NextFin

Kinder Morgan Commits $4.2 Billion Texas Gas Pipeline to Fuel the Data-Center Boom

Summarized by NextFin AI
  • Kinder Morgan committed $4.2 billion to build the 434-mile Permian Link pipeline, moving 2.8 Bcf/d of gas to power AI data centers by July 2030.
  • The company's expansion backlog hit $14.4 billion, including a record $13.2 billion in natural gas projects, with over 75% tied to power generation demand.
  • U.S. gas-fired power capacity in development jumped 50% to 378 GW by August 2026, with Texas alone accounting for 122 GW driven by data-center load.
  • Kinder Morgan shares traded near $34 with a 3.5% dividend yield, while Q3 net income rose 22% year-over-year to $764 million.

NextFin News - Kinder Morgan Inc. has formally committed to building a $4.2 billion natural gas pipeline across West Texas and the Panhandle, a 434-mile project designed to carry 2.8 billion cubic feet of gas a day to power a data center complex. The deal, announced Oct. 21, 2026, is the clearest signal yet that the U.S. gas-pipeline industry is being rebuilt around a single customer: the artificial-intelligence data center.

The project, dubbed Permian Link and routed along the Natural Gas Pipeline Company of America (NGPL) corridor, is among the largest natural gas pipeline projects to be constructed in decades, the company said. Its target in-service date is as early as July 1, 2030 — a timeline that puts it squarely in the window when grid operators expect data-center load to roughly double in Texas. The commitment lands on the same day Kinder Morgan reported record third-quarter results and a natural-gas project backlog that is the largest in company history.

At the announced size, Permian Link alone would move enough gas to support roughly 16 to 18 gigawatts of highly efficient gas-fired generation running around the clock — enough to power dozens of hyperscale facilities. That single number captures the scale of the shift: a midstream company is no longer sizing pipes to residential heating demand or industrial load growth. It is sizing them to the power appetite of the AI economy.

The Bottleneck Is Not Gas — It Is Pipe

The Permian Basin has never lacked gas. It lacks a way to move that gas, at firm pressure, to the exact node where a data center needs it, every hour of every day. That distinction is the entire economics of Permian Link.

West Texas gas has been trading at a steep discount for much of 2026 because takeaway capacity has not kept pace with production. Full-year 2026 pricing at the Waha hub — the Permian's benchmark — was still assumed at negative 26 cents per million British thermal units as recently as late July, after months of deeper dislocations that at times pushed prices below zero. Gas at the wellhead is cheap. Gas delivered reliably to a data center in the Panhandle is scarce. A pipeline is the machine that turns the first into the second, and the spread between them is what Permian Link monetizes.

The constraint is physical, not geological. New pipeline projects face multi-year permitting, steel and compressor lead times that stretched into 2027 and 2028, and a queue of competing expansions. While new capacity was being built, production kept rising. The result is a basin that is gas-rich and deliverability-poor — the exact condition that makes a 2.8 Bcf/d line with a committed anchor customer valuable enough to justify $4.2 billion of capital.

Kinder Morgan's own numbers show how decisively the business has pivoted. The company's expansion project backlog reached $14.4 billion in the third quarter, including $13.2 billion of natural gas projects — the largest gas backlog in company history. More than 75% of the total backlog is tied to projects supporting power generation and local distribution company demand. Two new additions alone — the Southeast Connector and Permian Link — are each designed to move more than 2.5 billion cubic feet a day.

"This quarter we increased our expansion project backlog to $14.4 billion, including the largest natural gas project backlog in company history at $13.2 billion," Chief Executive Officer Kim Dang said. "Two new major projects in the backlog include Southern Natural Gas Company's (SNG) Southeast Connector and Natural Gas Pipeline Company of America's (NGPL) Permian Link. Each project is designed to transport over 2.5 billion cubic feet per day (Bcf/d) of natural gas and will be among the largest natural gas pipeline projects to be constructed in decades."

The backlog math matters. It rose $4.7 billion in a single quarter, with $5.4 billion of additions net of $700 million of projects placed in service. The larger $13.7 billion portion is expected to generate a first-full-year project EBITDA multiple of about 5.7 times — the kind of contracted, toll-road return that midstream investors prize. The company ended the quarter with net debt-to-adjusted EBITDA of 3.5 times, at the low end of its target range, and guided 2026 adjusted EBITDA to $8.6 billion while expecting to finish more than 13% favorable to budget on adjusted EPS.

The storage connection built into the NGPL corridor is what makes the route attractive to data-center and power developers clustering along it, according to company commentary. For a facility that cannot tolerate even brief interruptions, storage-backed firm capacity is worth more than interruptible cheap gas. The pipeline is not a commodity bet; it is a reliability product.

