Associated gas, healthy storage and an LNG buildout are pulling gas prices away from oil, and contracts written for one price environment are being tested in another.

WTI crude has recently traded around $90 a barrel. Henry Hub natural gas futures settled at $3.03 per million Btu on September 30. In the Permian, many operators pull both out of the same wells, and right now the economics of the two look very different.

The forecasts show the same gap. The U.S. Energy Information Administrationā€™s September Short-Term Energy Outlook has Henry Hub averaging $3.43 per million Btu in 2026 and $3.28 in 2027. That second number is worth a second look. In January, EIA expected 2027 gas to average $4.59, because it expected demand growth, driven largely by LNG export plants, to outpace supply growth. The agency has since cut that forecast by $1.31.

Gas Comes Up With the Oil

Most Permian gas is associated gas, meaning it comes out of wells drilled primarily for oil. EIA describes it the same way: most Permian natural gas production is associated with crude-oil production. And the gas-to-oil ratio keeps climbing. The Permian averaged nearly 4,200 cubic feet of gas per barrel of oil in 2025, 15 percent higher than in 2021. As long as oil prices justify drilling, associated gas can keep growing even when local gas prices are weak or pipeline space is scarce.

Nationally, dry gas production is close to record levels. EIA expects it to average 111.7 billion cubic feet a day in 2026 and 115.9 Bcf a day in 2027.

The Permian has already shown what happens when that gas has nowhere to go. As of mid-April, prices at the Waha hub in West Texas had stayed negative for a record 47 consecutive days because limited pipeline capacity trapped associated gas in the basin. EIA notes that Energy Transferā€™s Hugh Brinson pipeline, which starts in the Permian, began moving gas into interstate markets in June, earlier than the agency expected.

Storage Is Comfortable

EIA reported working gas in storage at 3,415 billion cubic feet for the week ending September 25. Thatā€™s 2.4 percent above the five-year average and nearly 4 percent below a year ago.

EIAā€™s outlook has inventories reaching 3,969 Bcf on October 31, 5 percent above the five-year average, when the injection season ends. EIA also says the Permian and Haynesville regions together account for more than 70 percent of the production growth in its forecast.

January looked different. EIA then expected supply growth to run slightly ahead of demand growth in 2026 and fall behind in 2027 as LNG feedgas demand picked up, and it forecast Henry Hub prices rising about 33 percent in 2027. The September outlook has 2027 averaging slightly below 2026.

LNG Is Pulling the Other Way

LNG demand keeps growing. EIA expects U.S. gross LNG exports to average about 17 Bcf per day in 2026 and about 19 Bcf per day in 2027. In its January outlook, EIA attributed export growth to the ramp-up of three facilities: Plaquemines LNG, Corpus Christi Stage 3 and Golden Pass LNG.

Demand is still growing, and EIA still cut its price forecast. Anyone who leaned on the January outlook when pricing a long-term deal is now working from a very different number.

Where the Contracts Start to Matter

Most of the trouble starts with contract language nobody read closely. Hereā€™s where Iā€™d look.

Start with the index. Gas sales and gathering agreements can price off Henry Hub, a regional hub like Waha, or a formula that blends several, and those prices can move far apart, as Waha showed. Some agreements say nothing about what happens when prices go negative or when the sales price drops below what it costs to gather, treat and transport the gas. If yours is silent, the parties may end up arguing over how the pricing, delivery and payment provisions split the loss.

Associated gas adds a wrinkle. Because the gas follows the oil, a producerā€™s drilling schedule can drive its future gas volumes. That matters for acreage dedications, minimum volume commitments and deficiency payments. A producer with a strong oil program may push more gas into a weak market than it expected. One that slows drilling may struggle to meet commitments it made when its plans looked different. The definitions of committed gas, delivery points, minimum volumes and shortfalls can decide who pays.

Next, curtailment and shut-in rights. With a conventional gas well, an operator can consider shutting in production instead of selling at a loss. Associated gas is harder, because curtailing the gas can mean curtailing the oil well that produces it. The operator has to weigh the gas contract, any acreage dedication, the oil and gas lease, and operational or regulatory limits on what it can do with the gas.

Some leases let the lessee hold the lease through shut-in royalty payments if certain conditions are met, but the trigger language, payment mechanics and time limits vary widely. Donā€™t assume force majeure will rescue you. Under Texas law, the analysis starts with the language the parties negotiated, and economic hardship or a bad market, standing alone, generally isnā€™t enough to excuse performance unless the clause covers that situation.

Service companies face their own version of this. Baker Hughes counted 598 U.S. rigs working on October 2, 49 more than a year earlier, so drilling activity is up from this time last year. The mix matters, though: 456 were oil-directed, 133 were gas-directed and nine were miscellaneous. An oil-focused customer and a gas-focused customer can be in very different positions. If a contract ties you to a gas-weighted customer, review the minimum commitments, standby charges, early-termination rights and payment protections.

Finally, long-term supply agreements. LNG contracts can be priced off Henry Hub, tied to oil, or built on a hybrid. When gas and crude prices pull apart, that choice matters a great deal. Check the price-adjustment provisions, volume tolerances and the remedies that kick in when volumes change. A deal evaluated when EIA expected $4.59 gas in 2027 may look different against the agencyā€™s current $3.28 forecast.

What to Watch

EIA releases its next Short-Term Energy Outlook on October 6, and the weekly storage report follows on October 8. Both will show whether inventories are tracking the current forecast and whether EIA changes its 2027 price projection again.

The numbers will keep moving. The contracts wonā€™t, unless somebody changes them. Pull the ones that touch gas and read the pricing, curtailment and termination provisions before winter gets here.

Record nominal pump prices and a shrinking inventory cushion are colliding with fuel contracts that were priced for a calmer market.

The EIA now expects distillate inventories, diesel and heating oil, to drop below 100 million barrels this month and stay below the five-year low through most of 2027. That’s not a normal seasonal dip. Stocks fell below the five-year range back in April, and by the week ending August 28, they sat at 104.2 million barrels, 14 percent under the five-year average for that time of year.

Pump prices have made the point more bluntly than any inventory chart could. AAA put the national average diesel price at $5.85 a gallon on September 4, topping the old nominal record of $5.8159 from June 2022. It kept climbing from there. EIA’s own survey had the national average at $6.529 a gallon by September 21. If your fleet, your drilling rigs, or your frac crews run on diesel, this isn’t a headline you can skip past. It’s next month’s fuel bill.

Why the Cushion Disappeared

It’s not for lack of trying at the refinery. EIA data show refinery utilization at 98 percent for the week ending August 28, about as high as it gets, with distillate production running around 5.1 million barrels a day. Normally that pace rebuilds inventories fast. This year it hasn’t been enough.

Part of the reason is exports. U.S. net exports of distillate have run at or near five-year highs almost every month since February, because refiners overseas haven’t been able to match last year’s output. Every barrel that leaves the Gulf Coast for a buyer abroad is a barrel that never shows up in our domestic stockpile, no matter how hard U.S. refiners are running.

