This week, I’ll show that ~5.2 Bcfd of pipeline capacity will reach the constrained Southeast by year-end 2029,1 while new gas-fired power plants there will likely burn only ~1.1 Bcfd. But enabling demand growth is only part of the point: Southeast utilities contracted for this capacity ahead of their need for new gas supply, to reduce costs and fill shortfalls in their existing firm transportation portfolios.

How much pipeline capacity is coming online?

Six major Southeast-focused projects will add ~7 Bcfd of pipeline capacity, mostly backstopped by the region’s power utilities (Figure 1). But only ~5.2 Bcfd of that is effective capacity2 into the region, for two reasons:

• MVP Boost terminates at Transco Station 165, so its ~600 MMcfd only reaches Southeast markets via Transco or MVP Southgate.

• SoNat’s ~1.3 Bcfd South System Expansion 4 sources gas from TGP’s Mississippi Crossing project, so it doesn’t add supply into the region.

Figure 1 | Pipeline projects into the Southeast and capacity by shipper

Read more

EQT is paying a ~21% premium, in stock, for Range. On yesterday’s deal call, EQT CEO Toby Rice emphasized Range’s high-quality inventory, while Range CEO Dennis Degner pointed to EQT’s scale. But by paying a premium for Range’s inventory, EQT is implicitly conceding that its own wasn’t enough to keep growing capital-efficiently — just as Range, by accepting EQT paper, is conceding (as I highlighted in May) that it had no realistic path to scale in Appalachia.

Scale and upstream-midstream integration have been EQT’s flywheel: over the last seven years, it has outgrown every other major Appalachian operator in both volume and percentage terms.

Figure 1 | Appalachian equity production by operator

Scale is particularly important in a constrained basin,1 because managing those constraints means sometimes taking a hit — whether via shut-ins or out-of-the-money FT — on a fraction of your volumes, for the benefit of your broader portfolio. Only the largest operators can do that at enough scale to affect whether the basin is constrained consistently or only intermittently.

EQT has managed these constraints more holistically than any peer, and the Range deal reinforces that. It improves the economics of EQT’s curtailment strategy, strengthens the company’s negotiating position with gathering and processing counterparties like Energy Transfer and MarkWest, and enables more capital-efficient growth in southwest Appalachia by moving its best inventory into its cheapest outlets.

The central bank of Appalachian gas

Read more

Depending on which constraints bind, eastern prices fall into one of five paradigms, ranging from gas-on-gas competition at the low end to gas-oil competition at the high end. Now, two cohorts of pipeline projects are under construction to bridge these constraints: Northeast-to-Southeast (green and blue lines in Figure 1) and Gulf-to-Southeast (purple lines). They will reshape the region’s gas flows and how often each paradigm prevails, driving basis differentials in the years to come.

Figure 1 | Eastern constraint regions and forthcoming pipeline projects

Read more

This slowdown in structural gas growth is concentrated where CCs are most heavily utilized. And while today’s coal retirement delays reflect market fundamentals more than politics, the thin pipeline of new CCs is a lagged effect of the early 2020s’ decarbonization push. As a result, relief in power markets — and with it, faster gas demand growth — is still two years away.

Where is CC capacity heavily utilized?

Nationwide, CC capacity is running at its highest-ever utilization rates: ~60% annually and ~75% in July, up from ~50% and ~65% as recently as 2017. Even at these levels, CCs can ramp up during unusual conditions like extreme weather or low gas prices. But in most of the country, CC utilization is now too high to retire much additional coal capacity.

Figure 1 | US CC capacity factors by state

How is the fossil fleet running now?

CC utilization rates are higher still in most of the eastern half of the country, which spans both regulated and deregulated markets, gas-rich and gas-poor states, and varying degrees of political emphasis on decarbonization. Across the eastern US states bounded by Mississippi, Minnesota, Pennsylvania, and Georgia (top-left panel in Figure 2), CC capacity factors have risen sharply alongside a consistent buildout in CC capacity from 2003-23. Over the last 12 months, this ~107 GW fleet ran at a ~70% capacity factor — harder than the region’s coal fleet did in the mid-2000s.

Conversely, in California and Nevada (bottom-right panel), where sunny weather makes solar development more economic, CC utilization has fallen. In Florida (bottom-left panel), where strong load growth is nothing new and utilities have little interregional transmission to fall back on, the CC buildout has kept pace with demand and coal retirements.

Because load growth outran what generators expected, all dispatchable capacity — including coal — is now more valuable. For the first time in the shale era, coal capacity factors are also rising: from a (non-pandemic) low of ~42% in 2023 and 2024 to ~47% over the last year, the highest since 2019. Rising utilization, more than politics, is what’s keeping more coal plants open, at least until new gas capacity is in place.

