All five weekday entries (Monday August 3 through Friday August 7) are present in the research log. Substantial consolidation was required this week: the NERC Level 3 Alert / 1,800-MW data-center load-loss story appeared in both the August 3 and August 7 entries and is deduplicated below; the New Jersey BPU VPP straw proposal appeared in both the August 3 and August 6 entries (and was covered in last week’s digest) and is treated here only for its new detail; DTE’s $105/kW-year residential battery pilot surfaced in both the August 3 and August 6 entries via SEPA’s Q2 roundup and is consolidated; and the Bank of America 230-GW adequacy forecast (August 4 entry) is standing context previously carried, not a new item. Two log entries — August 6 and August 7 — carry the daily task’s own note that no qualifying articles published in the trailing 48 hours, so those items reach back to April–July source dates; they are reported here on the day they entered the log, with original publication dates shown. The week’s defining development is that the regulatory calendar this series has tracked for six weeks arrived: NERC’s Level 3 Alert response deadline and Governor Abbott’s ERCOT queue audit both landed August 3, PJM’s Reliability Backstop Procurement was filed at FERC July 31 (Docket ER26-3380-000), and PJM’s Interim Resource Adequacy Service filing is due today.
Three things converged this week, and together they change the shape of the argument. First, the large-load numbers utilities have been planning against started collapsing under scrutiny: ERCOT is tracking roughly 474 GW of large-load interconnection requests — more than five times the grid’s all-time peak — and Governor Abbott ordered every project in that queue verified for demand, water use, subsidies, and ownership, with ERCOT pausing its Batch Zero review until the audit finishes. Exelon supplied the utility-level confirmation the same week: its “high-probability” data-center load fell nearly 40% in a single quarter, from 18 GW to about 11 GW, once transmission security agreements imposed real credit obligations, and ComEd canceled a TSA for a 1.8-GW, $20 billion Joliet campus. Second, the obligations attached to large loads became binding rather than advisory: NERC’s Level 3 Alert response deadline passed August 3, converting EMT ride-through modeling for large loads from a request into a compliance record that feeds Project 2026-02 mandatory standards. Third, the procurement venues opened: PJM filed its one-time backstop auction at FERC on July 31 with a $555/MW-day cap and up to $20 billion at stake, and files the Interim Resource Adequacy Service — mandatory large-load registry, curtailment before Pre-Emergency Load Management, and a federally approved hourly credit rate — today. The synthesis for demand-side planners is sharper than in any prior week: the planning problem is no longer only that firm supply is scarce, it is that the load forecasts justifying firm supply are unreliable by roughly 40%, while the cost of guessing wrong is now a $20 billion backstop bill allocated to ratepayers. Incrementally-scalable, dispatchable flexibility is the only resource class that is correctly sized under both outcomes.
🔋 Energy Storage
Austin Energy launched its Power Partner Battery Pilot with EnergyHub as the orchestration layer, offering $500 upfront toward a qualifying home battery plus roughly $75/kW delivered on average seasonal peak reduction — an average of more than $300 per year — with a 1,500-system cap that yields only about 7.5–15 MW against the utility’s 78-MW-by-2027 and 270-MW-by-2035 demand-response targets. The compensation design is the useful part: Austin’s per-kW rate sits well below DTE’s proposed $105/kW-year and the ConnectedSolutions residential rate of roughly $275 per average event kW, but it is paired with a substantial hardware buy-down — a different lever mix for the same enrollment-friction problem, trading performance-payment richness for upfront capital subsidy. For utilities calibrating incentive structures, the three points now bracket the live 2026 range. The 1,500-system cap is the constraint worth noting for IRP purposes: this is a learn-then-scale pilot that builds telemetry, dispatch, and M&V record before the resource becomes countable, and at ~10% of the 2027 target it will not close the gap on its own. Approved hardware spanning Tesla, FranklinWH, SolarEdge, and Enphase demonstrates the vendor-agnostic device integration a mature DERMS has to deliver, and the structure lets a municipal utility use customer-owned storage in place of rate-based generation CapEx. (Source: EnergyHub, April 2026 — August 7 entry)
Atlantic City Electric’s proposed 500-MW/4-hour battery in New Jersey, claiming $1.36 in benefits per $1 spent, is the utility-owned-storage counterpart moving through the same state that just opened its VPP docket. Disclosed inside Exelon’s Q2 discussion alongside nearly 175 MW of approved BGE/Pepco VPP capacity in Maryland and a ComEd VPP program taking effect in March, it shows a holding company running utility-scale storage, utility-owned VPP capacity, and third-party aggregation simultaneously — and pricing all three against the same avoided-cost test. (Source: Utility Dive — August 5 entry)
⚡ Virtual Power Plants (VPP) & Demand Flexibility
