Executive Viewpoint: Water is the other megawatt

DAVID HODGES, Chief Operating Officer, LP Energy Services Group

Stand at the fence line of one of the new AI campuses going up in Texas and forget, for a minute, that the product is data. Look at what is actually being built: dedicated power, fuel supply, cooling plants, miles of process piping, water treatment, controls, redundancy, and a commissioning sequence where every system has to work together.

That is not an office building with servers in it. That is a refinery.

The product is computing capacity instead of gasoline, but the infrastructure is industrial. A year ago, we talked about data centers in megawatts—a 100-megawatt campus got everyone’s attention. Today, Crusoe’s Abilene campus is building toward 1.2 gigawatts, Vantage’s Frontier campus one county over is planned at 1.4, and ERCOT is tracking more than 438,000 megawatts of large-load requests, nearly 89 percent of them from data centers. Not all of it will get built. That is not the point. The point is that the scale changed almost overnight—at the top end, these campuses will draw more electricity than many conventional refineries. This is the Industrial Technology Revolution, and it runs on the same two inputs every refinery ever ran on: power and water.

I have spent more than three decades starting up energy and industrial projects—refineries, upstream, LNG, and now data centers—and I tell my team the same thing on every job: take what we learned in refineries and apply it to the data centers, because in most of the ways that matter, they mirror each other. So does the first rule I ever learned in this business: water and electricity never go together. Keep them apart, always. Yet on a project, they are never apart. Power runs the facility; water cools it, cleans it, tests it, and fills the piping. They sit in different work packages, under different teams and different budgets, and they meet at startup, on the same critical path, when the schedule has the least room left. Power is measured in megawatts. Water is measured in gallons. A project can lose a gigawatt because the gallons were never planned. That is why water is the other megawatt.

Power, at least, is getting the attention it deserves: boardroom oversight, dedicated contracts, interconnection studies, its own dashboard. The grid was not built to absorb this much new load this quickly, and no serious developer assumes the electrons will simply show up. But at LP Energy Services Group, we are not going to point at the grid and call it somebody else’s problem. The need is large enough that we are working to build our own power-generation division. The market is full of used generators—some of them good machines—but a used machine sitting on a pad is not a power solution. What these projects need is purpose-built generation engineered around the actual load: fuel, controls, switchgear, protection, emissions, the commissioning plan, and long-term service. Some campuses will run on utility power, some will need bridge power or permanent onsite generation, and many will need a combination. The answer, in every case, is a complete power system engineered to start, connect, carry the load, and stay online—behind the meter, taking part of the burden off the grid.

Water has the same problem with less of the spotlight, because it hides in the utility, mechanical, environmental, or commissioning scope. It should not hide. Berkeley Lab estimates U.S. data centers directly consumed some 66 billion liters of water—about 17 billion gallons—in 2023, before counting the water consumed to generate their electricity. Every campus and every cooling design is different, but this stopped being a small utility line item some time ago.

The oil field makes the point even clearer. A modern horizontal well can take 10 to 15 million gallons of water to complete. Once production begins, many Permian reservoirs give up three to five barrels of water for every barrel of oil—water often several times saltier than seawater—and the basin now produces more than 20 million barrels of it every day. That is too much water to keep calling waste. Not every stream needs to become drinking water; in the oil field, the objective may be to clean and condition saltwater so it can go back into the next completion, and Texas already allows and encourages exactly that. The question is not always whether water can be made potable. The question is what it needs to become useful again.

We are attacking that side of the problem with high-volume filtration and modular reverse osmosis. We engineered our filtration system around a refinery problem: roughly 2.5 million gallons of brackish water headed for the wastewater system in a region where water was scarce. We built a system that filters the full volume in 22 hours and puts it back to work. When potable quality is required, each of our modular RO units produces 250,000 gallons a day from brackish or salt water—500,000 a day as a pair. And we do not guess at finished quality: we pull the local tap-water analysis, set it as the baseline, and try to beat those readings. Water already on a site should not be thrown away simply because nobody planned to clean it.

We brought that same thinking inside the fence at a recent data center campus in the Dallas area. The original approach included a separate water-reclamation system with millions budgeted against it. Our process eliminated the need for it. We drained and captured the water in the piping, filtered it onsite, ran the precommissioning and commissioning sequence, checked the finished water against the local tap-water baseline, and returned it to the client to charge the lines. We have the process down to a science—in most cases we are in and out in about 30 days, and based on our field experience it uses roughly 90 percent less water than conventional methods for the same work. The reclamation system never got built. The money it would have cost covered our entire scope; the schedule came back, and the client kept the budget originally set aside for the piping work—no reclamation spread, and no paying to dispose of water only to buy it again. That is not a sustainability gesture. It is schedule, cost, and execution.

The technology already exists. The failure is waiting until the project is behind schedule to use it. In all my years in the field, I have rarely seen a project fall behind because technology did not exist; projects fall behind because nobody owned an interface. So put power and water on the same executive dashboard and decide early: what water must be potable, what can be reused, what needs RO and what only needs filtration—what power comes from the grid, what gets generated onsite, and what bridge carries the load until permanent infrastructure is ready.

The companies that win the Industrial Technology Revolution will be the ones that treat water and power as twin strategic resources—turning gigawatts into operating capacity by planning for both. Power is the megawatt everyone is now watching. Water is the other one—still hiding in the pipe rack. Both belong on the dashboard.

DAVID HODGES is Chief Operating Officer of LP Energy Services Group, a Houston-based industrial and energy services company supporting power generation, water treatment, EPC, precommissioning, and commissioning across upstream, LNG, data-center, and heavy-industrial projects in the United States and international markets.