooling technology and targeting Power Usage Effectiveness (PUE) ratings close to 1.1, where 1.0 represents perfect efficiency. Tech giants are signing massive renewable energy contracts to offset their consumption, and some facilities are pioneering direct integration with renewable generation.
Yet the overall footprint remains significant. Training a single large AI model can consume as much electricity as dozens of homes use in a year, and each major data center draws hundreds of megawatts continuously. The industry is increasingly focused on "sustainable AI"—not just ethical algorithms, but energy-efficient infrastructure.
The Infrastructure That Intelligence Built
The AI revolution, it turns out, is as much about concrete and steel as it is about code and algorithms. Every interaction with an AI assistant, every intelligent service we use online, quietly depends on an expanding network of data centers and power plants stretching from rural North Dakota to urban India.
This current building spree represents more than just technological progress—it's economic transformation on a scale that rivals previous industrial revolutions. The communities that successfully attract these facilities are positioning themselves as crucial nodes in the global AI economy. Those that don't may find themselves increasingly peripheral to the digital future.
As Chase Lochmiller put it when describing his Texas AI factory, these facilities will "enable intelligence to be manufactured with unprecedented speed and scale." It's a bold statement that captures something profound: our digital future is being built in physical infrastructure, and that infrastructure is reshaping both industries and communities worldwide.
The question isn't whether this transformation will continue—the billions already committed guarantee that. The question is which communities, which countries, and which energy strategies will power the next phase of human intelligence augmentation.
The AI revolution is happening in server farms and substations, in wind turbines and nuclear reactors, in small towns and megacities around the world. And for the first time in history, the manufacture of intelligence is becoming a global industrial enterprise.
Key Takeaways:
AI infrastructure represents the largest construction boom since the Interstate Highway System
Data centers are consuming city-level amounts of electricity, forcing utilities to completely rethink their capacity planning
Nuclear power is experiencing a renaissance as the only carbon-free baseload option for 24/7 AI operations
The economic impact extends far beyond tech hubs, transforming communities from rural Texas to central India
Creative solutions like wind-powered data centers are emerging to balance sustainability with massive energy demands
![]() | Nick WentzI've spent the last decade+ building and scaling technology companies—sometimes as a founder, other times leading marketing. These days, I advise early-stage startups and mentor aspiring founders. But my main focus is Forward Future, where we’re on a mission to make AI work for every human. |
Reply