The rest of the story is overseas, and it’s ugly. Tanker traffic through the Strait of Hormuz is still disrupted, alternative routes are seeing less use, and drone attacks on Saudi Arabia’s East-West Pipeline this month forced a precautionary shutdown of a line that can move up to 7 million barrels a day to Red Sea export terminals. Add strikes hitting Russian refining capacity and you get a global distillate market that’s tighter than it’s been in years. None of that happened here, but diesel doesn’t price itself on a national basis.

The Crack Spread Is Doing the Heavy Lifting

Crude oil prices only tell part of the story. The rest is the crack spread, the gap between the price of diesel and the cost of the crude used to make it. That spread hit an all-time high of $102.20 a barrel on August 17, then pushed past $106 in early September.

In plain terms, the crack spread is what refiners are earning for turning crude into diesel rather than something else. When it widens, wholesale diesel prices rise relative to crude, and once you layer in taxes, distribution costs, and the usual lag before that hits the pump, fleet operators feel it directly.

EIA breaks diesel pricing down the same way: crude cost, refining margin, distribution and marketing, and taxes. When distillate stocks are this tight and diesel commands a premium over other refined products, the refining-margin piece of that equation widens on its own, regardless of what crude is doing.

The Forecast Is Already Behind the Market

EIA’s September outlook has Brent crude averaging around $90 a barrel through the back half of 2026, easing to roughly $74 in 2027. The agency also expects distillate inventories to stay unusually tight thanks to strong exports and weaker global production.

Here’s the catch. EIA finished that outlook on September 3, before the national diesel average broke $6 a gallon. By September 21 it was at $6.529. The forecast wasn’t wrong when it was written. The market just moved faster than the ink dried.

EIA has flagged the risk itself: if Middle East disruptions run longer than expected, crack spreads could stay wider than the current outlook assumes. The next weekly petroleum report and the October 6 outlook will tell us whether the agency is catching up to where prices already are.

Where This Shows Up in Your Contracts

For oilfield services companies, this is a cost problem before it’s anything else. Rigs, frac fleets, workover units, and the trucks hauling sand, water, and crews all run on diesel, and a lot of that fuel gets bought under contracts signed when prices were a lot lower.

Start with your fuel surcharge or price-adjustment clause. Some track a published benchmark, like EIA’s weekly diesel survey, and adjust automatically. Others rely on a fixed band or a renegotiation window that was never built for a market moving this fast. A surcharge that resets monthly off last month’s average will always lag when prices jump this quickly, and that lag is where one side loses margin and the other picks up a windfall it didn’t earn.

Then there’s force majeure. Under Texas law, and most others, the contract language controls. Higher cost or economic hardship on its own generally doesn’t get you there. A pure price spike, without a qualifying provision or an actual physical inability to perform, usually won’t excuse performance no matter how dramatic the number looks. Read the clause the way opposing counsel will read it, not the way you’d like it to read.

Last point, and it’s the one people miss most: an escalation clause tied to an index only works if somebody is actually watching that index and pulling the trigger when it moves. A good clause nobody monitors is worth about as much as no clause at all.

What to Watch Next

Two things matter more than the headlines over the next few weeks. The weekly petroleum report will show whether distillate stocks actually cross below 100 million barrels as EIA expects, and the October 6 outlook will show whether the agency raises its price assumptions to match where the market has already gone.

Either way, if your fleet runs on diesel, now is the time to pull your fuel clauses and read them closely, not after the next report confirms what your invoices are already telling you.

On September 21, 2026, Governor Newsom signed into law seven data center bills, which primarily take effect on January 1, 2027. Three of the bills (SB 886, AB 2383, and SB 1168) form a single tariff regime administered by the California Public Utilities Commission (CPUC).

SB 886

California Technology Innovation and Ratepayer Protection Act

SB 886 adds a new Article 14.7 to the Public Utilities Code, providing that the CPUC must establish or update data center tariffs by January 1, 2028, covering interconnection, retail service, transmission, distribution, and generation. (SB 886, Ā§ 1; Pub. Util. Code Ā§ 945 et seq.) These tariffs must prevent stranded costs and cost shifts from transferring to nonparticipating customers. (Ā§ 945.2(b)(2).)

The bill also requires the CPUC to establish a transmission-level interconnection tariff (Ā§ 945.3(a)) that:

• Requires disclosure of duplicate interconnection applications in other utility territories or jurisdictions (Ā§ 945.3(a)(2)).
• Assigns cost responsibility for all transmission upgrades and usage to the data center under a CPUC methodology. Shared network upgrades will be included to the extent federal law permits (Ā§ 945.3(a)(3)).
• Limits refunds of upfront contributions to 75% of the utilityā€™s annual net transmission revenue from that customer, which is tied to actual load materialization (Ā§ 945.3(a)(4)).
• Imposes an early termination fee if the customer leaves within 10 years or fails to ramp load. The fee would not be less than the revenue gap over the 10-year term (Ā§ 945.3(a)(5)).

Pursuant to the new requirements, customers must certify compliance with the prevailing wage and skilled-and-trained workforce programā€™s labor standards (PRC Ā§Ā§ 25545.3.3(a)-(b), 25545.3.5(a)-(e)) whether or not the customer uses the related opt-in program (Ā§ 945.3(b)). Utilities must also publish maps showing where data centers can interconnect without major upgrades (Ā§ 945.3(c)). This interconnection tariff (Section 945.3), including the labor certification, applies only to data centers entering into new transmission-level interconnection agreements on or after January 1, 2027 (Ā§ 945.3(d)). The remainder of SB 886 is not limited as to the date. For data centers that are seeking service after January 1, 2027 but before the tariff is ultimately approved (Ā§ 945.9), utilities may seek CPUC approval of individual contracts.

AB 2383

Data Center T&D and Generation Tariffs

AB 2383 is a companion bill to SB 886, which adds certain defined terms, a transmission and distribution (T&D) tariff, a generation tariff, and community choice aggregator (CCA) and energy service provider (ESP) obligations (Pub. Util. Code Ā§Ā§ 945.1, 945.4, 945.5, 945.6, 945.8).

For investor-owned utilities (IOUs), the CPUC must require each IOU to file a T&D tariff (Ā§ 945.4) and a generation tariff (Ā§ 945.5(a)) that:

• Requires data centers to pay incremental generation costs caused by their load, through a mechanism lasting at least 10 years (Ā§ 945.5(a)(3)).
• Includes fees for failing to complete interconnection, early termination fees, upfront collateral or prepayment, and a minimum-payment floor based on projected load (Ā§ 945.5(a)(3)(A)-(C)).
• Lets data centers reduce their funding obligation by installing behind-the-meter zero-emission resources (Ā§ 945.5(a)(3)(D)).
• Requires reporting of planned onsite generation investments and bars compensation for onsite generation that shifts costs to others (Ā§ 945.5(a)(4)-(5)).