Figure 2 | Gas CC and coal fleet capacity factors by region

Read more

But the bullish gas demand themes haven’t come true yet. For both LNG and gas-fired generation, rising capex translates into accelerating demand only after a multi-year lag. As I’ll show, gas demand growth should set an all-time record in 2028 and average about twice the current pace over 2028-31. This year’s solid-but-unspectacular demand growth hasn’t pushed prices upward, but the much faster 2028-31 pace will likely require sustained prices above $4/MMBtu.

LNG: 2025 saw record FIDs, but demand acceleration comes in 2028-31

Almost 8 Bcfd of US liquefaction capacity took FID in 2025 alone, the highest figure on record, and more than total global FIDs in all but two years historically. Even if half of these FIDs would have landed in 2024 without former US President Joe Biden’s ill-advised pause on non-FTA export approvals, the US would still have seen a four-year streak of at least 3.5 Bcfd of liquefaction FIDs. Over the entire 2012-21 period, US LNG FIDs reached that level just twice.

The FIDs at the start of this window are already adding demand: US liquefaction capacity additions averaged ~3 Bcfd in 2025-26. But neither 2025 nor 2026 will match any year in 2028-31, and 2028 alone nearly equals the two combined.

Figure 1 | US1 LNG FIDs and demand growth

Power: Gas demand growth has softened in recent years but will accelerate in the medium term

While US LNG demand will accelerate from today’s already-strong levels, gas-fired generation will accelerate from a much weaker base. Weather and year-to-year swings in gas prices obscure the underlying trend, so I estimate structural gas demand from load growth and capacity changes for competing fuels.

Load growth has accelerated sharply since 2023, adding ~1.5 Bcfd of gas demand annually. But solar generation climbed alongside it, and coal retirements have recently slowed. The net result: after growing ~1.5 Bcfd annually in 2020-24, structural gas demand growth averaged just ~450 MMcfd in 2025-26.

Figure 2 | Structural gas-fired generation growth by driver

That coal retirement trend is unlikely to reverse soon. Over the last year, the Department of Energy issued 202(c) orders keeping ~3.5 GW of coal capacity slated for retirement online, and utilities have delayed or canceled another ~28 GW on their own. Rising CC utilization rates leave little spare capacity in many markets,2 so utilities need to build new gas-fired capacity before they can retire more coal.

Those investments are now underway. Gas combined-cycle capacity under construction (teal bars in Figure 3) collapsed from ~25 GW in 2016-17 to ~2 GW in 2023-24, before recovering to ~7 GW by mid-2026. Although CCs are faster to build than liquefaction, they are still multi-year projects, meaning that CC additions are unlikely to accelerate until the second half of 2028.

Figure 3 | Gas CC capacity under construction at year-end

Gas price implications

The market is weak this year mainly because of production — I agree with that much. But I’m still not worried about sustained sub-$3/MMBtu prices, for three reasons:

• The market isn’t that weak: the 12-month strip still trades ~$3/MMBtu. It almost never makes sense to talk about prompt prices for a seasonal commodity — especially one becoming more seasonal because storage is under-built — whether that’s the February contract at ~$7.50/MMBtu early this year or the November contract at ~$3/MMBtu now.

• During E&Ps’ capex planning season last fall, 2026 futures at ~$4/MMBtu called for strong growth. Now, the 2027 strip trades at just ~$3.10/MMBtu. Strong growth at $4 doesn’t worry me; strong growth next year would.

• Nothing this year rises to the level of the winter 2023-24 shock, but conditions have been modestly bearish: winter ran slightly mild, and Golden Pass kept slipping relative to the in-service date the market priced in last year.

In the medium term, gas-fired generation growth should accelerate again. Demand for artificial intelligence tokens is still rising, which points to continued fast load growth. Capacity trends will support gas burn too: solar and especially wind development are slowing now that subsidies are gone, and coal retirements should resume once CC capacity catches up. Across all sectors, US structural gas demand growth should peak in 2028.

Figure 4 | US structural gas demand growth by sector

That single year matters less than the sustained 2028-31 pace: ~6 Bcfd annually across 2028-31, almost twice this year’s rate. In that 2028-31 period, each year’s demand growth is likely to exceed the record to date. Consecutive years of strong growth compound the challenge for E&Ps: every year of growth steepens the following year’s base decline. That means more capex just to hold production flat, before funding any growth on top — and therefore higher prices.

So yes, gas demand has picked up in 2025-26, from the anemic 2023-24 pace back to the typical 2016-22 rate. But the truly bullish demand thesis is underway only in capex; the demand itself is still two years out. And if the 2028 acceleration isn’t priced into the 2028 strip by next fall, when E&Ps set capital budgets, then the 2028 price move will look less like 2025’s structural shift upward and more like the 2021-22 spike.

Storage is now so low that I expect pipelines into the region to run near capacity through the end of winter. Basis differentials will remain in the $0.50-1.00/MMBtu range, with heightened risk of spikes to $2+/MMBtu. New pipelines under construction will ease these constraints over the next year.

Where and why have prices been high?