This Is Structural, Not Cyclical

The right way to read Permian Link is not as a one-off win but as evidence of a regime shift in how the U.S. power system is being built. Three pieces of evidence point to a structural change rather than a cyclical spike.

First, the scale of the buildout is compounding, not peaking. U.S. gas-fired power capacity in development jumped 50% in six months to 378 gigawatts by August 2026, according to Global Energy Monitor. Texas alone accounts for 122 GW — nearly one-third of the U.S. total and more than any other country — with about 77 GW of that planned to directly power data centers. A separate tally put gas-fired capacity proposed, under construction, or announced specifically for U.S. data centers at more than 189 GW by mid-2026, nearly twice the 97 GW counted at the end of 2025 and more than 40 times the early-2024 estimate. Growth of that velocity is not a cycle; it is a new demand base.

Second, the demand is anchored in physics, not sentiment. A data center runs 24 hours a day, 365 days a year, at high load factor. Solar and wind cannot firm that load without storage at a scale that is not yet economical for most hyperscale sites. Natural gas combined-cycle plants can, and they can be built behind the meter or adjacent to the facility, avoiding transmission queues that have stretched to a decade in some regions. The gas plant is the fastest dispatchable answer to a load that cannot wait. That is why more than three-quarters of Kinder Morgan's backlog is tied to power and local distribution demand — the customers with the most urgent need for firm energy are the ones signing contracts today.

Third, the cash-flow structure has changed. Kinder Morgan's backlog is largely take-or-pay — the larger $13.7 billion portion is expected to generate a first-full-year project EBITDA multiple of about 5.7 times. That is utility-style contracted revenue, not merchant exposure to spot gas prices. The midstream company is not betting that gas will be expensive in 2030; it is collecting a toll from a customer that has already committed to pay.

"Our natural gas business in particular is benefiting from increasing LNG exports, rising power demand, and industrial expansion and is capitalizing on significant opportunities for investment across our footprint," Executive Chairman Richard D. Kinder said.

Peer behavior confirms the shift is industry-wide, not unique to Kinder Morgan. Williams has disclosed a portfolio of data-center power projects spanning gas supply, pipelines, and generation, with approximately 6 GW of potential projects in its backlog as of February 2026 and about $9.6 billion of projects in development. Energy Transfer and WhiteWater Midstream have advanced competing Permian takeaway lines, including the Blackcomb Pipeline, which reached a final investment decision in July with 2.5 Bcf/d of capacity. When three or four midstream companies are all racing to build gas infrastructure for the same end-use, the market has spoken louder than any single contract.

The cyclical counter-argument is real but narrower than it looks. Gas production, gas prices, and AI capital spending all cycle. If hyperscaler spending slows, or if the Permian overproduces and basis stays negative, a 2030 pipeline could face weaker-than-expected utilization. But the take-or-pay structure insulates the toll road from most of that risk — the shipper, not Kinder Morgan, absorbs the volume and price exposure once the contract is signed. The cycle that matters here is the AI capex cycle, and it is the shipper's cycle to manage, not the pipeline's.

The Second-Order Effect: Electricity Prices, Not Just Pipeline Tolls

The first-order effect of Permian Link is obvious: Kinder Morgan earns a contracted return on $4.2 billion of capital. The second-order effect is who pays for the electricity the pipeline enables — and that bill is larger than the pipeline itself.

The U.S. Energy Information Administration has warned that if data-center load growth outpaces planners' expectations, power prices will rise — and Texas is the most exposed grid in the country. Under a high-demand scenario, average wholesale electricity prices at the ERCOT North hub could be 78.9% higher than the $47.39 per megawatt-hour baseline forecast in the agency's Short-Term Energy Outlook. A pipeline that relieves a gas constraint lowers the fuel cost for gas-fired generators, but it does not relieve the transmission and generation capacity constraint that data centers impose on the grid. Cheap gas at the burner tip and expensive power at the meter can coexist — and in ERCOT, they increasingly do.

The EIA analysis put the mechanism in plain terms: regions managed by ERCOT and PJM Interconnection are expected to see the fastest data-center-driven demand growth through 2027, with ERCOT load forecast to grow about 10% annually between 2025 and 2027 in the baseline and up to 15% in the high-demand case. Compare that with the 0.1% average annual demand growth the U.S. saw from 2005 to 2019, or the roughly 1.7% annual growth from 2020 to 2025. The grid is being asked to absorb a load shock an order of magnitude larger than anything in recent memory, and the price signal is the rationing device.