Sunrun, Tesla, and Renew Home unveiled a framework to deliver up to 16.8 GW of distributed capacity to utilities and hyperscalers — roughly 9 GW from more than 8 million Renew Home-managed smart thermostats plus 7.8 GW from Sunrun- and Tesla-operated home batteries, aggregating as many as 12 million existing devices across 9 million homes including Alaska, Hawaii, and Puerto Rico. This is the largest single VPP commitment this series has carried, and its framing is the point: it is pitched explicitly as a “speed-to-power” answer to data-center interconnection delay, capacity that can be stood up faster than new firm generation or transmission. Set against the 116-GW gas-turbine backlog and four-to-five-year lead times documented last week, 16.8 GW of already-installed, behind-the-meter capacity needing only orchestration and a compensation signal is a direct substitute for utility CapEx in exactly the window where firm supply is unavailable. Two cautions belong with it. First, this is a framework announcement, not enrolled or accredited megawatts — the conversion from “devices reachable” to “capacity biddable” is the step that has bounded every VPP in this series, and North America’s whole fleet stands at 37.5 GW after growing only 13.7% last year. Second, coordinating 12 million heterogeneous devices for accredited, dispatchable performance is a demanding telemetry, forecasting, and settlement problem — which is precisely why platform capability and market access have to advance together. Note this resolves a standing watch item: the Tesla/Sunrun/Renew Home 16.8-GW concept, flagged as “no movement” in prior weeks, now has a defined framework. (Source: pv magazine USA, June 25 — August 7 entry; movement on watched item)
Massachusetts’ ConnectedSolutions extension to residential vehicle-to-grid arrives with the hard, audited numbers that make it the M&V reference case for the series: more than $5.4 million paid to 5,251 residential battery participants for the 2025 season, roughly 40 MW of net summer capacity delivered (28.9 MW residential, nearly 11 MW commercial), against a 2025–2027 plan targeting about 125 MW of battery performance capacity. The residential compensation rate — roughly $275 per average kW of summer event contribution, translating to $1,375–$2,750 per participant — is a performance-paid tariffed structure, not a wholesale-signal-dependent one, and the program’s demonstrated 375-plus MW shave during a June 2024 New England heat wave is the peak-coincident capacity that anchors avoided-capacity claims. The V2G extension itself is the new operational step: EnergyHub’s Edge DERMS manages dispatch, The Mobility House supplies the bidirectional chargers, Sunrun participates, and initially eligible vehicles include the F-150 Lightning, Nissan Leaf, Kia EV9, Polestar 3, and Volvo EX90. Because enrolled batteries can qualify under both ConnectedSolutions and the Clean Peak Energy Standard, the program is also the cleanest live demonstration of the value stacking that lifts a DR portfolio’s benefit-cost ratio. (Source: pv magazine USA, July 27 — August 6 entry; extends the V2G/ConnectedSolutions item covered July 31)
Rappahannock Electric Cooperative partnered with Virtual Peaker to launch a DERMS-backed DR/VPP program across 22 Virginia counties, enrolling member thermostats, batteries, EV chargers, and EVs for events dispatched on real-time grid conditions — with enrollment integrated directly into the existing MyREC SmartHub billing portal. The significance is the tier: a member-owned distribution cooperative, not a large IOU, responding to rising power-supply costs and data-center-driven load growth by standing up commercial-DERMS-orchestrated flexibility. The SmartHub integration is the design detail worth copying — putting enrollment inside the account and billing flow customers already use is a direct attack on the enrollment friction that determines whether a VPP reaches accredited scale. Framed inside REC’s “CORE Grid” strategy, it demonstrates that SaaS DERMS platforms have lowered the barrier far enough that avoided-cost pursuit no longer requires a bespoke utility build. As a vendor release it is light on MW targets and incentive levels; the named utility, platform, device classes, and billing integration are specific enough to mark a real launch. (Source: Virtual Peaker via Utility Dive, July 30 — August 5 entry)
New Jersey’s VPP straw proposal (deduplicated — appeared August 3 and August 6, covered in last week’s digest) adds one operative new fact: written comments under Docket QO26030099 are due Monday, August 17, 2026. That is the same date show-cause answers are due from all six RTOs — a genuine calendar collision for anyone participating in both. The program covers all four state electric utilities (PSE&G, JCP&L, Atlantic City Electric, Rockland Electric), enrolls home batteries, smart thermostats, and EV chargers, and compensates through bill credits, discounts, or direct payments under Governor Sherrill’s January 20 Executive Order No. 2 energy-emergency directive. (Sources: NJ BPU, July 27 — August 3 and August 6 entries; deduplicated)