Capacity into the region1 totals ~15 Bcfd across seven major pipeline routes: Transco (from the north and the south), TGP (mostly from the north through Station 860, with small volumes from the south through Station 523), Midcontinent Express, SoNat, Florida Gas, Southeast Supply Header, and Gulf South. Production in the region adds ~1 Bcfd.

Figure 1 | Southeast constrained area and key pipeline routes into the region

Last summer, that capacity comfortably served Southeast demand and refilled storage, with ~900 MMcfd of spare capacity into the region. Two fundamental shifts since then have left the region short:

Read more

National Fuel is unique among major US gas utilities in having such a substantial upstream and midstream business,1 and Seneca — National Fuel’s E&P subsidiary — is unique among US E&Ps in sitting inside a regulated utility. Even ~7% annual production growth gradually makes the combined entity more of an E&P and less of a utility. National Fuel has mitigated Seneca’s rising weight by bulking up its utility business, announcing the $2.6 billion acquisition of CenterPoint’s Ohio gas utility earlier this year, but future acquisitions would be necessary to keep pace with Seneca’s growth.

The cleaner fix is the one the company is now exploring: separating the Seneca assets. As Appalachia’s best inventory runs out and the basin returns to growth, Seneca has three credible paths:

• acquiring offset acreage to efficiently scale up as an independent E&P,

• being acquired by a large operator, to help it capture more of the basin’s future growth, or

• merging with a mid-sized operator, most likely one facing imminent production declines.

I see the last scenario as most likely, and that Ascent is most in need of such a merger, because its production is set to decline before its takeaway capacity rolls off.

Appalachia’s shift back to growth makes inventory more valuable

Cabot, Chesapeake, and Southwestern pioneered northeast Pennsylvania Marcellus development, ramping up activity and production sharply from 2008-13. Dwindling inventory eventually pushed both Cabot and Southwestern to merge into larger companies, and their assets now sit inside Devon and Expand, respectively. In recent years, northeast Pennsylvania production has remained steady in the ~11 Bcfd range, with declines on the legacy Cabot and Southwestern assets offset by growth from Repsol, Seneca, and others.

Figure 1 | Northeast Pennsylvania production by operator

Because northeast Pennsylvania sits inside a nested constraint within Appalachia, production is likely to remain at this ~11 Bcfd level until new intra-Northeast projects are developed, although individual northeast Pennsylvania operators can grow as legacy assets decline. Rising Northeast demand and new takeaway projects will also narrow Appalachian basis discounts, incentivizing production growth and strengthening the case for new intra-Northeast capacity.

But the quality of remaining Appalachian inventory has deteriorated since the basin was last in growth mode, so a different set of assets and operators will lead the next phase of growth. I expect operators to maintain production on a given asset2 as long as they can, and for the next-best assets — defined by both productivity and remaining inventory life — to ramp up once mature ones go into decline.

Good-but-not-great acreage gains value every year, as great inventory runs out, so companies like Seneca3 are well-positioned for long-term growth.

Figure 2 | Appalachian dry gas production forecast by sub-basin and operator

Read more

Morrisey’s announcement completes the puzzle posed by TC Energy’s surprise May sanctioning of the Appalachia Supply Project on its Columbia Gas (TCO) system, backed by a 20-year, ~800 MMcfd contract with an unspecified investment-grade utility. The project had not previously been announced, let alone held an open season.

Now, project details are crystallizing. TCO’s August 3-7 open season included a stylized map illustrating TCO Pool receipts and a delivery point in Grant County. The project targets an in-service date of November 1, 2030. As Appalachian gas-fired generation ramps up, demand at Mt. Storm could push TCO Pool back in line with Midwest hubs rather than Appalachian points.

Why does Mt. Storm need 800 MMcfd of pipeline capacity?

A 2.6 GW power plant only burns ~400 MMcfd at 95% utilization, and Dominion’s Mt. Storm permit application calls for 2033-34 in-service dates for each of the plant’s two units, considerably later than ASP’s 2030 start.

Read more

The project’s two shippers, Duke and Enbridge Gas North Carolina (formerly PSNC), are short FT relative to demand. EGNC should want the capacity as soon as it is available: MVP Southgate can immediately displace its Transco receipts and diversify its supply mix. But MVP Southgate does not reach far enough south to serve most existing Duke plants, and Duke’s Person County combined-cycle project is not due online until 2028.

MVP Southgate will free up Transco capacity to move gas further south. But with Southeast gas burn limited by combined-cycle capacity, that gas will mostly back off eastbound flows through Transco’s Station 85.

How do Duke and EGNC source gas currently?

Duke, EGNC, and Piedmont — a Duke subsidiary since 2016 and the largest North Carolina LDC — overwhelmingly source gas from Transco, directly or indirectly. All three have many physical Transco meters, but each nominates to a single virtual meter. Duke holds Transco capacity, but that gas reaches its plants via either Piedmont’s or EGNC’s distribution system, and sometimes also through Cardinal Pipeline.1

Figure 1 | North Carolina existing gas infrastructure and gas-fired power plants

Only two other interstate gas systems can deliver into North Carolina, but neither moves much. East Tennessee Natural Gas moves gas out of the state via its Transco interconnect, and Columbia Gas delivers ~10 MMcfd into Piedmont in northeastern North Carolina.

