Note: This digest synthesizes daily entries from Monday June 1 through Thursday June 4, 2026. The Friday June 5 daily entry was not present in the research log at the time of synthesis and is not reflected here.

Lede: The week’s defining development is the Google/Voltus “Bring Your Own Capacity” (BYOC) agreement announced June 2 — the first hyperscaler-funded virtual power plant, aggregating up to 100 MW/year of residential and commercial DERs in PJM with payments flowing to participating ratepayers. Paired with peer-reviewed Nature Energy/EPRI research demonstrating that a 256-GPU AI cluster in Phoenix achieved a 25% software-only power reduction for three hours during peak demand, hyperscaler load is being reframed from grid threat to potential grid resource at GW scale. In parallel, PJM accelerated its backstop reliability auction to September 2026, FERC rejected MISO’s Order 2222 transition model (pushing implementation to June 2029), and two independent analyses (Sightline Climate; Grid Strategies) argued that 30–50% of the 2026 data center pipeline may not materialize — strengthening the case for modular demand-side resources over speculative supply-side commitments.

🔋 Energy Storage

Wood Mackenzie 2025 LDES Outlook (Thu). Global long-duration energy storage deployments rose 49% YoY in 2025 to 15+ GWh, but LDES technologies face a “strategic squeeze” — lithium-ion has captured the economically critical 4–8 hour duration market (projected 85% of total energy storage through 2034) while multiday/seasonal storage lacks demand and price signals. Global LDES funding fell 30%; VC investment cratered 72%. China dominates with 93% of cumulative deployment. Three U.S. projects mark a commercialization pathway: Hydrostor’s 500 MW / 4 GWh Willow Rock facility in California (CEC approval December 2025, groundbreaking 2026); Form Energy / Google / Xcel Energy’s 300 MW / 30 GWh iron-air deployment in Minnesota (the world’s largest battery by GWh capacity, 100-hour discharge); and Google’s Energy Dome CO₂ battery partnership. WoodMac estimates average storage duration must rise from 2.5 hours to ~20 hours for reliability at high renewable penetration — a gap the market is not currently filling.

Minnesota Xcel Capacity*Connect (referenced Mon, Tue). PUC-approved $430M / 200 MW utility-owned, front-of-meter distributed battery program by 2028; implied installed cost ~$2,150/kW. First utility-owned VPP at scale; serves as a precedent that simplifies DERMS dispatch architecture (single owner, single control system) but raises competitive procurement questions.

⚡ Virtual Power Plants & Demand Flexibility

Google / Voltus “Bring Your Own Capacity” Agreement (Wed — flagship development of the week). Three-year agreement aggregating up to 100 MW/year of distributed batteries, smart thermostats, and flexible commercial assets into a Google-funded VPP within PJM. First BYOC agreement with a hyperscaler; Voltus pays participating DER owners, reversing the cost-shifting dynamic that has dominated the PJM backstop auction debate. Google separately disclosed it is working to unlock 1 GW of demand response capacity from its own data center operations through bilateral utility agreements — a direct commercial application of the Nature Energy / EPRI research findings (see below). Brattle Group estimate cited in the press release: $100B+ in consumer savings over the next decade from VPP-based grid optimization. For DERMS architects, the BYOC model demands multi-party orchestration spanning PJM’s 13-state footprint, accredited capacity bidding, real-time dispatch during grid stress, and dual settlement (hyperscaler buyer + DER owner).

SEPA Q1 2026 VPP & DER Policy Tracker (Tue). Most active quarter on record for state-level VPP policy action. 24+ states with VPP/DER policy activity in Q1; two states enacted dedicated VPP legislation. Headline actions: Massachusetts 3.5 GW load-management target including VPPs; Virginia legislature directed Dominion to develop a 450 MW VPP pilot; New Jersey Gov. Sherrill EO mandating BPU to develop a VPP program within six months; Colorado PUC-approved 125 MW Xcel VPP; Minnesota Capacity*Connect approval (referenced Mon). The political environment for demand-side resources has shifted from permissive to prescriptive — utilities are increasingly under legislative mandate rather than encouragement.