That tension is why the watch groups are pushing back. "This frenzy of gas proposals is running headlong into the hurdles of an already tight gas market," said Jenny Martos, interim director of oil and gas at Global Energy Monitor. "It is nearly impossible nowadays to guess what is a pie in the sky proposal, and what has a real chance of getting built. The projects that eventually clear those hurdles are paying top dollar for turbines, locking in emissions, and pushing up electricity prices."

The emissions point matters for the AI industry's own narrative. Several hyperscalers have built their brand on clean-energy procurement, yet the fastest way to power a data center today is a gas turbine. Permian Link and its peers are a quiet admission that the AI buildout is being powered by fossil fuel at the margin, regardless of renewable purchase agreements signed elsewhere on the grid. A pipeline with a 2030 in-service date locks in gas dependence for decades, because pipeline assets are built to run for 40 to 50 years.

The Counter-Thesis: What If the AI Buildout Slows?

The strongest case against this story attacks it at the foundation: the data-center load that justifies 378 GW of gas-fired capacity may not all materialize. If AI revenue fails to keep pace with infrastructure spending, hyperscalers will cut capital expenditure, and projects still in development — as opposed to those already under take-or-pay contract — will be the first to stall. Global Energy Monitor's own caution that most proposals are "pie in the sky" is a direct challenge to the assumption that today's queue becomes tomorrow's throughput.

There is also a technology risk. Battery storage durations are lengthening, and small modular reactors and geothermal pilots are targeting the same 24/7 load. If storage economics improve faster than expected before Permian Link's 2030 in-service date, some data-center customers could have cheaper, cleaner alternatives than a gas pipeline locked in today. A data center sited in a region with abundant hydro or nuclear could bypass gas entirely.

The answer to both objections is the same: contract quality trumps queue size. Kinder Morgan's earnings do not depend on the 378 GW being built; they depend on the portion already under long-term contract, and the company says more than three-quarters of its backlog supports power and local distribution demand with utility-like multiples. The risk is not that every project in development gets built. The risk is concentrated in the developers who signed the contracts — and in whether their power purchase economics survive a slower AI revenue curve.

Even so, the counter-thesis deserves its due weight. If a major hyperscaler cuts capital expenditure guidance while Permian-basin gas basis remains deeply negative for four consecutive quarters, the market will begin to discount the backlog's convertibility. That is the scenario in which today's structural call would prove wrong.

What to Watch

For investors, the near-term signal is the conversion of backlog into final investment decisions and then into service. Kinder Morgan expects the backlog's average in-service date to cluster around the first quarter of 2028 for its largest projects, with Permian Link itself targeting July 2030. The company's third-quarter net income attributable to KMI was $764 million, 22% higher than a year earlier, while adjusted EPS and adjusted EBITDA rose 21% and 9%, respectively. Natural gas transport volumes were up 3% in the quarter, driven by LNG deliveries and stronger Texas Intrastate demand; gathering volumes rose 14%.

The falsifying signal is specific: if ERCOT's data-center interconnection queue growth stalls for two consecutive quarters, or if a major hyperscaler cuts capital expenditure guidance while Permian-basin gas basis remains deeply negative, the structural thesis weakens materially. Until then, the base case is steady conversion of contracted backlog into earnings through 2028-2030; the upside case is a fresh wave of final investment decisions as more developers hit the gas constraint; the downside case is an AI capex slowdown that leaves non-contracted projects stranded.

Kinder Morgan's shares were changing hands near $34 on the announcement day, close to the 52-week high of $34.81, as the market digested the backlog and the new project. The stock's roughly 3.5% dividend yield and a one-year analyst target near $36 suggest investors are pricing in steady growth but not a dramatic re-rating — a valuation that assumes the backlog converts without major slippage.

The deeper takeaway is this: Permian Link is not a bet on higher gas prices. It is a bet that the AI economy will need firm, fossil-fueled power for the rest of the decade — and that the companies owning the pipes, not just the electrons, will collect the toll.

Explore more exclusive insights at nextfin.ai.

Insights

What is Permian Link pipeline project?

How costly is pipeline construction now?

Why build gas pipes for data centers?

When will Permian Link start service?

What is Kinder Morgan project backlog?

Why is West Texas gas so cheap?

How does AI drive gas demand growth?

What risks face the pipeline plan?

How do peers compare to Kinder Morgan?

Will electricity prices rise in Texas?

What is gas supply bottleneck cause?

How does storage help data centers run?

What if AI spending slows down fast?

Are emissions a concern for AI power?

How does take or pay contract work?

How large is gas capacity needed today?

What drives 2026 gas project growth?

Is shift structural or cyclical trend?

What is the Waha hub price trend?

How does ERCOT handle load growth?

Search, ask or analyze anything...