LNG exports vs. data centers: The fight for Gulf Coast natural gas is here
Everyone is talking about data centers and natural gas demand. The hyperscale builds, the AI power load, the projections that stretch into the terawatt-hours.
It is a real story. But here is the part that is getting lost: LNG is winning, and it is not close.
The actual numbers
The U.S. Energy Information Administration is forecasting LNG gross exports to average 16.3 billion cubic feet per day in 2026. That is up from 14.9 billion cubic feet per day in 2025 and 11.9 in 2024. The Strait of Hormuz closure knocked Qatar, the world's second-largest LNG exporter, effectively offline for global markets. Gulf Coast terminals responded by running flat-out to fill the gap. Europe and Asia are actively seeking replacement supply, and American LNG is the primary answer.
Data centers, for all the attention they generate, are expected to drive roughly 0.5 billion cubic feet per day of incremental gas demand in 2026 through increased power burns. LNG export growth will drive 3.7 billion cubic feet per day of incremental demand in the same period. That is more than seven times the data center impact, on a shorter timeline, with contracted volumes already locked in.
At RBN Energy's GasCon 2026 conference in Houston in February, analyst Rusty Braziel ran the math on data center gas demand through 2030 and concluded that new LNG capacity additions will demand roughly three times as much gas as new gas-fired power plants serving data centers. The headlines favor data centers. The molecules favor LNG.
Where the gas is coming from
Meeting both sets of demand at scale requires supply growth, and the Haynesville Shale is doing most of the heavy lifting right now.
The Haynesville has some of the cheapest breakeven costs outside of the Northeast and Permian, and it is geographically positioned to feed Gulf Coast LNG terminals directly. East Daley Analytics projects Haynesville production to exit 2026 nearly 3 billion cubic feet per day higher than at the end of 2025. That ramp is designed to serve LNG demand.
The Permian is also adding supply, but pipeline constraints have been a limiting factor. New egress capacity, including the Blackcomb and Hugh Brinson pipelines scheduled for 2026 startup, will begin to relieve that pressure. Once those corridors open, Permian gas will have better access to the Gulf Coast market, which means more competition for storage capacity, more pressure on midstream infrastructure and more options for LNG terminals looking to diversify their feed gas sources.
Why natural gas power plant costs matter to this conversation
Data center demand is real, but the path from data center to gas consumption is longer and less certain than LNG's direct feed gas relationships.
A TechCrunch report published last week found that natural gas power plant costs have surged 66% in two years and that new plants are taking 23% longer to build than they did previously, driven by data center electricity demand outpacing generation capacity. That supply constraint on the generation side means some data center operators will struggle to get the power access they need on the timelines they want, regardless of how much gas is available.
LNG terminals, by contrast, have long-term contracted demand, interconnected pipeline infrastructure and feed gas agreements already in place. They do not have to wait for a power plant to be permitted, built and commissioned before the gas demand shows up. The molecule goes straight to the liquefier.
For Permian and Haynesville producers deciding where to direct production growth, that certainty matters. Long-term contracts with LNG terminals are structurally more predictable than the more variable demand signal from data center power loads. Several upstream producers, including EQT Corporation, have said publicly that they find data center power demand attractive precisely because of that consistency. But in 2026, LNG contracts still represent the larger, more contracted portion of new demand growth.
The data center story is real, just not the dominant force yet
None of this dismisses data center demand. Texas and Louisiana are among the states best positioned to absorb data center growth because they generate more power than they consume and have access to significant gas supply within their borders. The Gulf Coast will benefit from both demand streams. The question is sequencing.
LNG wins the near-term gas competition because it was already winning before the Hormuz crisis made the margin wider. Data center power load will be additive, and over the course of the decade, it could become a more significant factor than current projections suggest. AI compute demand has consistently surprised analysts on the upside. But in 2026, the gas is going to the terminals.
What it means for Gulf Coast operators and contractors
For midstream operators and pipeline developers, the implication is to build for LNG demand first and let data center load add on top of it. The infrastructure required to serve LNG terminals, high-volume pipeline capacity, storage flexibility and reliable interconnects, is the same infrastructure that will eventually serve data center-adjacent power generation. It is a build-once, serve-both opportunity.
For upstream producers in the Haynesville and Permian, the combined demand signal from LNG exports and data center power load is supportive of sustained production growth. The risk is on the infrastructure side. If pipeline and storage development lags the pace of demand growth, bottlenecks will appear, as they have before.
For Gulf Coast industrial operators more broadly, the key takeaway is that the regional natural gas market is becoming structurally tighter. The LNG export buildout is locking in long-term demand that will not flex when prices rise. Data center power load is adding on top of that. Facilities that have not reviewed their gas supply contracts and pricing exposure recently should do so.
The gas is there. The question is who has the infrastructure to move it and who locked in their supply before the market got this competitive.