Compensation benchmarks, consolidated (deduplicated): DTE Electric’s proposed two-year Michigan Residential Battery VPP Pilot at $105/kW-year, capped at $504 annually, using customers’ existing batteries with no utility-owned hardware, appeared in both the August 3 and August 6 entries via SEPA’s Q2 2026 roundup and was covered July 31. Read against the series’ ~$66/kW-year Brattle avoided-capacity anchor, DTE’s premium implies either valuation beyond capacity alone (energy plus T&D deferral) or a deliberate enrollment premium. The SEPA roundup’s wider catalog — Illinois’s ComEd/Ameren Rider SDVPP scheduled-dispatch tariffs and ComEd’s bring-your-own-device program under S.B. 25 (cost-effectiveness review due end-2026), Maryland’s third-party-accessible DER registry and contested utility DERMS timelines, Hawaii’s competitively-procured edge-DERMS agent with a five-year minimum term and performance-prioritized compensation, Entergy Louisiana’s approved $81M 2026–2030 DR suite, and Virginia’s four new VPP/non-wires laws (H.B. 1467/285/429/434) mandating pilots up to 150 MW and DR/storage IRP scenarios — was covered in the July 31 digest and is standing context. (Sources: SEPA/NCCETC, July 23 — August 3 and August 6 entries; DTE via SEPA — August 6 entry; deduplicated, previously covered)
🔌 DERMS & Grid Integration Technology
FERC’s May 29 approval of ISO-New England’s revised Order 2222 market rules re-entered the log this week with the participation scope stated plainly: aggregations of batteries, smart thermostats, EV chargers, demand response, and rooftop solar can bid into ISO-NE energy, capacity, and reserve markets, with full regional implementation targeted by year-end 2026. This is the conversion step the series repeatedly flags between enrolled megawatts and accredited, biddable capacity — and it is what makes the ConnectedSolutions battery above worth more than its utility DR payment alone, since wholesale capacity and reserve revenue can now stack on top of the program payment and the Clean Peak credit. The DERMS procurement consequence is direct and unavoidable: bidding aggregations into ISO-NE requires telemetry, forecasting, schedule optimization, and settlement-grade M&V, so market access and platform capability advance in lockstep. The regional contrast is the planning fact worth carrying: ISO-NE’s year-end-2026 target against MISO’s Phase 1 (June 1, 2027) and Phase 2 (June 1, 2029) means New England aggregators face a monetization window roughly two to three years ahead of their Midwestern counterparts, which should shape where multi-region platforms and aggregators prioritize integration spend. (Source: Troutman Washington Energy Report, June 2026 — August 6 entry; movement detail on a previously covered item)
The orchestration-layer architecture for large-load flexibility got its clearest description this week, and it is a DERMS pattern in all but name. Emerald AI sits between the utility and the data center, translating grid requirements into dispatch targets and returning telemetry, verification, and event-compliance reporting, while the utility “retains full dispatch authority” within pre-agreed guardrails — demonstrations have delivered up to 40% flexibility (SRP ramped a workload 25% over a three-hour peak; PGE simulations cut load 20%), and Boston University modeling found AI workloads can offer 18–55% flexibility while meeting quality-of-service. EPRI’s FlexMosaic framework supplies the valuation grammar: five flexibility classes from Class A (infrequent extreme stress) through Class E (frequency stabilization), keyed to notification time, duration, frequency, and depth of response, with the finding that Class D/E resources mitigating local thermal overloads or voltage drops unlock the most system value — which is granular, locational value-stacking of exactly the kind that strengthens avoided-T&D arguments. The binding constraint is named explicitly and matches the series’ standing thesis: what is missing is a standardized binding interconnection agreement guaranteeing utility “visibility and dispatchability.” Utilities (PGE, Silicon Valley Power) insist on it plus a non-negotiable loadside-breaker safety net; data centers resist instantaneous control. The EPRI–Open Compute Project MOU and the 96-MW Aurora AI Factory in Manassas, Virginia (online late 2026) are the standardization steps that would convert this from pilot to procurable resource. (Source: Utility Dive — August 4 entry)
🏗️ Data Centers & Large Load Growth
Governor Abbott ordered a comprehensive verification audit of every data center in ERCOT’s interconnection queue on August 3, and ERCOT paused its Batch Zero large-load review until it completes — against a queue ERCOT reported on July 29 at roughly 474 GW across more than 1,800 projects, about 90% data centers, and more than five times the grid’s all-time peak-demand record. Each project must now substantiate its power demand, water use, public financial assistance, community protections, and ownership before advancing; no completion deadline was announced. This is the strongest state-level statement yet that headline queue figures are not planning data, and it is direct movement on the ERCOT Batch Zero watch item — though not the movement expected, since the August applicant classification notices this series was watching for are now suspended rather than issued. For IRP work the implication runs one way: if a large share of a 474-GW queue evaporates under verification, utilities that committed lumpy, long-lead firm generation against it have stranded ratepayer capital, while dispatchable DR, DSM, and storage sized incrementally to load that actually materializes carry no equivalent exposure. The audit also reinforces that curtailability, visibility, and verified demand are becoming interconnection conditions rather than afterthoughts — and it is a template other states with runaway queues will copy. (Source: Power Magazine, August 3 — August 5 entry; movement on watched item)