What does MVP Southgate do?

After years of legal battles, the ~2 Bcfd Mountain Valley Pipeline (light blue line in Figure 2) entered service in 2024. MVP Southgate (red line in Figure 2) was originally proposed as a 375 BBtu/d downstream extension of MVP, with EGNC (then PSNC) as its lone 300 BBtu/d shipper. When MVP stalled, Equitrans2 paused MVP Southgate development.

Atlantic Coast Pipeline’s cancellation revived MVP Southgate’s prospects, and MVP’s 2024 in-service made it viable. Without ACP, Duke contracted for 250 BBtu/d on MVP Southgate, and EQT upsized the pipeline to 550 BBtu/d. In Duke’s filings with the North Carolina Utilities Commission, the company justifies its capacity commitment as needed to support its existing fleet and backstop future coal retirements and combined-cycle buildout — 6.8 GW of new CCs by 2033, per Duke’s latest integrated resource plan.

In the near term, though, Duke has only one new CC — 1.3 GW at Person County (red diamond in Figure 2) — under construction, with a 2028 in-service date. EGNC is building a new distribution pipeline, T-15 (yellow dotted line), from MVP Southgate to the plant, due in service in 2027.

Figure 2 | Current and under construction North Carolina gas and power infrastructure

Read more

Antero’s Columbia Gas/Gulf path features its lowest rates to LNG-connected markets but also the earliest expiration dates. The fate of Antero’s Columbia Gas/Gulf contracts expiring in October 2027 and June 2028 will reveal both whether the company is optimizing for near-term savings or long-term cost structure and how willing Antero’s midstream counterparties are to negotiate.

Antero’s three paths to the Gulf Coast

On its earnings call, Antero CFO Brendan Krueger said that over the next five years, the company would move from selling two-thirds of its gas in the “LNG fairway” to about half. These disclosures on volume changes and dollar savings, coupled with the latest index of customers data and FERC filings on negotiated rates, narrow down which specific contracts Antero intends not to renew.

Currently, the company holds ~1.9 Bcfd of capacity to the “LNG fairway” via three stacked routes. Most of the capacity — all but the TGP 100 leg deliveries — terminates in premium coastal markets, as shown in Figure 1.

Figure 1 | Antero’s LNG-connected capacity

Read more

But since then, Antero has made three announcements that begin to address this mismatch:

1. A new intra-basin East Side Express pipeline project to move dry-gas volumes from legacy HG acreage.

2. The $315 million acquisition of 125 MMcfed of production and 15 locations in the dry-gas window near East Side Express.

3. A detailed one-off company presentation outlining how declining to renew expiring pipeline contracts would improve the company’s margins and free cash flow.

As I’ll show, East Side Express is the first step in a re-plumbing that will eventually let dry gas volumes feed pipeline capacity originating at Sherwood/Smithburg. Then next week, I’ll break down how the company’s latest disclosures confirm which capacity Antero will let go.

How East Side Express fits

This earnings season, Antero Resources and Antero Midstream announced a 1.5-2.0 Bcfd intra-basin gas transmission project, East Side Express. Antero Midstream said the project would cost ~$100 million annually for the next three years, implying a 2H28 or 1H29 in-service date. A contract with Antero Resources underwrites the project.

The companies did not publish a map but said the project would run 30 miles east-west across Antero’s dry-gas acreage position and feature seven interconnects with gas transmission systems. HG’s legacy gathering system (coral line in Figure 1) also runs east-west, delivering volumes into EGTS (teal line), SGG (purple line), and Equitrans (brown line).

Figure 1 | Antero FT and likely East Side Express route

Read more

Accounting for rising in-basin demand and exports to Mexico, filling that capacity would require 18 Bcfd of dry gas production growth, more than the ~17 Bcfd Appalachia managed in its fastest five-year stretch, from 2011-16. Every gas E&P fears what Permian gas growth could mean for Henry Hub prices. But a record pace of Permian gas production growth would require operators to drill more gas-weighted targets, which Permian E&Ps demurred on this earnings season.

Rather than an expectation that Permian gas production will match the pace of post-FID pipeline buildout, I read this project’s FID as revealing skepticism about Desert Southwest’s effective capacity and its ownership structure as suggesting a change in how Permian E&Ps view the risk-reward of gas takeaway commitments.

Who are the likely shippers?

Read more

Over the first half of this year, the company ran more rigs and as many frac crews as any other operator in the play. With seasonal spreads again as wide as last year’s, Apex’s growth is likely to surge again in early winter.

How did Apex’s production grow last winter?

