The true scale of prospective projects that want to plug into the Texas grid is hard to hold in your head. As of August 3, the Texas governor’s office put the figure at more than 474 gigawatts of requests to connect, more than five times the grid’s record peak demand, and said roughly 90 percent of those requests are data centers. Many of those projects may never connect. Against that backdrop, the Public Utility Commission of Texas (PUC) published a proposal on July 24 that would place substantial payment obligations on the large customers seeking that capacity. It is a serious piece of work, but one important part of its scope remains unsettled.
On August 3, Governor Greg Abbott then directed the PUC and the Electric Reliability Council of Texas (ERCOT) to audit every data center moving through the interconnection process and said the audit must be finished before any project moves forward. Among other things, it asks the extent to which data centers are “paying their own way.” That question is aimed at tax breaks, grants, and public assistance. It does not reach who pays for the grid work needed to serve these projects. (That inquiry is in the proposed rulemaking.) In June, the governor told the PUC to require data centers to fully fund the electric infrastructure needed to serve them. How far that reaches is the open question.
Start with what the new PUC proposal would require, because it asks more of data centers than the public argument usually assumes. Before it connects, a qualifying large-load customer tells the utility how much power it wants served at that site. The rule calls that its contracted peak demand, and it sets the floor for what the industrial customer is billed. The bills start once the power is ready for it, whether or not it has switched anything on, and they run at least 20 years.
It would also put money down before ERCOT has even finished studying whether the project works. If it walks away, that money does not simply come back. The utility first takes out what it has already spent. Of what is left, the data center gets one-fifth back and the rest goes toward holding transmission rates down. A data center that actually runs at full size for five years gets its money back, and it would pay cash for the wires and substation work that connect it. It gets no discount on that bill, and the utility cannot turn around and charge those costs to anyone else. The commission says the point of all this is to keep ordinary customers from being unfairly burdened by the cost of connecting these projects.
Those are real obligations, and they are durable. This is not a proposal that waves large-load projects like data centers onto the electric grid for free. But it leaves one big question unanswered. When the grid needs work to serve a particular data center, does that data center pay for it, or does the cost show up on everyone’s power bill?
The proposal is clear about the wires and substation work right at the data center’s doorstep. That, the customer pays for. But ERCOT reviews these projects in batches, and it sometimes finds that serving one of them takes more than a hookup. A single large data center can draw an enormous amount of power, and the lines running toward that site may not be built to carry it. The fix may sit well beyond the property line, out on the grid everyone shares. The proposal does not clearly say who pays for that.
Think of it as the driveway and the highway. The PUC proposal is clear that the data center pays for its own driveway. What it does not settle is whether certain improvements, such as adding lanes to the connected highway, the ones identified as necessary to accommodate that new traffic, should also follow the direct-payment route.
That line matters in practice. Say a data center pays upfront for the work at its doorstep. Under the new rule, the utility cannot then charge everyone else for that same work through their rates. That is a sensible protection against getting paid twice for the same infrastructure. It also means where the line falls decides which work gets handled that way, and which does not.
This is not a problem someone had to go digging for; indeed, the PUC raised it itself. In the same proposal, the agency asked whether the definition of transmission interconnection costs should be broadened to include the transmission projects recommended at the conclusion of a batch study to serve the capacity needs of a particular load. The proposal leaves that question open and invites comments in Project 58000.
That split already exists elsewhere in the commission’s own rules. Under the the commission’s proposed companion connection rule, a data center pays outright, in cash, to connect itself. For bigger upgrades further out that are needed to serve it reliably, that rule has the customer put money down instead and says nothing about who ends up paying. The question now is where the first bucket stops and the second begins.
None of this means data centers would pay nothing toward the rest of the grid. They would pay through those minimum bills and through the rates every kind of customer pays into.
That is why this reaches past the utilities and the data centers. The costs that get spread out land partly on households and small businesses, alongside everyone else. This is not a prediction that anyone’s bill is going up. Nobody knows yet how much money is at stake here because ERCOT has not named these projects, let alone priced them out.
There is a real difficulty here, and it deserves to be stated at its strongest. Some broader projects identified through system-wide planning may serve multiple customers or strengthen the grid generally, which can make it genuinely difficult to assign their cost to a single load. Planning does not always sort neatly into who caused what.
But difficulty is not a reason to treat the physical point of connection as the automatic outer limit of a customer’s direct responsibility. That is the position worth arguing: Broader projects identified as necessary to serve a particular load should not be excluded from that customer’s direct responsibility merely because they extend past the wire that connects it. If the planning process names a project as necessary to serve a specific customer’s capacity, the fact that the steel sits farther down the road is a poor reason to categorize it differently.
Comments closed August 11, and here is where the line should start. When ERCOT’s planning identifies work as necessary to serve one data center, that data center should pay for it, unless the commission finds the work does enough for everyone else to justify sharing the cost.
The post Who Pays for the High-Powered Grid that Data Centers Need? appeared first on The Texas Observer.
Leave a Reply