The sense that “gas is back” has permeated the data center and energy worlds this year, fueled by surging demand from data centers and a rush to secure reliable power — and by the very pro-gas Trump administration. While gas still faces some challenges, gigawatts are now speeding to grid connection via “fast-track” interconnection processes, leaving cheaper, renewable alternatives in their wake.
Faced with a deluge of data center load, PJM, MISO, and SPP have all opened these faster queues to allow new generation to bypass the bottleneck of the standard study process. These are designed to get gigawatts of capacity online in the next three to five years, and specifically target the windows where grid operators project acute resource adequacy shortfalls.
These programs are nominally resource-agnostic, open to all forms of new generation. But early data shows that the biggest user of these express lanes, at least so far, are gas projects. In MISO, which has the longest-standing fast-track program of the three regions, gas accounted for 82% of the first round of projects. In PJM and SPP, whose fast-track programs were both approved earlier this year, gas projects make up 69% and 62%, respectively.
That’s not a surprise; the programs are structurally biased toward “firm” resources via accreditation. Each grid operator has its own accreditation process, but in general, right now they’re ranking gas more highly than other forms of generation, explained Patrick Huang, a senior research analyst at Wood Mackenzie.
While natural gas is facing historically long lead times for critical equipment and soaring prices, fast-track queues are getting existing projects online faster. “For many of these gas projects being pushed along these fast-track queues, they already had a turbine order secured in advance,” said Huang. “I would be very surprised if there were any projects that didn’t have a turbine order in place that were being considered.” In other words, the programs aren’t spurring net new generation, just getting gas that would already be coming online sooner.
Many of the gas projects in MISO, for example, are in Entergy’s territory, where the utility has already been doubling down on gas to serve data center load. MISO’s expedited resource addition study, which is currently evaluating the second cycle of projects, has effectively provided those gas projects with a mechanism to bypass the years-long process that still traps their renewable energy competitors.
These fast-track initiatives are arguably serving their intended purpose of getting more power to the grid faster, but their preference for “firm” power may be coming at the expense of market competition.
But there’s a debate over whether ISOs should be diverting time and resources to facilitating “side queues” when there are so many challenges facing the primary interconnection queues, Huang explained. There’s certainly a role for gas to play, particularly now when customers like data centers are willing to pay such a premium for baseload power, he added, “but from a cost point of view, if you want to serve more gigawatts of demand growth, renewables and storage are going to be the cheaper way to go.”
That remains true even after federal tax credits were phased out via Trump-backed anti-renewables legislation, while turbine shortages and rising capital costs have pushed the cost of new gas generation higher and higher.
There’s concern among consumer advocates that, in the rush to secure reliability, regions are over-correcting toward the most expensive, long-term option, leaving cheaper resources to flounder in the main queue, while locking in infrastructure payments for a much more expensive asset class.
While hyperscalers are increasingly signing bilateral deals for power, much of this fast-tracked capacity is utility-owned and rate-based. This is central to the conversation over how to ensure ratepayers don’t bear the brunt of the AI boom: Gas isn’t a temporary or “bridge” solution to data center demand, because the cost of a plant built on a fast-track timeline today will be amortized across a rate base for decades.
The implication, then, is that if projected AI power demand fails to materialize or softens, utilities and their customers could be left subsidizing reliability that doesn’t meet the needs or the reality of the grid.