After the acquisition, Apex immediately ramped up drilling activity, from two rigs in winter 2024-25 to seven rigs by June 2025. The company ran just one frac crew in summer 2025, building DUCs and keeping production flat at ~600 MMcfd. But by December, Apex’s Haynesville volumes surged to ~1.4 Bcfd, deploying several tactics simultaneously to boost production alongside rising seasonal prices.

Figure 1 | Apex production and upstream activity

Deferred TILs and excess DUCs

Although market lore holds that Apex accumulated a large backlog of excess DUCs and deferred TILs, in fact deferred TILs and excess DUCs explain less than half — ~300 MMcfd — of the company’s late-2025 production surge.

Of the 13 pads Apex turned in line between September 2025 and February 2026, only five had completed wells sitting idle for more than 21 days after the last well was completed (purple boxes in Figure 2). These TIL delays were generally modest: less than two months for all but one. And only one of these pads (yellow shading) also saw an abnormally long gap between the rig’s departure and the crew’s arrival.

Figure 2 | Apex spuds, completions, and TILs by pad

Aggressive chokes

An equally large driver was Apex’s aggressive choking of wells that came online in 2H25. Across multiple vintages, those wells’ production peaked in December. Expand’s wells, by contrast, peaked in their second month online.

Figure 3 | Apex and Expand type curves1 by vintage

Read more

Reduced coal-gas competition, capital-discipline-era supply inelasticity, 15 years of storage underinvestment, and rising power and LNG demand combine to drive continued increases in gas price volatility. And this rising volatility creates outsize profits for companies positioned on the right side of it, which Expand now clearly aims to be. In conjunction with the deal, Wichterich announced an increase in target incremental marketing profits from $500 million to $750 million annually.

But realizing that value depends on future strategic decisions around contracting and risk management, not the book of deals that Twin Eagle currently holds. If successful, Expand will differentiate itself from smaller gas E&P peers, none of whom has the scale to execute a similar playbook. If not, the volatility-induced profits will remain concentrated among the funds and consumers (and EQT) that are willing to shoulder the associated risk.

Limited immediate FT integration

Producers typically contract for FT to backstop needed takeaway capacity from constrained basins, whereas marketers and traders view pipeline capacity as an option and make contracting decisions accordingly. Expand’s interstate FT — from both legacy Southwestern and legacy Chesapeake — moves gas from Marcellus and Haynesville producing areas to markets in the Northeast, Midwest, and Gulf Coast. Twin Eagle’s capacity is for smaller quantities (thinner lines in Figure 1) and is more geographically diffuse.

Immediate integration opportunities are likely limited to the Midwest. Some of Expand’s Rover and REX volumes might be able to flow into Twin Eagle’s capacity, although it’s uncertain because the receipt points are out of path. If these volumes are successfully scheduled further downstream, Expand’s realized prices would improve during peak winter heating months.

Figure 1 | Expand and Twin Eagle US interstate FT (2026Q3)

But in that scenario, Expand’s improved realized prices cannibalize what would otherwise have been Twin Eagle marketing profits. Pre-deal, Twin Eagle could buy either Expand’s or another producer’s gas and capture the value of its capacity. To the extent that these volumes came from another producer, the pro forma entity could unlock perhaps a couple cents per MMBtu of marketing margin. But the bigger incremental benefits to the combined organization depend on future FT or other strategic marketing decisions.

Strategic rationale

The strategic rationale, therefore, is more about the Twin Eagle team and its capabilities than the specific capacity agreements. That capacity position itself is likely to change: Twin Eagle’s FT is not just for smaller capacities but also for much shorter tenors, as shown in Figure 2.

Figure 2 | Expand and Twin Eagle US interstate FT history

Read more

I was never anywhere near that bullish; my own outlook called for Henry Hub prices reaching the mid-$4/MMBtu range by 2030. But an analyst’s job is to keep re-evaluating, and the evidence since then suggests I made the same error Chronometer did — underestimating elasticity — albeit on a much smaller scale. Given recent Haynesville activity levels, I now expect Henry Hub prices to rise to the low-mid $4s by 2030.

What is the purpose of this whitepaper (and media tour)?

After 18 years in this business, I can tell you that written research — even paid written research — is almost always a loss leader for something else, typically banking or software. Here it’s likely marketing, for raising additional capital and/or recruiting partners and analysts. In both cases, your incentive is to gain attention by positing interesting, thought-provoking ideas. In other words, to identify a real problem and sell it hard. If Henry Hub prices briefly go to $8/MMBtu and then settle at $4.50, Chronometer still looks great, even though that’s nowhere near a structural doubling or tripling.

What’s wrong with Chronometer’s analysis?

Energy FinTwit immediately and comprehensively documented the flaws1 in Chronometer’s analysis — even Kimmeridge’s Ben Dell weighed in — so I won’t rehash those here. (Kimmeridge itself is a good example of a firm that produces serious, deep research as a loss leader.)