Exelon proved the point at the utility level within days: “high-probability” data-center load fell nearly 40% in one quarter — from 18 GW to about 11 GW — and the broader interconnection pipeline dropped from roughly 43 GW to about 25 GW, after the company began requiring transmission security agreements that impose credit obligations, committed-revenue contributions, and shortfall payments. Only about 4 GW of the remaining high-probability load has signed TSAs, backed by $1 billion in posted collateral, and ComEd canceled a previously approved TSA tied to a planned 1.8-GW, $20 billion Joliet campus. This is the cleanest available demonstration that requiring financial commitment collapses paper demand — and the single most citable number for any load-forecast challenge in a rate or IRP proceeding this year. Pair it with PJM’s summer auction, which cleared at the cap, missed the reserve-margin target by 6.8 GW, and drew only 525 MW of new generation: the market cannot fill even the verified gap at current price signals, which is why Exelon’s own answer runs through demand-side and utility-owned resources rather than merchant supply. (Source: Utility Dive, July 31 — August 5 entry)
NERC’s Level 3 Alert on data-center load loss hit its structured-response deadline August 3 (deduplicated — appeared in both the August 3 and August 7 entries), converting the problem from advisory warning into a planning obligation. The 2026 State of Reliability report documents the trigger: a single normally-cleared transmission fault in the Eastern Interconnection caused 1,800 MW of data-center load to trip offline in milliseconds when UPS systems and automatic transfer switches sensed the voltage sag and moved the facilities to backup before the grid recovered — with four further Eastern events (428, 227, 540, and 1,300 MW) and nine separate ERCOT crypto-mining load-loss events over 100 MW each in the same year. Nothing failed; the protection logic worked exactly as designed, which is what makes clustered campuses running identical logic a correlated multi-gigawatt risk. The alert directs Transmission Planners, Planning Coordinators, Transmission Owners, Balancing Authorities, and Reliability Coordinators to build and validate electromagnetic-transient models of large-load behavior and share them across large loads, transmission owners, and planners. Alongside it: the new PERC1 power-electronics model requiring field-validated rather than datasheet parameters, the draft Computational Load Entity registration category, and mandatory standards under Project 2026-02, with the August 3 responses to 33 implementation questions feeding directly to FERC. The reliability-side argument for controllable load follows directly — load that can drop 1,800 MW in seconds is precisely the volatility flexible DER portfolios buffer, and NERC’s own diagnosis moves large loads into the same compliance framework as generators. Also logged: conventional-generation weighted forced-outage rate rose to 9.2% in 2025, with coal at 14.1% (up from 11.2%). (Sources: Tech Times, July 4 — August 3 entry; Utility Dive — August 7 entry; deduplicated)
Adequacy context (previously covered): Bank of America’s forecast — 230-plus GW of new capacity needed over five years against roughly 93 GW of accredited utility additions, data centers alone adding ~125 GW at a 4.1% CAGR through 2030, large gas turbines sold out to 2030 with GE Vernova’s backlog at 116 GW, 7.5-plus GW of behind-the-meter data-center generation under construction and 60-plus GW pre-construction, and coal retirement delays across Maryland, Wisconsin, Indiana, Utah, Kansas, Nebraska, and Mississippi — re-entered the log August 4 and was covered in the July 24 digest. It remains the standing scarcity citation. U.S. data-center demand is separately projected at 66 GW in 2027 (from 31 GW in 2025) and up to 17% of U.S. electricity by 2030 per EPRI. (Source: Utility Dive, July 17 — August 4 entry; previously covered)
📋 Regulatory & Policy
PJM filed its Reliability Backstop Procurement at FERC on July 31 (Docket ER26-3380-000) — a one-time auction running September 30 to October 21, results by December 2, to fill the 6.8-GW shortfall from a base auction that cleared at the price cap for the third straight time while attracting only 525 MW of new generation — with a $555/MW-day cost cap, up sharply from $325, and an NRDC estimate that it could commit up to $20 billion for new plants serving data centers built through 2027. Two design features matter more than the headline for demand-side planners. First, the procurement target is reduced to reflect documented bilateral contracts and self-supply arrangements for new capacity serving new load — meaning every megawatt of accredited demand-side flexibility that clears, or that a bilateral removes from the target, directly reduces the backstop bill. Second, cost allocation is explicitly contingent on states defining which “large loads” bear the costs, which puts the ratepayer-protection fight in state proceedings rather than at FERC alone. PJM’s projection that large-load demand could reach 70 GW by 2038 establishes this as structural rather than a one-off. (Sources: Utility Dive, August 3 — August 5 entry; PJM FERC filing 20260731-5214, ER26-3380-000 — external docket verification)