The generation-tariff threshold is set by the CPUC but cannot exceed 25 MW of peak demand (Ā§ 945.5(c)). The T&D tariff has no stated threshold (Ā§ 945.4).

CCAs and ESPs must adopt their own data center generation tariffs by January 1, 2028, with parallel requirements and the same 25 MW cap (Ā§ 945.6(a)-(b)).

The bill exempts certain facilities from the generation tariff, including publicly funded research, public safety, publicly funded national security (as identified by the CPUC), publicly owned, and utility facilities (Ā§ 945.5(d)). The T&D, CCA and ESP tariffs do not have corresponding exemption lists.

SB 1168

Data Center Rate Structures

SB 1168 directs the CPUC to assess rate structure options, but does not mandate any specific tariff (Pub. Util. Code Ā§ 901). The stated goal of the bill is to ensure data centers pay a reasonable share of T&D costs regardless of interconnection level, pay their proportionate share of load growth and procurement, and help ease cost pressure on residential ratepayers (including CARE and FERA customers) (Ā§ 901(a)-(c)). The bill cites that CAISO expects that energy use will grow by 2.3 GW by 2030 (in the context of data center demand), noting that in 2024 there were approximately 2,300 MW of data center capacity applications through PG&E alone (SB 1168, Ā§ 1(a)).

AB 1577

Data Center Reporting to the CEC and Local Agencies

AB 1577 provides certain reporting requirements and applies to facilities of 10 MW of capacity or more. Obligations apply to the owner or operator running the entire facility, not to individual tenants (Pub. Resources Code Ā§ 25302.10(a)(2)(B), (a)(3)).

The bill requires a one-time California Energy Commission (CEC) filing at the time of energization (and at any subsequent time thereā€™s a substantive change) that includes the name, addresses, operator contact, start date, total and IT floor area, peak capacity per the interconnection request, and onsite generation (Ā§ 25302.10(b)(1)).

An annual CEC filing is also required, which must be broken out by month and specify peak load, installed IT capacity, total kWh, PUE, demand flexibility participation, refrigerant types, cooling degree days, onsite generation, fuel burned by type, and energy storage details (Ā§ 25302.10(b)(2)).

The CEC publishes this data annually in anonymized, aggregated form (Ā§ 25302.10(d)) and must include a data center load assessment in the 2029 Integrated Energy Policy Report and in later biennial reports as the CEC determines (Ā§ 25302.10(c)).

Operators can identify information they claim is exempt from disclosure under existing law (although the bill creates no new exemptions). Operator submissions do not waive any privilege or trade secret protections, and customer-specific energy data cannot be released (Ā§ 25302.10(e); Gov. Code Ā§ 7927.303(a), (d)).

When applying for a discretionary local permit, operators must provide to the city or county their estimated annual kWh, onsite generation by type, and average and maximum sound levels (dBA) at the property line (Gov. Code Ā§ 65909.3(a)).

Similar to the exemptions under AB 2383 regarding generation tariffs, AB 1577 exempts certain public and utility facilities (also including cable and internet service providers) (Ā§ 25302.10(f); Gov. Code Ā§ 65909.3(d)).

AB 2469

Water Disclosures as a Condition of Local Approval

AB 2469 establishes mechanisms to facilitate water disclosures by providing that cities and counties cannot approve a discretionary or ministerial permit for a new data center, or an expansion that raises peak water use (Gov. Code Ā§ 65964.7(b)), unless the applicant provides:

• A water supply assessment under Water Code Part 2.10 (Ā§ 10910 et seq.) (Ā§ 65964.7(b)(1)(A)).
• A water scarcity plan with staged drought measures (required starting January 1, 2028) (Ā§ 65964.7(a), (b)(1)(B)).
• Projected water use and efficiency measures (Ā§ 65964.7(b)(1)(C)).
• Workforce disclosures, including composition, job classifications, duration, wage ranges and benefits, and the share of workers living in the county and in the state (Ā§ 65964.7(b)(1)(D)).

The applicant must assume the full cost of water conveyance, treatment, storage, or distribution upgrades, as determined by the water supplier (Ā§ 65964.7(b)(2)). However, simply satisfying these conditions alone does not entitle the applicant to approval (Ā§ 65964.7(b)(3)). Notably, this bill contains no definition of data center nor any express size threshold (Ā§ 65964.7).

AB 2619

Water Reporting Through Business Licenses

AB 2619 sets forth additional water reporting requirements and covers data centers of all sizes across three tiers:

• Type I Data Center (Hyperscalers >10,000 servers or >25 MW power consumption)
• Type II Data Center (2-25 MW power consumption)
• Type III Data Center (<2 MW power consumption)

(Bus. & Prof. Code Ā§Ā§ 16000.4(a)(1), 16100.4(a)(1)).

Before applying for an initial city or county business license, or an equivalent permit, the operator must provide its water supplier a good-faith estimate of its expected water use, source, and projected volumes for max day, max month, and average year, under penalty of perjury (Ā§Ā§ 16000.4(b)(1), 16100.4(b)(1)). The same estimates must appear on the license application itself (Ā§Ā§ 16000.4(b)(2), 16100.4(b)(2)).

At the time of renewal of the permit, the operator must report the prior yearā€™s total and direct water use by source (potable, nonpotable, recycled) and identify the cooling system type (Ā§Ā§ 16000.4(b)(3), 16100.4(b)(3)).

The bill also provides that urban water suppliers must now factor data center demand into their annual supply and demand assessments (Wat. Code Ā§ 10632(a)(2)(B)(i)).

SB 887

CEQA Treatment of Data Centers

SB 887 defines ā€œdata centerā€ for CEQA and confirms CEQA applies to data center entitlements (Pub. Resources Code Ā§Ā§ 21061.6, 21080.08). The bill bars categorical exemptions for data center projects, and exempts publicly funded research, public safety, national security, publicly owned, utility, and independent-college research facilities (Ā§ 21084(f)(1)-(2)).

Pursuant to the bill, the Governor may certify a data center as an environmental leadership development project (Ā§ 21180(b)(5)(A)) if the lead agency certifies that the project satisfies the following requirements:

• Prepays full interconnection costs (Ā§ 21180(b)(5)(A)(i)).
• Does not increase in-state fossil fuel consumption (Ā§ 21180(b)(5)(A)(ii)).
• Maintains zero-carbon storage of at least 4 hours at 100% of forecasted peak, used for demand response (Ā§ 21180(b)(5)(A)(iii)-(iv)).
• Relies on behind-the-meter zero-carbon generation to the maximum extent feasible (Ā§ 21180(b)(5)(A)(v)).
• Has an enforceable commitment to pay all grid investments, including new generation, with an early termination fee (Ā§ 21180(b)(5)(A)(vi)).
• Uses recycled water and water-efficient or waterless cooling (Ā§ 21180(b)(5)(A)(vii)).
• Runs on 100% zero-carbon hourly electricity within 5 years of operation, of which 75% must be newly built (Ā§ 21180(b)(5)(A)(viii)).
• Enters into a binding community benefits agreement with local organizations (Ā§ 21180(b)(5)(A)(x)).
• Meets the Energy Commission opt-in program requirements (PRC Ā§Ā§ 25545.3, 25545.3.3, 25545.3.5) (Ā§ 21180(b)(5)(A)(xi)).