At a high level, though, this analysis falls down by consistently understating elasticity — in gas supply, LNG exports, gas-fired generation, and gas pipeline development — and thereby overestimating the impact of real market changes (Figure 1). And then all of these overstated partial equilibria compound into an even more ridiculous general equilibrium: that Henry Hub prices will double or triple structurally.

Figure 1 | Chronometer’s claims and my views on them

To be fair, Chronometer is right that complacency about long-term gas prices itself increases the odds of a spike, that Silicon Valley is taking gas-price risk lightly, and that rising gas and power prices are likely to become more politically salient. But it’s a long way from any of those to a structural doubling, which none of Chronometer’s analysis supports.

Why I was already getting less bullish

But even before Chronometer gave all of us gas bulls a worse name than we already had,2 market fundamentals had me reevaluating that mid-$4/MMBtu view. Starting in mid-2025, weather-adjusted gas-fired generation flattened, while production growth accelerated despite modest prices. The former is likely transient, but I am revising my thinking on prices based on the latter — and I’ll take each in turn.

Gas-fired generation: likely to accelerate over the next year

So far this year, gas-fired generation is roughly flat year-on-year, after normalizing for gas prices and fossil generation levels (Figure 2). The slowdown in coal retirements has ended a long-term tailwind for gas burn.

Figure 2 | Year-over-year change in gas-fired generation

But while gas’s competitiveness with coal (holding gas prices constant) is unlikely to shift much in the short or medium term, fundamentals for fossil generation look stronger. Data-center development is diffuse and opaque, so it’s hard to know exactly why total generation levels slowed in the first half of this year after accelerating sharply in 2023-25. But hyperscalers (Microsoft and Google) and frontier labs (OpenAI and Anthropic) all point to a common explanation: demand for AI tools outstrips computing capacity.

I expect that weakness in load growth from mid-2025 through 1Q26 reflected constraints on data-center capacity, and that load growth is likely to accelerate as new capacity comes online. Some evidence of this is already visible in US power data: weather-normalized generation since April has grown at a ~1.5% annualized rate, up from ~0.4% over the prior three quarters.

Figure 3 | Weather-normalized run-rate US electricity generation (30-day MA)

Add in a decelerating solar build following the tax-credit phase-out, and I’m not worried about medium-term fundamentals for gas-fired generation, even though gas burn this year has been weak.

Haynesville activity: stronger than I expected

Permian gas production growth filling available takeaway capacity this summer is unsurprising, and still consistent with a bullish long-term view. But ~2.4 Bcfd of Haynesville gas production growth since September 2025 is more than I expected, as was the high-50s rig count in February-March.

However, the implications of these higher drilling levels are less severe than they seem at first glance. The currently active 41 rigs don’t necessarily mean that operators see that activity level as optimal relative to prompt-month prices of $2.75; rather, operators made those drilling decisions last fall, when the 2026 strip traded at ~$4. But the 52 rigs active on average so far this year do mean that operators saw $4/MMBtu as a sufficient price to ramp up Haynesville drilling and production, which is a lower threshold than I expected (and lower than major operators guided).

Figure 4 | Haynesville activity and year-ahead Henry Hub prices

Implications for long-term gas prices

Since I wrote about my bullish view last fall, the 2027-30 Henry Hub strip has sold off from ~$3.85/MMBtu to $3.60. Given this year’s stronger-than-expected Haynesville activity, I’d fade my medium-term outlook by a similar amount, to the low-mid $4s.

In 2024-25, operators ran ~40 rigs following ~$3.25-3.50/MMBtu planning prices, so I expect a similar 2027 Haynesville activity level with the 2027 strip at ~$3.40/MMBtu. With Haynesville drilling already down to an average of 45 rigs in June, this seems likely — but if it’s not, I’ll need to revise further.

And this trend is likely to continue. Most public gas E&Ps operate in Appalachia, where production growth will likely remain slow, limited by modest incremental takeaway capacity and in-basin demand growth. The largest Permian operators are public, but these companies often do not break out dry-gas production guidance separately from their rich-gas or Boe figures. Haynesville acreage, meanwhile, has consolidated, but mostly into the hands of Japanese conglomerates that don’t give detailed Lower-48 gas production guidance, either.

As a result, accurately anticipating near-term Lower-48 gas production inflection points will increasingly depend on synthesizing satellite and state-reported data.

First, estimating current production

Before we can synthesize activity levels into an estimate of run-rate growth, we need to understand current well-by-well production. To do that, I developed an algorithm that estimates likely unreported spuds and completions1 in a multi-step process:

1. Estimating a rig’s (or frac crew’s) capacity for additional spuds (or completions), beyond what has been reported, based on how long it was observed on a pad and typical drilling (or completion) times in that play

2. Netting out reported spuds or completions on the pad, and

3. Constraining this by the availability of candidate wells on that pad that have been permitted (or drilled)

Which companies drove recent growth?