The paired Interim Resource Adequacy Service filing — renamed from “Connect and Manage” and due at FERC today, August 7 — is the more consequential of the two for demand-side practice, because it creates the registry and the compensation mechanism at once. Per PJM’s July 27 Board decisional letter and CIFP executive summary, the proposal defines a “Large Load” as an end-use customer with peak load of 50 MW or more at a single point of interconnection, or at multiple points within a one-mile radius; establishes a PJM-maintained mandatory Large Load Registry to improve load forecasting and support both new services; and provides that new large loads that have not brought their own generation by June 1, 2027, and have not otherwise secured supply, will be curtailed before Pre-Emergency Load Management deploys. Compensation for directed reductions runs through Electric Distributors in coordination with states, at a federally approved hourly credit rate set equal to the Non-Performance Assessment Interval rate for existing Pre-Emergency Load Management events — approved by FERC on June 26, 2026, and equal to 50% of the PAI rate. That last detail is the one to carry into program design: PJM is pricing mandatory large-load curtailment at half the performance rate of voluntary load management, which preserves — deliberately — a compensation premium for resources that commit in advance rather than being curtailed by obligation. It is the clearest signal yet of how the boundary between compelled curtailment and paid flexibility will be priced. (Sources: PJM Board decisional letter and CIFP Interim Resource Adequacy Service executive summary, July 27; Foley Hoag; White & Case — external verification of an item flagged in the August 5 log entry)
Texas drew the same boundary the other way, and harder. The PUCT’s July 23 order approved a 260-MW AI data center co-located with a roughly 265.5-MW wind farm — but in one of the first major tests of Senate Bill 6, it rejected the developer’s argument that a second co-located load (bringing site demand to ~525 MW against 265.5 MW of generation) should escape curtailment, requiring instead that the data center shed its full load within 30 minutes during grid emergencies, with physical breaker disconnection if necessary. Critically, the order bars the project from paid demand-response programs tied to the arrangement, explicitly separating mandatory emergency curtailment as a reliability obligation of operating behind generation from voluntary compensated grid services. ERCOT gets 60-minute advance notice “when practicable,” with an optional 10-minute voluntary response. As GridTracker’s Chris Talley put it, this is not a death blow to co-location but it effectively requires full backup capacity — reshaping behind-the-meter DER-plus-load economics and setting the standard the pending Amazon/Vistra campus at Comanche Peak will be measured against. For utilities drafting large-load tariffs elsewhere, Texas has established that curtailability can be mandated as a condition of interconnection without triggering DR compensation, which shrinks the paid-DR value pool for large loads even as it expands the flexibility obligation. Read against PJM’s 50%-of-PAI credit rate, the two jurisdictions are converging on the same principle from opposite directions: obligated curtailment is worth less than committed flexibility, and the design question is only how much less. (Source: Utility Dive, July 30 — August 4 entry)
PJM’s Independent Market Monitor put the cost-allocation stakes in dollars: data centers accounted for $6.3 billion — 38% — of the $16.4 billion in charges from PJM’s latest base capacity auction, and $29.4 billion (46%) of $63.6 billion across the last four base auctions. Joseph Bowring argues PJM “is continuing to act like it’s business as usual” while data-center growth imposes higher capacity, energy, and transmission costs on other customers, and presses for a separate data-center capacity auction under 15-year contracts — a proposal this series has been watching for treatment in the RBP filing. Those same capacity charges are the avoided-cost pool dispatchable DR, storage, and VPP capacity offsets, which is what makes the backstop auction a venue where demand-side flexibility competes against new firm supply rather than merely observing it. Morningstar DBRS’s warning of rising data-center opposition, cancellations, taxes, and moratoriums compounds the forecast-uncertainty argument that Exelon and the ERCOT audit made empirically this week. (Source: Utility Dive, July 20 — August 4 entry; partial movement on the IMM 15-year-auction watch item)
DOE’s Section 202(c) emergency authority is now a routine summer instrument, and its cost and ineffectiveness are both documented. The log’s August 7 entry records a 202(c) order for PJM signed June 30 and extended through July 6, authorizing PJM to direct any customer drawing at least 50 MW at a single delivery point onto onsite backup generation within 15 minutes of an emergency signal — data centers and hyperscale AI campuses the explicit targets, with hospitals, 911 centers, water-treatment plants, and air-traffic control exempt. Against a historical baseline of roughly 26 uses across all US grid operators from 2000 through early 2026, multiple orders on one regional grid in a single year is unprecedented; note that source accounts differ on the ordinal count and exact dates across the June–July sequence (this series previously logged a PJM order July 14–21 and Order 202-26-37 to SPP July 26), so the pattern rather than the tally is the reliable fact. The supply-side backstop those orders lean on is failing on its own terms: five of the six plants DOE ordered to delay retirement produced just 1.5 million MWh in Q1 2026, down 65% from 4.3 million MWh a year earlier, with TransAlta’s 730-MW Centralia running zero hours and Craig Unit 1 running about two weeks; CenterPoint estimated up to $20.5 million and 14 weeks offline to make its 104-MW Culley Unit 2 compliant, calling extended life “neither practical nor financially responsible”; and the Sierra Club puts the program’s cost at roughly $550 million a year. When the marginal firm resource is a forty-year-old coal unit kept alive by emergency order at high cost and low availability, the avoided-capacity and CapEx-deferral value of dispatchable demand-side flexibility rises by exactly that much. (Sources: Utility Dive, July 2026 — August 7 entry; Utility Dive, June 23 — August 3 entry)