Pew Charitable Trusts DER Policy Playbook (Mon). Landmark 18-month playbook with bipartisan DER Advisory Council, organized around three goals and six policy actions. Key data: U.S. projected to add 217 GW of new DER capacity 2024–2028 (nearly equal to all U.S. coal capacity, more than 2× forecast data center demand through 2035); North American VPP market only 37.5 GW in 2025 with <20% of DER capacity enrolled — massive untapped potential. VPPs deliver peak power at 40–60% of conventional resource cost. Investor-owned utilities plan $1.1T in grid spending 2025–2029. Playbook explicitly couples interconnection reform with VPP expansion.

🔌 DERMS & Grid Integration Technology

Nature Energy / EPRI Field Demonstration on AI as Grid-Interactive Asset (Wed — landmark publication). First peer-reviewed validation that AI data centers can deliver demand-side flexibility through software alone. EPRI, Boston University, and industry partners ran a 256-GPU cluster in Phoenix executing representative AI workloads and achieved a 25% power reduction for three consecutive hours during peak demand, with AI quality-of-service maintained and no hardware modifications. Companion Nebius / NVIDIA / National Grid / EPRI demonstration at a London AI Factory (March 2026) confirmed the approach is cross-platform. If extrapolated across the 75.8 GW U.S. data center fleet (2026), the theoretical flexible capacity pool is ~19 GW — approximately equal to PJM’s full 14.9 GW backstop auction target. Software-only dispatch means existing ISO/RTO and utility DR signaling can be used; integration cost is dramatically lower than hardware-based DR. This finding underpins Google’s 1 GW DR commitment.

FERC Rejects MISO Order 2222 Transition Model (Mon). January 16, 2026 decision found MISO’s proposed DRR Type I model non-compliant because its 1 MW minimum size requirement is 10× Order 2222’s 100 kW threshold, effectively excluding the residential and small commercial aggregations the order was designed to enable. MISO implementation now pushed to June 1, 2029 at earliest. The widening RTO implementation gap (ISO-NE Nov 2026 → PJM Feb 2028 → MISO June 2029 → SPP Q2 2030) creates a fragmented national landscape: identical DER assets across regions will face vastly different wholesale market access for the next four years. Steptoe analysis notes FERC Chair Christie has publicly discussed sunsetting Order 2222 entirely; political future of the order is uncertain. MISO serves 42M customers across 15 states.

🏗️ Data Centers & Large Load Growth

Hyperscalers as Vertically Integrated Power Companies (Tue). Utility Dive analysis catalogs 19+ GW of hyperscaler-locked generation capacity: Microsoft (Three Mile Island restart 20-year PPA + 10.5 GW Brookfield partnership, $10B+); Meta (6.6 GW across TerraPower, Oklo, Vistra + 433 MW nuclear uprates); Amazon (1.92 GW Susquehanna + SMR exploration); Google (500 MW Kairos SMR by 2035 + $40B Texas buildout). Global data center electricity consumption approaching 1,050 TWh — roughly 3× 2024 levels; AI server racks at 40–100+ kW vs. 5–15 kW traditional. The March 2026 Ratepayer Protection Pledge signed by Microsoft, Google, Meta, Amazon, and xAI commits to covering 100% of facility transmission/infrastructure costs. Paradox for DR program designers: the largest, most flexible loads are also the ones most likely to leave the grid, simultaneously reducing the DR enrollment base and increasing stress on remaining customers.

Grid Strategies Critique of NERC 2025 LTRA (Thu). Consulting firm — commissioned by Earthjustice, NRDC, Sierra Club, and EDF — argues NERC’s combination of high demand forecasts (160 GW by 2030, 90 GW data centers), low generation-additions assumptions (only projects with signed IAs), and minimal interregional flow credits systematically overstates the reliability gap. The 90 GW data center forecast may be inflated by double-counting across utility territories and queues, supply chain constraints on chips and transformers delaying actual energization, and AI revenue uncertainty. During Winter Storm Uri, MISO imported 13 GW — far more than NERC’s firm-contract-only assumptions. Adding likely-to-connect queue projects “resolves the majority of identified seasonal adequacy shortfalls.” For IRP analysts, the critique provides a credible counter-narrative to the “build everything immediately” framing crowding out demand-side alternatives.