Additionally, the bill provides that the Office of Land Use and Climate Innovation must write uniform statewide standards for these conditions (Ā§ 21180(b)(5)(B)).

Ā鶹ֱ²„ā€™s Data Center team will continue to monitor and report applicable updates in due course.

On September 24, the Texas Attorney General’s office announced an investigation into hundreds of data center developments across the state, targeting facilities that have not responded to a water use survey administered by the Texas Water Development Board. The investigation spans 18 counties, including Harris, Travis, Dallas, Tarrant, and El Paso, and follows a series of other state actions on the same underlying issue during the preceding seven weeks.

None of this is happening in isolation. It is the latest major state action in a rapidly evolving regulatory response to a legal requirement that has existed for years but drew relatively little enforcement attention until the recent surge in data center development.

What the Law Actually Requires

The Texas Water Code requires TWDB to survey entities that use groundwater or surface water for municipal, industrial, power generation, or mining purposes, and it requires survey recipients to complete and return the survey. TWDB has sent the survey annually to major water users, including data centers, since 2020, asking for monthly water use figures. Under the statute, failure to return a completed survey is a Class C misdemeanor. A recipient that fails to timely complete and return the survey is also ineligible for certain TCEQ permits, permit amendments, or permit renewals, and ineligible for TWDB financial assistance, until the survey has been completed.

TWDB says the data matters beyond simple compliance. The survey results feed directly into the state’s 50-year water demand projections, which Texas regions use to plan future water infrastructure. A facility that does not report leaves a gap in those projections. TWDB has also noted that the survey is not sent to data center projects that are still in planning and have not yet been built, meaning current data does not capture future demand from facilities that are not yet online.

A Rapid Series of State Actions

The sequence started August 3, when Governor Abbott directed the PUCT and ERCOT to audit data center projects in ERCOT’s interconnection queue before any additional projects could move forward, citing noncompliance with the PUCT’s own water and power usage survey as one basis. ERCOT responded immediately, pausing the Batch Zero study process and delaying final classification of large loads, while seeking good-cause exceptions to extend related deadlines. As of late August, ERCOT told regulators it did not yet know how the resulting delay would affect its interconnection study timeline.

On September 14, Abbott followed with a letter directing TWDB specifically to use its enforcement authority against survey noncompliance, stating that some major water users, including data centers, “appear to have committed civil and criminal violations.” TWDB has said publicly that it is a non-regulatory body without independent authority to prosecute violations itself, leaving criminal enforcement to county or district attorneys, and that it shares compliance information with law enforcement upon request. Abbott asked TWDB for a compliance update by October 14.

On September 21, Abbott escalated the response further, directing TCEQ to halt all permits sought by data center projects until ERCOT completes its interconnection audit and the required information from ERCOT, the PUCT, and TWDB becomes available. The Governor’s office stated that no state agency should advance regulatory approvals related to data center development until that information has been obtained. TCEQ must provide an update to the Governor’s office on its compliance with the directive by October 19, 2026.

The Attorney General’s investigation, announced three days later, is the first action in this sequence initiated by an office with direct prosecutorial and civil-enforcement authority rather than a planning, regulatory, or information-gathering role.

The Scale of the Underlying Growth

Estimates of exactly how many data centers operate in Texas vary by source, which is itself part of what these reporting requirements are meant to address. The Texas Tribune puts the figure at 335 to 341 operating facilities, with roughly 248 more planned. Texas A&M’s Real Estate Research Center separately reported about 405 operating or under construction, with another 442 planned. Tracking services differ on where Texas ranks nationally: one national tracker counted 238 Texas facilities this year, the most of any state, while a separate tracker, Data Center Map, put Texas in second place nationally with more than 600 facilities operating or planned, behind Virginia’s 674. Real estate data from JLL has separately reported that Texas recently surpassed Virginia in the pace of new data center construction, even if not yet in total count.

ERCOT estimates of pending large-load interconnection requests have varied as the queue has expanded. An April 2026 ERCOT presentation cited approximately 410 gigawatts of pending requests, with about 87 percent associated with data centers. By August 2026, Governor Abbott cited approximately 474 gigawatts of pending requests, about 90 percent of which were attributed to data center projects. These figures represent interconnection requests rather than approved or constructed demand and should be understood as snapshots taken at different points in time. ERCOT has also said that a substantial share of projects in the queue have not yet reached the interconnection study stage.

ERCOT’s long-term forecast separately projects that data centers will consume more than 12,700 megawatts in 2026. Reported individual facility demand varies widely, from approximately 5 megawatts to proposed projects measured in the thousands of megawatts. Using the commonly cited rule of thumb that one megawatt can serve roughly 200 Texas homes during peak conditions, even a single large facility can represent substantial electric demand.

On water, a University of Texas at Austin white paper, produced as part of the Bureau of Economic Geology’s COMPASS research consortium, found that data centers currently account for less than 1 percent of the state’s water use but could reach 3 to 9 percent by 2040. That study counted 484 data centers in Texas that were operating, under construction, or planned as of September 2025. A separate white paper from the Houston Advanced Research Center projected that data center water use could reach 29 to 161 billion gallons a year by 2030, as much as 2.7 percent of the state’s total water use, and concluded that the state’s official water planning process did not then account for that growth.

TWDB’s year-by-year figures show the compliance gap persisting even as the survey pool has grown. The agency surveyed 22 data centers in 2023, and 32 percent responded. In 2024, it surveyed 67, and 28 percent responded. In 2025, it surveyed 329, and nearly 30 percent responded, though among the 267 facilities newly added to its tracking that year, only 22.5 percent responded.

At a June 23 hearing of the Texas House Committee on Natural Resources, lawmakers were separately told that only 17 percent of Texas data centers had responded to the 2025 survey. That figure does not align neatly with some of TWDB’s other reported response-rate statistics, and the agency has not publicly reconciled the apparent discrepancy. The difference may reflect variations in survey populations, reporting dates, or counting methodologies. At an August 2026 Texas Senate committee hearing, TWDB’s Temple McKinnon testified that the agency had sent surveys to 341 facilities and, as of that hearing, had achieved a roughly 30 percent response rate.

What the Investigation Says, and What It Doesn’t

The Attorney General’s announcement frames the investigation around the same statutory survey-response requirement referenced in Abbott’s letters and states that the office intends to ensure that data centers operating in Texas provide accurate, current water use information. The release does not name specific companies or allege that any particular facility has violated the law. Instead, it describes an ongoing, county-by-county investigation into developments that have not returned a survey response.