The FERC show-cause calendar advanced two steps this week. August 3 was the deadline for RTOs and transmission owners to request full or partial abeyance of the Section 206 show-cause proceedings (Dockets EL26-67 through EL26-72) in order to develop Section 205 filings — abeyances capped at 90 days, with FERC on record that it will scrutinize requests heavily and disfavor extensions. August 17 is the substantive deadline: all six RTOs/ISOs and their transmission owners must either show cause why their tariffs remain just and reasonable without large-load reforms, or state what tariff changes would remedy the Commission’s concerns. Note the collision — August 17 is also the NJ BPU VPP comment deadline. MISO’s Zero Injection GIA filing, expected on or about July 31, appears to have slipped: reporting as of mid-July indicated MISO delayed the filing by roughly a month over stakeholder apprehension while still targeting implementation by end-2026, and no July 31 filing was confirmed. That remains an open verification item for next week. (Sources: Day Pitney; McGuireWoods — external docket verification; RTO Insider, July 19 — external verification)
🔬 EPRI Research Spotlight
No new EPRI-sourced research entered the log this week, and a direct check confirms no major EPRI announcement since the Flex MOSAIC framework (March) and the nine-site DCFlex expansion announced at DTECH in February — making this the fourth consecutive dry week in the log. EPRI’s substantive influence this week is nonetheless load-bearing in three places. First, the FlexMosaic five-class framework (Class A infrequent extreme stress through Class E frequency stabilization, keyed to notification time, duration, frequency, and depth, with Class D/E identified as unlocking the most system value) is the valuation grammar underneath the August 4 orchestration piece and is the most likely reference framework for any binding flexibility interconnection agreement. Second, EPRI research is the identified source of the ride-through root-cause diagnosis behind NERC’s 1,800-MW and 1,300-MW load-drop findings, which now feeds the Computational Load Entity registration record. Third, DCFlex’s nine demonstration sites — with Compass Datacenters, Constellation, Emerald AI, Google, National Grid, Nebius, NVIDIA, Oracle, and PADO AI — plus the EPRI–Open Compute Project MOU and the 96-MW Aurora AI Factory in Manassas (online late 2026), constitute the standardization pathway from pilot to procurable resource.
The standing watch tightens: NERC’s Computational Load Entity Rules of Procedure comment period is expected to open in August, following the initial 45-day period that closed May 15. The proposed criteria — loads of 20 MW and greater, connected at 60 kV, containing more than 1 MW of IT load — would pull a large population of facilities into NERC registration and Reliability Standards compliance, and EPRI’s DCFlex data is the most credible technical record available to that proceeding. Board approval is targeted for December 5. Check EPRI.com directly again next run.
🚩 Utility-Sector Relevance Flags
⚑ The 40% Load-Forecast Correction: Exelon and ERCOT Made the Same Point in One Week
Topic: Load Forecasting / IRP Planning Discipline / Capital Allocation
Relevance: Exelon’s high-probability data-center load fell from 18 GW to ~11 GW in one quarter under transmission security agreements, its pipeline from ~43 GW to ~25 GW, with a canceled 1.8-GW Joliet TSA as the concrete case; ERCOT is auditing a 474-GW queue that exceeds its peak record fivefold. Planning firm CapEx against unfiltered queue totals is now demonstrably unsound, and the correction runs in the direction that favors incrementally-scalable demand-side resources over lumpy, long-lead generation.
Action Signal: Implement — Cite the Exelon quarter-over-quarter figures in any load-forecast challenge or IRP demand-side justification this cycle; apply a financial-commitment filter (credit obligations, committed revenue, shortfall payments) to your own large-load pipeline before it enters a capital plan, and size demand-side procurement to the filtered number.
⚑ PJM Prices Obligated Curtailment at Half of Committed Flexibility
Topic: DR Compensation Design / Large Load / Market Design
Relevance: The Interim Resource Adequacy Service credit rate is set at the Non-PAI rate — 50% of the Performance Assessment Interval rate — for large loads directed to reduce consumption, administered by Electric Distributors in coordination with states. Read alongside Texas barring its co-located 260-MW data center from paid DR entirely, two major jurisdictions have now established that compelled curtailment earns materially less than pre-committed, accredited flexibility. That spread is the economic case for enrollment.