Read more

But as I’ll show, New England LDCs are well-served with existing capacity, while the unit economics of space-heating conversions only pencil out with a subsidy from existing gas customers. Power generators, meanwhile, only need additional gas on a couple dozen days per winter and don’t have a rate base to pass through the costs of additional capacity. In the long run, the region’s consumers would be better off backstopping more LNG imports into Canaport LNG1 than paying for additional 365-day pipeline capacity.

Who holds capacity into New England?

For the first time since 2017, a new pipeline expansion into New England is on offer: Algonquin’s Project Beacon recently launched an open season. Algonquin is one of five pipelines bringing gas into New England on three major routes:

• Iroquois and Portland Natural Gas connect to the TC Mainline at the border and bring gas south into New England.

• Intermittently, Maritimes & Northeast brings gas south from Canaport.

• TGP and Algonquin bring gas northeast from Appalachia.

Repsol holds most of the capacity on M&NP, and LDCs hold ~80% of the contracted capacity on the other four.

Figure 1 | Pipeline capacity into New England

Even though New England power generators hold just ~200 MMcfd of capacity into the region, they burn ~1.3 Bcfd on average. Because utilities contract for peak-day needs, their capacity is under-utilized most of the year, meaning that most days it is available for merchant power generators to free ride.

Figure 2 | New England gas demand by sector

Where and why does New England burn oil?

Read more

When I launched this project late last year, I posed three questions that I was hoping to answer:

• Whether research alone is a viable business

• And if it is, whether subscribers would want something broader or deeper, whether they’d want more attention on production or power or LNG

• Which of my potential product ideas could have a market

This week, I’ll cover what I’ve learned since then and ask for your feedback, both generally and on a few specific questions. I've put together a short reader survey, and I’d really value your free-response feedback. I’ll comp a month’s subscription for responses I find especially helpful.

Is research alone a viable business?

Eight months into this endeavor, and four months into paid subscriptions, Measured Depth has:

• Three enterprise subscriptions. I’m very grateful to them, all of whom I knew as clients prior to launching this endeavor.

• An additional ~25 individual subscribers, each from a different company.1

• ~800 more subscribers to the free version

Relative to what I expected at the outset, Measured Depth has more free subscribers but fewer paid subscribers. And the paid subscription trends also differed from what I anticipated; I thought perhaps 5-10% of free subscribers would pay initially, but that hardly anyone would upgrade afterward. Instead, only about 10 people upgraded to a paid subscription in March, but about half that many have signed up each month thereafter. Realistically, I probably need to double the paid subscription levels for the weekly publication to be worth it, but I’m optimistic that I’ll reach that level within the next 6-12 months.

But perhaps that confidence is misplaced! To that end, I’d like to understand how each of you is thinking about subscribing.

What do subscribers want?

From page views and subscription rates, I have a sense of which topics resonate most strongly with subscribers. But honestly, the readership statistics have been more similar across posts than I expected, suggesting that there aren’t universally “most interesting” topics, but rather a range, each of which resonates with a different subset of Measured Depth subscribers.

And while I’m proud of this range — several angles around LNG exports, coal-gas competition and coal retirements, analytical methodology, Permian production, Appalachia takeaway and new projects, load growth and more — I’m less sure whether it’s the right direction for this publication, relative to something more focused.

I’m not ever going to have a daily fundamentals product like RBN or Criterion or Genscape, or comprehensive coverage of the energy value chain like WoodMac or CERA. I do think I’m offering differentiated depth and quality of analysis and providing unique strategic and tactical insight. And as I build more foundational tools, I’ll have more time to incrementally add functionality and improve the offering. The question I can’t answer from readership statistics, though, is what latent interest subscribers have, and what direction you’d like to see those product improvements take.

Which product ideas could have a market?

About a year ago, my cousin — who works for a small private Permian E&P — came to Houston for a board meeting and wanted to talk about two things: the prospects for higher Waha prices and AI. I told him I more or less wasn’t using AI at all, and he was stunned. At the time, I had a team of 10 or so people, so I spent a lot of time in meetings and wasn’t writing much code myself.

When I wrote in November about launching this venture, I had seen enough to know that AI would make it easier, but in retrospect, I had no idea how much easier. As a parent, I often wonder how AI will reshape jobs by the time my kids are interviewing.2 As an analyst, though, I feel like these tools were made for me: someone with subject-matter expertise who thinks algorithmically, but whose ideas for new models far outpaced her time (or capability3) to build them.

I’ve never set up a database before, or scraped data at scale, or digitized pipelines, but I’ve used these tools enough to know what I’m looking for, and what to ask Claude for direction on. I know it’s trite, but truly, what a time this is to be alive. I’ve been able to build products much faster than I expected, and am excited to share some of those with you this fall.

In the weekly posts, you’ve seen some graphics illustrating pipeline capacity by location, which I can now do algorithmically with location, shapefile, and index of customers data. I want to add more functionality to the application, but I have enough pipelines (Figure 1) digitized to be useful.