How the Industry Has Responded

The Data Center Coalition, an industry trade association, issued a statement following the August 3 ERCOT and PUCT directive saying that its members “strive to be good actors where we operate” and that the group hoped the audit would distinguish responsible operators from noncompliant ones. The statement noted the scale of investment and employment tied to data center development in Texas and urged the PUCT and ERCOT to move quickly so that serious, committed projects are not delayed alongside speculative projects still working through the interconnection queue.

That distinction, between projects that are further along and actually operating and those that remain speculative or in early planning, runs through much of the industry and regulatory commentary on this issue. Some projects within the broader data center universe may not yet have been built and therefore would not have received a TWDB survey. TWDB has acknowledged that its current process does not fully capture planned facilities that are not yet online.

What Comes Next

Several deadlines and review processes are converging. TWDB owes Abbott’s office a compliance update by October 14. TCEQ has been directed to provide the Governor’s office an update on its implementation of the permitting directive by October 19. ERCOT has not formally requested a change to its April 9, 2027 deadline for delivering Batch Zero study results, and multiple filings describe no confirmed impact to that date. At an August 20 PUCT meeting, however, an ERCOT official acknowledged the agency would not have the study complete by that date and said it was still determining a revised timeline. ERCOT’s audit and verification process was separately expected to conclude in early December 2026, according to an August 20 ERCOT filing. The Attorney General’s investigation has no stated public deadline.

For data center developers with projects in the ERCOT interconnection queue, the effect of these actions is cumulative: a paused and delayed classification process, TWDB compliance scrutiny, a statewide permitting halt pending completion of ongoing audits, and now an Attorney General investigation, all tied to the broader state effort to obtain more complete information about data center impacts on electricity and water resources.

Project timelines, financing agreements, and utility contracts tied to anticipated interconnection or permitting dates are now exposed to a review process whose full scope and duration the participating state offices have not yet defined.

In this episode of Energized by Ā鶹ֱ²„, Deanna Reitman speaks with David Dardis, Senior Executive Vice President and Chief External Affairs and Growth Officer at Constellation Energy, about his path from lawyer to business leader and what that perspective means in todayā€™s energy market. David discusses how his background in law, regulation, and policy has shaped his approach to growth, dealmaking, and strategy, and shares insights on the rising demand for power, the role of nuclear energy, new generation development, and the policy and market dynamics influencing the industryā€™s next chapter. The conversation offers a thoughtful look at leadership at the intersection of law, business, and energy, and the challenges and opportunities facing the sector as demand continues to grow.

On September 21, 2026, Governor Greg Abbott issued a new directive to the Texas Commission on Environmental Quality (TCEQ), ordering an immediate halt to the issuance of all environmental permits for data center projects.

This action marks a significant expansion from the stateā€™s initial grid interconnection audit and related directives to the Texas Water Development Board (TWDB), into a statewide, multi-agency permitting standstill spanning power, water, and environmental approvals.

1. Who is the TCEQ and What Permits Do They Oversee?

The TCEQ is the primary environmental regulatory agency for the State of Texas. It is tasked with protecting public health and natural resources by issuing permits, monitoring compliance, and enforcing state and federal environmental laws.

For data center developers, TCEQ authorization is typically required across several key operational areas prior to construction and energization, including:

• Air Quality Permits: Authorization (such as Standard Permits or New Source Review permits) for backup diesel or natural gas generators, fuel storage tanks, associated emissions, and possibly for co-located power generation.
• Industrial Wastewater Permits: Authorization for the discharge of cooling tower blowdown, hydrostatic test water, and other industrial wastewater.
• Stormwater Permits: General permits for stormwater discharges associated with construction activities and permanent site operations.
• Water Rights & Protection: Regulatory oversight regarding surface water use and diversions.
  

2. Key Provisions of the September 21 Directive

1. Immediate Permit Freeze: TCEQ is instructed to pause the issuance of all pending and new permits for data center projects until the Electric Reliability Council of Texas (ERCOT) completes its interconnection review and audit that Governor Abbott directed in August.
2. Whole-of-Government Standstill: The Governor explicitly states that no state agency may move forward with regulatory approvals for data centers until the grid and water audits conducted by ERCOT, the Public Utility Commission (PUC), and the TWDB. It is not yet known whether this standstill will extend to agencies that regulate other aspects of data center development, such as the Texas Department of Licensing and Regulation (which regulates industrialized building manufacturers that provide buildings for data center projects).
3. Cross-Agency Water Enforcement: The Governor reiterates that TWDB has been directed to compel data centers to comply with water use reporting requirements, enforce legal penalties for past non-compliance, and partner with ERCOT on water use audits.
4. Strict Funding, Consumer Protection, and Operational Mandates: The directive confirms non-negotiable conditions for data centers in Texas previously mentioned by the Governor:
   1. Infrastructure Costs: Projects must fully cover all electrical infrastructure costs and yield lower residential electricity bills.
   2. Water Resources, Power Consumption, & Setbacks: Data centers must not draw water needed by local communities, must report all electricity and water consumption, and must adhere to local community setback requirements.
5. Elimination of Financial Incentives: Governor Abbott announced plans to partner with the Texas Legislature to eliminate all financial incentives for data centers.
6. Agency Compliance Deadline: Engaging TCEQ to align permitting decisions ahead of upcoming legislative changes, the Governor instructs TCEQ to submit a formal progress and compliance report to the Governorā€™s office by Monday, October 19, 2026.

3. What Data Center Developers and Financiers Need to Know

• Comprehensive Project Delays: Obtaining an ERCOT queue position or power agreement are no longer the sole bottlenecks for data center projects. Even projects with secured power cannot construct or operate without required TCEQ permits, effectively stopping development timelines across all state regulatory bodies.
• Dual ERCOT & TCEQ Dependency: ERCOTā€™s verification auditā€”which involves Requests for Information on commercial readiness and community impacts for loads of 25 MW or moreā€”is expected to take several months. TCEQ permits will remain locked until ERCOT and TWDB publish their final findings.
• Re-evaluating Project Economics: Lenders and sponsors must factor in the potential loss of local tax abatements and state grants, along with higher capital expenditure requirements to cover 100% of grid interconnection infrastructure and non-potable water or closed-loop cooling systems.
• Immediate Audit Compliance: Developers must ensure that all historical and current water and energy usage data submitted to TWDB and ERCOT is complete and accurate. Submitting false or incomplete information carries severe regulatory penalties and queue disqualification.
• A Delay, Not a Death Knell:  The Governorā€™s directive is a forceful action which will undoubtedly cause delays to some projects.  However, it should not be read as state government opposition to data centers; rather, the Governor sets out general principles for data center development that some in the industry already support.  As these principles become more widely adopted, the pace of project development in Texas should normalize.