Action Signal: Implement — Use the 50%-of-PAI differential in customer-facing large-load flexibility offers: the value proposition is now quantified by the market operator itself. Review any DR tariff that fails to price a premium for advance commitment.
⚑ PJM’s Mandatory Large Load Registry and the June 1, 2027 Bring-Your-Own-Supply Cliff
Topic: Large Load Interconnection / Registry / Curtailment Priority
Relevance: New large loads (≥50 MW at a single POI, or multiple POIs within one mile) that have not brought their own generation by June 1, 2027, and have not otherwise secured supply, will be curtailed before Pre-Emergency Load Management deploys — with a PJM-maintained registry making them visible. This is the visibility-and-dispatchability infrastructure that converts enrolled flexibility into a countable reliability resource, and it arrives with a hard date.
Action Signal: Engage — PJM-footprint utilities should inventory customers meeting the 50-MW/one-mile threshold now and brief them on the June 1, 2027 cliff; the registry requirement is also the natural hook for offering DERMS-orchestrated flexibility as the alternative to being curtailed uncompensated at half rate.
⚑ NERC Level 3 Deadline Passed — EMT Ride-Through Modeling Is Now a Compliance Record
Topic: Bulk-System Reliability / Large-Load Modeling / Standards Trajectory
Relevance: The August 3 structured response to 33 implementation questions feeds directly to FERC and Project 2026-02 mandatory standards. Validated EMT models of large-load behavior, the PERC1 field-validated power-electronics model, and the draft Computational Load Entity category together move large loads into the generator compliance framework. The 1,800-MW correlated trip is also the reliability argument for controllable, buffering DER portfolios.
Action Signal: Implement — Confirm your organization’s Level 3 response is filed and consistent with your interconnection-study assumptions; begin scoping EMT model acquisition from large-load customers as an interconnection-agreement requirement, and audit protection-logic diversity across clustered campuses on your system.
⚑ NERC Computational Load Entity Comment Period Opens This Month — 20 MW / 60 kV / 1 MW IT
Topic: Registration / Compliance Scope / Data Centers
Relevance: The proposed registration criteria are broader than most operators assume: 20 MW and above, connected at 60 kV, containing more than 1 MW of IT load. That captures a substantial population of facilities beyond hyperscale campuses, including many colocation and enterprise sites, and pulls them into Reliability Standards compliance. Board approval targeted December 5.
Action Signal: Engage — Run the 20 MW / 60 kV / 1 MW-IT screen across your service territory now to size the affected customer population, and prepare comments when the August window opens; utilities with large colocation footprints have the most at stake in where the threshold lands.
⚑ 16.8 GW Announced vs. 37.5 GW Installed: The Sunrun/Tesla/Renew Home Framework
Topic: VPP Scale / Speed-to-Power / Third-Party Aggregation
Relevance: A single framework offering nearly half the size of North America’s entire existing VPP fleet, built on 12 million already-installed devices across 9 million homes, pitched directly at the interconnection-delay problem. It is a framework, not accredited capacity — but it establishes third-party aggregation as a procurement channel competing with utility CapEx, and it resolves a standing watch item that had shown no movement.
Action Signal: Watch — Track conversion from announced framework to signed utility and hyperscaler offtake; the first accredited megawatts under it will indicate whether device reach translates to biddable capacity at this scale. Utilities should clarify now whether third-party VPP capacity in their territory competes with or complements their own programs.
⚑ ConnectedSolutions Is the Audited M&V Reference Case — $5.4M, 5,251 Participants, 40 MW
Topic: DR Program Evidence / Compensation Benchmark / V2G
Relevance: Preliminary administrator data filed with the Massachusetts Energy Efficiency Advisory Council gives regulator-facing, audited numbers — $5.4M paid, 5,251 participants, 40 MW net summer capacity, ~$275 per average event kW — that most VPP claims lack. The V2G extension on EnergyHub Edge DERMS with The Mobility House chargers moves vehicle-to-grid from economics into an operating program with named eligible vehicles.
Action Signal: Implement — Cite the filed ConnectedSolutions performance data alongside GMP’s $6M realized savings in avoided-cost filings: one is audited program-level M&V, the other is realized system-level value. Use the ~$275/avg-kW rate and Austin’s $500-plus-$75/kW structure as the two poles when designing residential storage incentives.
⚑ ERCOT Batch Zero Is Paused — Recalibrate Any Texas Large-Load Timeline
Topic: Interconnection Process / State Regulatory / Queue Management
Relevance: The August applicant classification notices this series was watching for will not issue on schedule; the Batch Zero collective study of 75-MW-plus loads is on hold pending an audit with no announced completion deadline. Every Texas large-load interconnection timeline built on the prior roadmap needs revision, and the audit template is likely to spread to other states with inflated queues.
