A smarter way to bring data centers to Colorado

 

Texas — yes Texas — may offer an approach that could best serve Colorado’s needs for clean energy and low-water impacts

 

David Hurlbut

Again, a bill to incentivize data centers failed to make it through Colorado’s legislative session. Criticisms of the bill were on point, and the issues aren’t going to change next year.

• The state budget can’t afford to give tax breaks to big data centers.
• Everyone’s electricity bills would go up because utilities would need new infrastructure to meet data centers’ appetite for power.
• Colorado’s progress towards decarbonization would stall.
• Using water consumptively for cooling would strain the state’s already limited supply of water.

In his remarks at the end of the session, bill sponsor Rep. Alex Vasquez (D-Denver) spoke longingly of the economic growth Texas has seen from data centers. But the Texas story is complicated, and it has lessons that many in Colorado miss.

Texas offers data centers two carrots. One is tax breaks, which the state comptroller estimates will amount to $1.3 billion this year. That’s huge even for Texas, and it’s beginning to worry pro-business legislators.

The other carrot is letting data centers negotiate their own electricity supplies instead of dealing with a utility. This isn’t controversial because everyone does it already. Even if the tax breaks disappear, data centers can still shop for savings on the wholesale electricity market and be better off than they would be in other states.

I helped put in place many of the rules that made electricity choice possible in Texas. I was a senior economist with the Texas Public Utilities Commission when the Electric Reliability Council of Texas (ERCOT) became a fully functioning wholesale electricity market in 2001. ERCOT operates the grid serving most of Texas, including Dallas, Fort Worth, Houston, Austin and San Antonio, where today most of the state’s big data centers are located.

We broke up the state’s largest vertically integrated electric utilities, set up rules and systems enabling customers to choose their electricity suppliers, and ensured that competitive generators and retail electricity providers had access to the ERCOT transmission system. Large customers like data centers have economic leverage and can shop for the lowest-priced electricity service. That’s a big reason industrial customers in Texas pay 28% less for electricity than those in Colorado.

The model works, but it’s difficult to replicate in Colorado—not because it can’t be done, but because state leaders don’t think in terms of wholesale electricity markets. Like other western states, Colorado is politically inured to a cost-of-service regulatory model. Electric utilities are vertically integrated (one company generates the power, delivers it to customers, and handles everything between) and regulators approve all aspects of utility operation (customer rates, types of service, new power plant and transmission investments, profits returned to investors). In exchange, the utility has a monopoly franchise that protects it from competition.

Texas, on the other hand, lets market economics do much of the heavy lifting with respect to investment decisions and customer prices. The only monopoly franchise is for the wires. Many generators compete on the supply side to create electricity. Many retailers compete on the demand side to serve end-use customers. ERCOT is the market platform bringing them together physically and commercially.

This is how Texas led the nation in the growth of wind generation during the past two decades and solar generation since 2020, two trends borne out by data from the Energy Information Administration. In Texas, renewables are often the go-to option for large-load customers such as data centers, not because the state orders it but because they can reduce the cost of doing business.

It’s possible for Colorado to follow suit, but as Steve Jobs said, you have to think differently. Here’s how.

• Give large-load customers like data centers the option to leave their electric utility and negotiate their own power purchase agreements, provided they connect to the transmission grid directly and participate in one of the wholesale power markets operated in Colorado by the Southwest Power Pool (SPP).
• Direct the Colorado Public Utilities Commission (PUC) to create new transmission corridors connecting competitive renewable energy zones (CREZs) to the grid.
• Provide regulatory clarity on state expectations for data center design (including cooling), siting, and community engagement. Knowing the public policy guardrails in advance helps a business address them cost effectively.

The economic engine behind this approach is that the costs of new wind and solar are less than what Xcel Energy currently charges large customers for energy. This would be especially true for new generators in a CREZ, where economies of scale and competition would push costs lower.

The SPP markets would connect data centers with their suppliers and would provide any supplemental energy needed in real time. Like any large off taker, the data center would pay its share of transmission and operating costs.

The PUC would not review contract prices, but it would set rules to ensure data centers don’t take away from renewable energy serving other customers. It would also set reporting requirements for tracking greenhouse gases.

Grid reliability—keeping the lights on—would be managed better under this approach than under last session’s bills. That’s important. In May, the North American Electric Reliability Corporation (NERC) issued a rare level 3 alert concerning the growth of large data centers across the country. The issue: treating large data centers as ordinary demand could compromise grid planning and operations. NERC said big data centers are unique by virtue of the size and variability of their demand, and planners need a better understanding of how swings in their power usage can affect the grid’s reliability and stability.

NERC’s warning was aimed primarily at operators such as SPP, which plans transmission, operates markets, and coordinates reliability across a 17-state region in the Western and Eastern Interconnections. SPP recently established special requirements for so-called high-impact large loads that address some of NERC’s concerns, and just last week it received federal approval for non-firm transmission service to high-impact large-load customers (where SPP can cut service to a data center in order to prevent disruptions to other customers).

Therefore, a state policy that puts large data centers in Colorado under the SPP umbrella would provide a developed, geographically larger, and more efficient framework for making sure data centers don’t jeopardize grid reliability.

Xcel has offered data centers a baby step in the direction of choice. Its proposed new Transmission Large Service tariff, submitted to the PUC in April, would require direct connection to the transmission system and would charge the data center for Xcel’s cost of building it. The proposal would charge the data center for energy used, and it includes an option allowing the data center to negotiate power purchase agreements with outside suppliers.

But the choice of outside suppliers is limited to clean energy resources located in Xcel’s service territory that the utility doesn’t want. Perhaps the bids were too high, perhaps the available transmission capacity is too limited. Likely, the leftovers would be more expensive than the clean energy resources Xcel took for itself. Wind and solar not connecting to Xcel’s transmission lines would be off the table. It would be choice managed by a monopolist, nothing that would foster competition and entrepreneurial innovation.

Nor does Xcel’s proposal do much for transmission. A CREZ, on the other hand, would attract competition and boost the supply of low-cost wind and solar for everyone. Choice would actually mean something because there’d be a critical mass of suppliers and a contestable market.

Colorado’s data center problem isn’t going away and is bound to return in the next legislative session. The best way to solve it isn’t to grab the state money shovel, but to strategically link data centers’ need for electricity with new clean energy supplies in a regional market. No need for state tax breaks, no collateral costs for utility ratepayers.

Best of all for climate change, data center demand would become a powerful engine for renewable energy growth in Colorado. This—and not tax breaks—is the smart way to bring data centers to Colorado.

David Hurlbut retired from the National Renewable Energy Laboratory in 2024 and was the lab’s senior expert on wholesale power markets.