About Ā鶹ֱ²„ā€™s Data Center Team

Ā鶹ֱ²„ā€™s Data Center team advises developers, operators, investors, lenders, power providers, and other stakeholders on all aspects of data center development and operations. Leveraging experience across real estate, construction, energy, finance, regulatory, tax, and environmental matters, the team helps clients navigate the complex challenges and opportunities shaping the future of digital infrastructure.

About Ā鶹ֱ²„ā€™s Texas Government Solutions Group

Ā鶹ֱ²„ā€™s Texas Government Solutions team provides clients with unparalleled legal and legislative representation and counseling. The attorneys and lobbyists manage legislative and regulatory proposals at the Texas Capitol and before state agencies. The group counsels clients regarding key governmental processes and areas of law based upon decades of collective legal or governmental experience. Our Austin office is also home to a unique federal, state, and local government enforcement defense and investigations (GEDI) and litigation defense team, as well as several members of Ā鶹ֱ²„ā€™s national State Attorneys General practice.

About Ā鶹ֱ²„ā€™s Energy & Infrastructure Sector

Ā鶹ֱ²„ā€™s Energy & Infrastructure team provides clients with strategic legal counsel across the traditional and renewable power, oil & gas, infrastructure, and energy transition sectors. The team advises on the development, financing, acquisition, operation, and regulation of energy and infrastructure projects, helping clients navigate an evolving energy landscape and capitalize on emerging opportunities.

The U.S. power system ended 2025 with 43.6 gigawatts of utility-scale battery storage online. In just the first six months of 2026, operators added another 8.3 gigawatts, pushing total capacity to nearly 52 gigawatts. Over the prior three years, EIA data shows an annual average growth rate of 70 percent. Few segments of the energy industry are growing this fast, and few are being built on contract templates this unsettled.

That gap between deployment speed and legal infrastructure is where the real risk sits right now, for developers, for the utilities and data centers buying storage capacity, and for the lenders financing it.

What’s Actually Driving 52 Gigawatts

Storage is being built for two different reasons that happen to be converging at the same time. The first is the traditional pairing with intermittent renewables, storage smooths out solar and wind output and lets a project sell power when prices are higher rather than only when the sun happens to shine. The second, newer driver is data centers. Standalone battery projects are increasingly being sited to manage congestion and reliability challenges tied to rapid data center load growth, particularly in interconnection markets that are already constrained.

The industry expects roughly 24 gigawatts of new storage capacity in 2026 alone. A meaningful share of that is being financed and sited specifically around data center clusters, with ERCOT’s Dallas and Houston-area projects a leading example, positioned to capture both merchant revenue and behind-the-meter offtake arrangements with co-located facilities.

The Revenue Structure Isn’t the Same in Every Market

A storage contract that works in one market can be the wrong structure entirely in another, because the underlying revenue mechanics differ by region. ERCOT runs an energy-only market, so a battery there earns most of its money from arbitrage, charging when power is cheap and discharging when it’s expensive, with no guaranteed capacity payment underneath that strategy. California’s Resource Adequacy framework works differently: contracted capacity payments can represent 30 to 40 percent of a storage project’s annual revenue, a floor ERCOT projects don’t have. PJM’s recent capacity price spike signals its own acute need for dispatchable resources, changing the math again. A financing agreement or offtake contract built around one market’s structure doesn’t transfer cleanly to another.

The Regulatory Gap FERC Is Still Closing

The clearest sign that the legal framework hasn’t kept pace with deployment is a specific finding FERC made about PJM, the grid operator covering all or part of 13 states and the District of Columbia. FERC determined that PJM’s own tariff lacked sufficient rates, terms, and conditions of service governing co-located generation and load arrangements, the exact structure many storage-plus-data-center deals are built on. FERC ordered PJM to file a compliance filing by February 17, 2026, spelling out the specific terms an interconnection customer must follow to serve a co-located load, including how to designate the eligible customer that takes transmission service on the load’s behalf.

That’s a regulator telling one of the country’s largest grid operators, mid-buildout, that its existing rules don’t actually cover the arrangements companies are already signing. Storage developers and their counterparties negotiating co-location deals in PJM territory right now are working against a rulebook that was still being written as of early this year.

The FEOC Problem Nobody Priced In Two Years Ago

Foreign entity of concern restrictions now affect eligibility for the investment tax credit and the Section 45X manufacturing credit that much of the storage industry’s financing model depends on. Battery supply chains run heavily through a small number of international manufacturers, which means FEOC compliance documentation has become a real part of deal timing and cost, not a formality. Industry guidance for 2026 now treats FEOC documentation protocols as something that needs to be ready at signing, not something to work out after a customer relationship is already underway.

Companies that priced financing models two or three years ago, before FEOC rules were finalized, are discovering that the compliance framework underneath their tax credit assumptions has changed since the deal was structured. That’s the kind of gap that surfaces during due diligence on a financing round or an acquisition, not before.

What This Means for Anyone Building or Buying Storage

Four things are worth checking against current deals rather than assuming they were handled correctly when signed. Does the revenue structure match the market the project actually sits in, since an ERCOT arbitrage assumption built into a PJM capacity deal is a mismatch that eventually shows up in the financials. Do co-location and interconnection terms reflect the grid operator’s current tariff, not a template drafted before FERC forced operators like PJM to fill gaps in their own rules. Is FEOC documentation actually in place and current, given how much of the industry’s tax credit eligibility now depends on it. And have safety and siting requirements been checked against the newest state rules, since incidents at storage facilities have pushed several states toward new requirements that can affect permitting mid-project.

The Pace Isn’t Slowing Down

Nothing in the current data suggests this growth curve is flattening. Illinois enacted a wide-ranging energy law this year addressing battery storage payment structures and consumer protection requirements, one example among several states actively legislating in this space right now, and grid operators including the Southwest Power Pool have separately reformed their own interconnection queue processes to accommodate storage.

The companies moving fastest into battery storage right now are, almost by definition, moving faster than the legal and regulatory framework around them. That’s not a reason to slow down. It’s a reason to make sure the contracts, the interconnection terms, and the tax credit documentation are being built by people who are tracking where the rules actually stand today, not where they stood when storage was a much smaller part of the grid.

Ā鶹ֱ²„ partners Nick Peters, Deanna Reitman and Glenn Reitman recently presented a webinar updating on recent M&A trends across the energy sector.Ā At a high level, energy M&A activity in 2026 continues to reflect a market shaped by diverging sector dynamics, evolving demand patterns, and heightened strategic selectivity. In the oil and gas space, transaction activity remains meaningful, but buyers are proceeding with greater discipline amid commodity price volatility, geopolitical uncertainty, and persistent valuation gaps. Premium assets, particularly those with high-quality drilling inventory, strong development potential, and strategic infrastructure value, continue to attract interest, while natural gas and midstream assets are benefiting from renewed focus on domestic energy security, LNG demand, and the need for reliable power supply.

