When Paradigms Collide: Interconnecting Energy and Data in the Industrial Compute Age
So much noise being made at the intersection of artificial intelligence (AI), hyperscale data centers and the legacy power grid. Most of us hear it loud and clear like a clarion call.
Some voices are telling us that trouble is ahead, that the energy sector isn’t ready for AI. A few are saying there’s no problem, nothing to see here.
Few of those experts in AI or energy have worked extensively across both sectors, much less at the same time. Everett Thompson, founder and CEO of data center site broker WiredRE, has worked with hyperscalers, utilities, colocation developers, power engineering firms and communications infrastructure builders throughout his career.
Thompson clearly recognizes the sense of criticality facing the nation with dozens of gigawatts forecast to come online this decade even though the electric utility sector was anticipating flat load growth. No doubt there’s a challenge happening as power generation and delivery utilities grapple with artificial intelligence and quantum computing, yet maybe one answer does not fit all when it comes to satisfying data center load of the future.
We are entering an Industrial Compute Age, Thompson says. And the U.S. must stay ahead to survive as a technology leader in this unprecedented era.
The national interest cannot be served by software or private industry alone. The nation needs a new paradigm to meet the challenges of supplying enough energy while also leading the new age of AI, quantum and cloud computing.
“When you show up with experience in all those different layers, your perspective is often quite different,” Thompson said in an exclusive interview with EnergyTech. “It’s not a simple answer, and it’s almost always a regional answer.”
Creating a new paradigm for energy and big tech
The new paradigm may revolutionize the power grid like nothing since the first half of the 20th century. Data centers may revert to on-site power and become their own mini-utilities, Thompson predicted, and governments will need to play a major role as they did the evolution of railroads and power plants of the past.
Thompson and WiredRE are engaged in huge movements around the data center and cloud-based expansion happening now and anticipated for at least the next decade or beyond. The company recently acquired, re-developed and then sold a 1-GW AI-focused site west of Dallas that will house Google’s servers, creating perhaps the biggest single interconnection yet into the Electric Reliability Council of Texas grid system.
One gigawatt is massive by anyone’s calculations yet only a fraction of what’s coming, according to data center growth forecasts from researchers such as McKinsey, Goldman Sachs, Bain and others. At the same time, and with an inverse pace of movement, the electric utility sector spent a decade accepting the notion of flat load growth, now finding itself potentially outmatched for AI and quantum computing demand.
“Nobody was paying much attention to it even two years ago,” Thompson said. Interconnection challenges are time-consuming and that hasn’t changed.
“Everybody still wants an interconnection agreement,” he said, referring to energy resources connected into the primary macro grid, “but dealing with a slow-moving utility is difficult. That’s not a cost issue, that’s bureaucratic. How do we overcome that in a way that our entire nation can be elevated?”
AI and quantum computing certainly are key to national security and economic leadership. China recently unveiled its DeepSeek AI technology that showed that the energy density of computing demand doesn’t have to be an consumption hog.
Creating opportunities to boost resources and renew infrastructure
And perhaps solutions in the national interest are, like most politics, truly local at heart. Data centers are a cluster industry, often intensely focused geographically in regions like northern Virginia and Texas, which facilitate interconnection into massive grids.
The location of data centers, however, also offers opportunities for microgrids, engaging a multitude of resources from gas-fired power to renewables, battery storage and both conventional, restarted nuclear plants and future small modular reactors. Texas and West Virginia are among the states looking into this.
“It’s going to be a boon for a lot of different resources,” Thompson replied. “I think self-generation is being adopted. . . This is an amazing opportunity to renew our infrastructure and our regional industries.”
Revolutionizing power at the edge of the grid is a good move, many say, but not the only one. Collaboration is key as federal, state and local political leadership are necessary to help fund an industrial revolution-level change on the seismic level of AI and quantum, he added.
What was not known then must be learned now and at hyperspeed pace as well as hyperscale volume.
“How do you turn the ship overnight? It must be government leadership,” Thompson pointed out. “It’s going to take public-private partnerships. If this is a moon shot, which I think it is, the government needs to create some sort of support.”
AI forces acceleration of all the game plans
In his quarter-century career cutting data center, financing and infrastructure deals, Thompson has consulted on behalf of customers as diverse as Iron Mountain, AT&T, Amazon, Goldman, Citibank and the U.S. Department of Energy. His cloud and data expertise were polished working for and with site developers such as Equinix, CoreSite, Rackspace, Digital Realty and others.
Thompson started his career with global engineering giant Foster Wheeler, spending ample time within power plant projects. These experiences create the bridge into both worlds of data center load and power generation capacity.
And yet, while he tends to accept the forecasts of 50-GW future load growth because of AI and cloud computing, Thompson also sees opportunities for micro-efficiency and local generation.
“AI is like a different voltage,” he said. “Acceleration is making the existing infrastructure more powerful.”
In fact, he’s high on the notion of on-site power for this so-called Industrial Compute Age.
Sometimes the new is forced to use the old when alternatives are not quickly available. In one controversial example, Elon Musk’s $6 billion Colossus AI supercomputer in Memphis, Tennessee, is being powered by a host of simple-cycle gas-fired turbines which are drawing the ire of local and national environmentalists.
Meanwhile, in a show that natural gas must play a role in the near future of AI expansion, major energy players such as Chevron are exploring gas-fired interconnections while others, such as DC Grid, are looking at direct-current and off-grid projects.
No one size, resource or capacity fits every data center capacity challenge. China’s DeepSeek has shown that efficiency can be part of power, too.
“The U.S. is so resource rich that sometimes it just throws resources at things,” Thompson noted.
But the resources are there or can be, with public-private partnerships developing an “all of the above” approach utilizing renewables, historically rich gas reserves, future nuclear, carbon capture, battery storage and perhaps even hydrogen.
Not all of this hype will play out the same but it will play out fast
Aim for the stars and settle for the moon. One true, but lesser known factor is that not all of these data center developments are going to work out and survive the design, construction or interconnection phases. Money talks and hype walks.
“I’m talking about developers who have no money who have made applications to AEP (American Electric Power) and other utilities,” he said. “Some of these projects are not going to happen.”
Many will, though, because competitive momentum, national and economic interests will propel AI, quantum and cloud capacity forward into yet unknown realms. The future of disruptive technology cannot and has never been reversed or silenced.
Disruption in data will just force things to be different, including the energy sector.
“Enterprise data centers are going away, and everybody is using the cloud; it’s a very fascinating time,” Thompson said. “There will be some swings and misses, but no question the genie is out of the bottle and we’re in an Industrial Compute Age.
“It’s hard to imagine that’s going to stop,” he added. “Quantum computing is coming faster than people expected, and it could be a wonderful stairstep, changing the relationship between computing and power in a positive way.”
The Global Battery Storage Race is On
Read more with this free EnergyTech E-Book