Action Signal: Watch — Reset Texas project timelines and advise affected customers; utilities in other states with large speculative queues should anticipate similar verification demands and prepare the demand, water, subsidy, and ownership documentation their own commissions may soon request.
📌 Sources
August 3, 2026 Entry
– Tech Times — AI Data Centers Triggered 1,800 MW Grid Drop; NERC Issues Highest Alert (July 4, 2026)
– New Jersey Board of Public Utilities — VPP Straw Proposal and Stakeholder Docket QO26030099 (July 27, 2026)
– SEPA / N.C. Clean Energy Technology Center — VPP and Supporting DER Policy Developments, Q2 2026 (July 23, 2026)
– Utility Dive — DOE 202(c) Plants Producing Far Less Than Before: Centralia, Campbell, Schahfer (June 23, 2026)
August 4, 2026 Entry
– Utility Dive — Bank of America Warns AI Data-Center Growth Will Blow a 100-GW-Plus Hole in U.S. Supply (July 17, 2026)
– Utility Dive — Hyperscalers Ready to Trade Load Flexibility for Faster Interconnection; EPRI FlexMosaic and Emerald AI (June 26, 2026)
– Utility Dive — PJM IMM: Data Centers Drove $6.3 Billion of Latest Capacity Auction Charges (July 20, 2026)
– Utility Dive — Texas Approves AI Data Center Co-Location Next to Wind Farm With Curtailment Requirement (July 30, 2026)
August 5, 2026 Entry
– Power Magazine — Abbott Orders Full Audit of Texas Data Center Interconnection Queue (August 3, 2026)
– Utility Dive — PJM Files Backstop Capacity Auction Plan at FERC (August 3, 2026)
– Utility Dive — Exelon Data-Center Load, New Jersey Battery, Q2 Earnings (July 31, 2026)
– Virtual Peaker via Utility Dive — Rappahannock Electric Cooperative Turns to Smart Device Technology (July 30, 2026)
August 6, 2026 Entry
– New Jersey Board of Public Utilities — Statewide VPP Straw Proposal; Comments Due August 17 (July 27, 2026)
– pv magazine USA — Battery VPP That Paid Homeowners $5.4 Million Now Expanding to EV Batteries (July 27, 2026)
– SEPA — VPP and Supporting DER Policy Developments Q2 2026 (DTE $105/kW-Year Pilot)
– Troutman — Washington Energy Report: FERC Approves ISO-NE Updated Market Rules for DERs (June 2026)
August 7, 2026 Entry
– Utility Dive — NERC Issues Rare Level 3 Alert Over Data-Center Load Losses (May 2026)
– Utility Dive — PJM/DOE Emergency Order to Curtail Data Centers (July 2026)
– pv magazine USA — Sunrun, Tesla, Renew Home Announce Plans for 16.8 GW Virtual Power Plant Program (June 25, 2026)
– EnergyHub — Austin Energy Partners With EnergyHub to Expand Multi-DER VPP With Power Partner Battery Pilot (April 2026)
External Verification (targeted searches, not from the daily log)
– PJM — Reliability Backstop Procurement FERC Filing, Docket ER26-3380-000 (July 31, 2026)
– PJM Board of Managers — Decisional Letter on CIFP Reliability Backstop Procurement and Connect and Manage (July 27, 2026)
– PJM — Interim Resource Adequacy Service Executive Summary (July 27, 2026)
– Foley Hoag — PJM Board Directs FERC Filing on Reliability Backstop Procurement and Interim Resource Adequacy Service (July 2026)
– White & Case — PJM Proposes to Carve Out New Services for Co-Located Data Centers
– Day Pitney — FERC Issues Show Cause Orders to Six RTOs/ISOs on Large Load Integration
– McGuireWoods — FERC Issues Section 206 Show Cause Orders Directing All Six RTOs/ISOs to Justify or Reform Large Load Integration Rules (June 2026)
– RTO Insider — MISO Presses Ahead With Zero-Injection Gen Agreements (July 19, 2026)
– NERC — Computational Load Entity, Summary of Changes (Rules of Procedure Appendix 5B posting)
– Steptoe — NERC Releases Proposed Registration Requirements for “Computational Load” Customers
– EPRI — DCFlex Initiative Expands to Nine Demonstration Sites Across U.S., Europe
Next digest: week ending August 14, 2026. Carried verification items — MISO Zero Injection GIA filing status (apparently delayed past July 31); PJM Interim Resource Adequacy Service docket number once posted; which RTOs filed August 3 abeyance requests; CAISO large-load straw proposal (August 12); August 17 show-cause answers and NJ BPU VPP comments; NERC Computational Load Entity comment window opening; FERC response to the five-state “but-for” rehearing; ERCOT audit scope and duration.