By contrast, M&A activity across the broader power, utility, and infrastructure landscape has accelerated significantly, driven in large part by surging electricity demand associated with AI and data center development. As developers, utilities, infrastructure investors, and technology companies seek scalable and dependable sources of power, existing generation, transmission access, and utility platforms have become increasingly valuable. In many cases, acquisitions offer a faster and more certain path to capacity than greenfield development, particularly given permitting delays, interconnection backlogs, and the growing premium placed on speed to power.

This dynamic is reflected in some of the yearā€™s most notable transactions. In oil and gas, buyers have continued to pursue scale, inventory depth, and strategic supply positions through transactions involving upstream and natural gas-focused assets. In the broader energy sector, large utility and infrastructure deals, along with acquisitions by technology companies seeking to secure dedicated energy resources, underscore a market increasingly centered on reliability, grid access, and long-term power availability. Taken together, these trends suggest that while traditional commodity-driven considerations remain important, the broader energy M&A market is being redefined by infrastructure constraints and demand growth tied to electrification and digital expansion.

Ā鶹ֱ²„ is actively advising clients across this evolving landscape, including in connection with upstream and midstream joint ventures, fund formations, M&A, pipeline development projects, renewable and conventional power transactions, and data center-related infrastructure investments. Our experience reflects what the market is increasingly demanding:  practical legal guidance at the intersection of energy supply, infrastructure development, capital deployment, and strategic growth. As 2026 continues, we expect successful transactions to be shaped by a common set of priorities, including access to power, infrastructure readiness, and the ability to move quickly in a highly competitive environment.

Read Part 1 here.

As utility-scale and commercial energy storage deployments continue to grow in record numbers and the universe of technology providers has expanded, more sponsors are finding themselves negotiating first-time Battery Energy Storage System (ā€œBESSā€) supply contracts and long-term services agreements with new counterparties. These negotiations do not occur in a vacuum, but rather in parallel with offtake agreements, engineering, procurement, and construction (ā€œEPCā€) contracts, financing goals, and evolving tax credit compliance requirements. Each parallel agreement presents assumptions and obligations concerning project performance, schedule, remedies, and liability. It is important to ensure these contractual structures remain aligned during live negotiations as project documents evolve, since misalignment can create gaps in risk for project sponsors. Once pricing is fixed and manufacturing slots are secured, addressing those gaps can be difficult, costly, or threaten to delay a project schedule.

In this two-part article, Part 1 focused on certain core structural provisions that shape the projectā€™s risk architecture from the outset. This Part 2 will focus on ongoing risk mitigation related to construction and operations, including key tax compliance, which requires a coordinated commercial and legal approach across tax, engineering, and operational disciplines that can be particularly challenging for firstā€‘time parties navigating evolving regulatory guidance and market practices.

Performance Guarantees and Capacity Buydown

Sponsors will typically structure EPC and BESS contracts to require minimum performance standards at commissioning and during operations that match corresponding sponsor performance obligations under offtake contracts. In the BESS supply agreement, these performance requirements will include things like guaranteed BESS capacity (in megawatt-hours), power (in megawatts), and round-trip efficiency, but may also include technical specifications for ramp rate, response time, and other specific operational expectations. These are incorporated into the commissioning and testing requirements in the supply agreement, and are typically continued in BESS service contracts as recurring performance requirements. Sponsors should consider the extent to which the supplier will have any ā€˜buydownā€™ rights to permit adjustments to its guaranteed performance values in order to achieve commissioning or later supplier milestones. For example, it is not uncommon for a supplier to propose a capacity buydown right that permits the supplier to achieve its commissioning milestone when testing shows the BESS failing to achieve the agreed target capacity but still achieving a lower minimum capacity level. Any supplier buydown rights in connection with its performance obligations under the supply agreement and/or services agreement should be carefully assessed on a cumulative basis and compared against ongoing offtake requirements to ensure a sufficient cushion in performance. This also includes confirming that, under any BESS services contract, any adjusted performance guarantees during operations are paired with sufficient performance liquidated damages.

Ambiguity in Commissioning and Testing Documentation

Generally, equipment documentation will include both detailed, product-level engineering documentation for the BESS equipment, plus project-specific documentation tied to construction, commissioning, testing, and operations. To the extent practicable, parties should seek to finalize as much project, commissioning and testing documentation at contract execution. Ambiguity around scope and responsibilities frequently emerges, particularly among the Owner, EPC contractor, and BESS Supplier during commissioning. Similar attention should be paid to operations and maintenance documentation, which will typically condition warranty coverage and performance guarantees on ongoing compliance with supplier technical and operational manuals. To the extent project timing considerations require that additional design or engineering work follow contract execution, the contract should establish a clear framework for the supplier to propose updates and for the project Owner to have a meaningful process to review and approve deliverables within well-established timelines. Earlier resolution of these materials will also be preferred by financing parties conducting due diligence.

Warranties and Performance Carveouts

Warranty and performance guarantee frameworks are most effective when any carveouts or exceptions are clearly stated and set forth the consequences for a project operating outside of warranted values. In practice, misalignment can arise from slightly different exclusions or terms among general equipment warranty provisions,  supply agreement performance guarantees, and longā€‘term services agreement guarantees, especially if a counterparty’s form has gone through various iterations over time without a comprehensive update. Counsel will need to work closely with technical leads to ensure that terms are applied uniformly across the documentation.

Serial Defects

In addition to confirming whether a supply agreement includes a serial defect provision, parties should carefully consider the details of how the provision is structured and the overall sequence of events under which it can be invoked. Key questions include what portion of the project ā€“ and what specific components ā€“ must be affected before a suspected defect is deemed ā€œserial,ā€ as well as the process and timing for conducting a root cause analysis. While agreements less frequently specify a fixed timeframe for full remediation given the potential wide scope and impact of a serial defect, project owners should pay close attention to their review and approval rights over any root cause analysis, remedial action plan, and remediation schedule. Approval rights are particularly important where extended or uncertain corrective measures could affect construction sequencing, financing, or project operations.

Tariffs and Price Adjustment Mechanisms

Trade policy volatility has required parties to adopt more creative approaches to allocate tariff-related risk in supply agreements. Approaches may include the use of an assumed tariff rate as of a specified point in time coupled with defined mechanisms to adjust pricing for subsequent tariff increases or decreases. While the allocation of incremental tariff risk is ultimately a matter of commercial negotiation, parties should pay close attention to how much exposure the supplier retains above any specified assumed tariff rate, when and how the owner begins to share in that exposure, and whether larger specified thresholds trigger a re-evaluation of the underlying pricing.