Figure 1 | Lower-48 contracted interstate capacity

Then, over the next few months, I want to use this contracted capacity as the backbone of a gas flow model. Maintaining infrastructure data is a major challenge in the traditional approach to gas flow forecasting, so my hope is to streamline this process using the IOC and nodal structures. And then there’s my white whale — a supply model — which I’ve been developing with a partner and hope to begin scaling in the coming months.

If you or your firm would be interested in hearing more about any of these products, please get in touch so I can set up a meeting. Of course, I’d love to hear your thoughts on my survey questions, too. In the meantime, happy belated 4th, and thanks for reading!

Subscribe now

But now, three major Permian takeaway pipelines are under construction to Gulf Coast markets, and Energy Transfer is also developing a new-pipeline-sized expansion of Transwestern. These four projects add ~10.5 Bcfd of capacity over the next four years, and the call on Permian gas from Mexico and within the basin is likely to rise ~1 Bcfd over the same period.

That ~11.5 Bcfd of new market access would accommodate both the currently curtailed ~1.2 Bcfd and 2.5-3.0 Bcfd of growth each year thereafter. Public Permian midstream operators agree that the ongoing Permian gas takeaway buildout will un-constrain the basin, but the market is increasingly worried that Permian gas supplies will fill this takeaway capacity, weighing on Henry Hub prices.

But as I’ll show, accelerating base declines mean that even a consistent pace of Permian activity — itself unlikely — won’t fill forthcoming takeaway capacity unless the pace of single-well gas productivity accelerates sharply.

Declines from existing wells

Because the Permian is now such a mature play, forecasting gas production from existing wells is much more straightforward than forecasting either future activity levels or the productivity of future wells. However, traditional Arps-based approaches to building type curves systematically under-forecast gas volumes in the middle of a well’s life, when the rate of gas decline slows dramatically relative to that for oil and gas-oil ratios rise rapidly.

Instead, I use an Arps-based forecast for oil volumes and forecast gas volumes based on the typical trajectory of gas-oil ratios through the well’s life. This approach adds ~800 MMcfd to medium-term gas production compared with a typical Arps-based approach. However, even this approach points to continued acceleration in base declines. Rising production levels mean that even if the pace of PDP declines remains consistent, the level of base declines increases.

Figure 1 | Permian PDP gas volumes and first-year declines from wells TIL by 2025

Read more

In its April fundamentals presentation, Enterprise pegged curtailments at 2.0-2.5 Bcfd, and a growing subset of analysts now expect the ~4.6 Bcfd of Permian capacity due online over the next year to fill immediately. Whether that happens depends on two factors:

1. How much production is curtailed currently

2. How much production is likely to grow from this point forward

Given the complexity of this topic, I’ll cover the latter in a follow-up post. The former is more challenging in the Permian than it would be in any other basin for two reasons: pipeline scrapes have more noise than signal for Permian production, and while most wells report within 2-3 months of when they begin producing, the long tail of Texas well data trickles in over more than a year.

To address these issues, I’ll triangulate between frac-crew tracking, pricing data, and meta-analysis of state reporting to estimate what Permian dry gas production “should” be currently and, therefore, how much volume is likely curtailed. As I’ll show, Enterprise’s curtailment estimate was likely conservative as of April, due to a surge in March TILs. Since then, easing completion activity and the GCX expansion coming online likely pushed curtailments down to ~1.2 Bcfd.

Volumes (including wellhead shut-ins) from reported wells

As of yesterday, state regulatory agencies reported ~28 Bcfd of Permian wellhead gas production for March 2026, which translates to ~22 Bcfd of dry gas.1 Although most Permian production has oil-weighted revenue streams and therefore little incentive to shut in when Waha prices are negative, some wells are more gas-weighted and respond to gas price signals. These shut-ins mean that true wellhead production potential is understated even in reported data. I analyze cohort-specific gas-oil ratios to identify where reported wellhead gas volumes fell short of what we would expect given oil production levels, indicating price-driven wellhead gas shut-ins.2

Altogether, these curtailments add ~400 MMcfd of missing wellhead production in state-reported data for February and March. Because Arps-based type curves tend to understate gas volumes in oil plays, I use a combination of oil type curves and the typical trajectory of gas-oil ratios over a well’s life3 to estimate production from these wells since March. This translates to 22.4 Bcfd in June, which in-basin demand and takeaway capacity would easily accommodate. But this estimate is incomplete because it does not account for wells that are already producing but have not yet been reported by state regulatory agencies.

Figure 1 | Permian gas production and takeaway capacity

Reported completions not yet reported producing

These wells constitute the biggest source of uncertainty in estimating current Permian gas production because we don’t know how many there are. The last two months are instructive but by no means unique. Over that period, most of the Permian wells newly disclosed by the RRC (medium and light gray bars in Figure 2) started producing in March or April. But an additional four of them started producing in July 2024, meaning it takes more than a year to have a complete picture of Permian wellhead volumes.

Read more