From a financing perspective, flexible mechanisms are not necessarily problematic, but it is generally preferable for any adjustment mechanics to be clearly memorialized up front and any ultimate adjustments to be memorialized before key debt and tax equity milestones to provide lenders and tax equity investors with greater visibility and mitigate the risk of late-stage contract price increases. A related consideration is whether the contract also permits contract price adjustment provisions tied to changes in critical BESS raw materials, such as lithium. In many cases, those raw material price indexes reflect the impact of tariff increases, so parties should take note of whether the indices referenced in raw materials price adjustments could have the effect of double-counting tariff impacts for purposes of the overall contract price.

To the extent offtake agreements include price adjustment mechanisms, information requirements and adjustment deadlines should be aligned with the supply agreement to mitigate any documentation and/or timing gap and ensure the sponsor has the necessary information for the offtaker when needed.

PWA Reporting and Compliance

Parties should carefully assess the supplierā€™s understanding of, and contractual obligations with respect to, prevailing wage and apprenticeship (ā€œPWAā€) requirements, particularly where the new supplier or its counsel has limited experience with U.S. projects and/or federal tax credit requirements. Generally, Ā鶹ֱ²„ has routine PWA reporting exhibits specific to supplier and EPC contracts, which allocate responsibility to each counterparty for its own and its subcontractorsā€™ PWA compliance, with corresponding obligations to cure nonā€‘compliance and cover associated penalties where failures occur. Common points for extended negotiation include the cadence and level of detail required for recurring PWA reporting, as well as the overall breadth of related indemnity coverages. From a riskā€‘management perspective, clarity around these mechanics at the outset can help mitigate uncertainty for sponsors and financing parties and reduce the potential for lateā€‘stage disputes around tax credit eligibility.

Domestic Content ā€“ Reporting and Risk Allocation

Unlike PWA rules, the 10% ITC domestic content adder does not provide cure opportunities for non-compliance once a project is placed in service. Parties should pay close attention to how much liability is allocated to the BESS supplier and EPC contractor, particularly given that the full value of the domestic content adder is likely to represent a significant percentage of overall project economics. As a result, it is often an open commercial point for parties to determine any one particular partyā€™s overall liability if their non-compliance threatens the entire adder amount.

Conclusion

For firstā€‘time counterparties, a deliberate, crossā€‘disciplinary approach to these issues is critical as BESS projects move from contract execution into manufacturing, construction, and operations.

Capital expenditures across the largest U.S. oil and gas producers fell 49 percent year over year even as production hit a study-period high. That’s the central finding of EY’s annual US Oil and Gas Reserves and Production Benchmarking Study, released September 9 and based on five years of data from the 30 largest publicly traded exploration and production companies, a group representing about 43 percent of total U.S. oil and gas production.

The two numbers don’t usually move in opposite directions this sharply. Oil production climbing to a record while spending is cut nearly in half points to an industry drawing more out of the wells it already has rather than chasing new ones.

Where the Money Stopped Going

Exploration spending fell 11 percent to $4.8 billion and now represents just 3 percent of total capital expenditures across the group. M&A spending dropped 70 percent as the industry moved past the wave of megadeals that reshaped the sector over the prior several years. Total capital expenditures, combining exploration, development, and acquisitions, fell 49 percent year over year.

Revenue still grew 7 percent, but pretax operating results fell 2 percent anyway, as lower commodity realizations compressed margins even as the top line improved. The only way that combination makes sense is if the extra revenue came from pumping more, not from getting paid more for it.

The Number That Should Worry Someone

Reserve replacement fell below 100 percent for the first time since 2021, meaning the group collectively pulled more oil out of the ground than it added back in new proved reserves. Natural gas told a different story: gas reserves reached a five-year high, supported by new discoveries, rising exports, and growing electricity demand, according to EY’s own release.

Matt Melnar, EY Americas Oil & Gas and Chemicals Assurance Leader, framed the split directly: oil production and reserve replacement are now moving in different directions, and reserve replacement metrics alone no longer tell the full story. “Producers are engaged in a balancing act between production goals, shareholder returns, and long-term portfolio resilience as they make investment decisions,” Melnar said.

Is This Actually a Problem?

A 49 percent capex cut sounds like retreat. In context, it reads more like a specific strategic choice: after years of consolidation and portfolio building through M&A, the industry is shifting its attention to running what it already owns more efficiently rather than continuing to buy or explore for more. That’s consistent with the broader capital discipline theme investors have rewarded across the sector for several years now, producers who spend less to produce the same or more, rather than producers who spend aggressively to grow reserves for their own sake.

The exploration figure is the sharpest illustration of that shift. Spending only 3 percent of total capital on exploration, down from levels that would have looked unremarkable five years ago, signals a group of companies betting that developing known assets more efficiently beats the higher-risk, higher-cost work of finding new ones. That bet works as long as commodity prices and drilling efficiency both hold up. It gets harder to defend if either one turns against producers at the same time reserve replacement is already running below 100 percent.

The bigger picture is straightforward: record production and lower spending can coexist, at least for now. Whether that balance holds, or reserve replacement eventually forces a different strategy, is the real question hanging over this data.

Who Actually Feels This

None of this is really about the operators. It’s about the companies that get paid to drill, complete, and service the wells those operators own. A 49 percent cut in total capital spending, weighted heavily toward exploration and new development, is going to be felt hardest by whoever depends on new wells getting drilled.

Start with the exploration number. Three percent of capital across a group that produces 43 percent of the country’s oil and gas is not a rounding error. It’s a group that has, for the moment, largely stopped looking for new reserves. Seismic work, exploratory drilling, first-time completions, that end of the business is going to feel thinner for a while.

The M&A slowdown cuts a different way, and it’s not all bad news. Fewer deals means fewer newly merged companies picking winners and losers among their inherited service contracts, which has been a real source of churn for OFS companies over the past few years. Less M&A activity, at least for now, probably means more stability in existing vendor relationships than upheaval.

Where there’s real opportunity is in what EY is calling a focus on existing assets. Operators spending less to find new oil and more to squeeze more out of what they’ve already got is good news for workover crews, well optimization work, artificial lift, anything that keeps a producing well productive longer. It’s a worse environment for companies built around new-well drilling and completion, and a better one for companies built around keeping older wells running.

The reserve replacement number is the one worth watching longest. Pulling oil out faster than you’re replacing it isn’t something an operator can do forever. At some point that either forces a change in strategy or gets bailed out by higher prices. Service companies with multi-year contracts tied to a particular operator’s production plan should treat today’s capital discipline as the new normal rather than a phase, and keep an eye out for the moment reserve pressure starts pushing spending back toward exploration.

None of This Accounts for the War

EY says as much itself: the study covers 2025 and doesn’t capture anything that’s happened since the Strait of Hormuz conflict started. Brent has traded above $100 for much of this year. That’s a very different price environment than the one these numbers were generated in.

So which way does that cut? Higher prices could easily talk some of these same producers into loosening up again, chasing bigger margins while they’re on the table. Or the discipline holds anyway, if this really was a shift in how the industry thinks about capital rather than just a reaction to a rough couple of years. Next year’s study will tell us which one it was.