Sightline Climate 2026 Data Center Outlook (Wed). Tracking 777 projects >50 MW (190 GW aggregate planned capacity), Sightline finds 30–50% of the 16 GW of large data center capacity planned for 2026 globally is unlikely to come online this year. Only 5 GW currently under construction; 25% of projects haven’t disclosed their power strategy; community moratoriums have been proposed in 10+ U.S. states including Virginia. The 2025 precedent: 26% of 110 projects expected online were delayed. Developers often announce projects in parallel and test which can clear local regulations fastest — flooding utility interconnection queues with “phantom” load requests that inflate demand projections. The Grid Strategies and Sightline analyses converge on a single planning insight: utilities are being asked to commit $1.4T in capacity (PJM through 2030) to serve loads that may materialize at half the projected rate.

POWER Magazine on Data Center / Grid Infrastructure Mismatch (Thu). 300–600 MW hyperscale campuses are sited, permitted, and built in 18–36 months, while transmission substations and high-voltage feeds (230 kV / 345 kV) required to serve them take 5–10 years. Northern Virginia exceeds 3 GW of data center load. Power transformer lead times exceed 2 years. The article notes growing interest in behind-the-meter solutions including SMRs (50–300 MW), hydrogen fuel cells, and enhanced geothermal — which if adopted would reduce utility load growth. Fundamental structural argument for demand-side investment: when supply-side pipelines run on 5–10 years but load arrives in 18–36 months, only DR, VPPs, and distributed storage can scale to bridge the gap.

📋 Regulatory & Policy

PJM Backstop Reliability Auction Accelerated to September 2026 (Wed). PJM board moved the one-time backstop auction up from March 2027, responding to FERC Chairman Laura Swett’s pressure (PJM “potentially too big to function”) and a joint Trump administration / PJM-state governors statement demanding faster action on data center capacity shortfalls. Target reduced from 14.9 GW to ~9 GW of shortfall from the 2028/29 base capacity auction. PJM board letter warning to stakeholders: “After PJM runs the Backstop procurement, if states have not established frameworks to appropriately allocate costs to new data center loads, it is unclear to which customers those costs would be assigned” — effectively acknowledging that absent state action, residential ratepayers across 13 states could bear infrastructure costs built primarily to serve hyperscalers. PJM is also developing “connect and manage” rules that would require large loads to accept curtailment as a condition of interconnection — creating a mandatory DR obligation for data centers and requiring DERMS-grade monitoring to verify. Technical conference on PJM governance scheduled July 2026.

DOE Section 202(c) Emergency Orders Institutionalized (Tue). DOE now maintains dedicated year-specific 202(c) tracking pages (both 2025 and 2026 active) — a procedural development that itself signals normalization. 2026 orders include PJM Order 202-26-06 and Duke Energy Order 202-26-07, building on January cold-snap orders and the May 18 Order 202-26-23. Congressional Research Service report R48568 provides comprehensive analysis, reflecting growing legislative scrutiny. The institutionalization of what was designed as extraordinary authority converts every emergency order into a quantitative argument for expanded DR — each represents a grid condition where adequate demand-side resources could have reduced or eliminated the need for federal intervention. PJM May 18 event: <5,800 MW reserves with 40+ GW generation offline for maintenance against expected 134–136 GW peak loads; PJM DR capacity sits flat at 7,864 MW.

🏭 Utility Programs & Deployments

Xcel Minnesota Capacity*Connect VPP (referenced Mon). $430M / 200 MW utility-owned distributed battery deployment by 2028, first utility-owned VPP at scale. PUC conditions include independent program evaluation, priority placement in underserved communities, and Building Strong Communities apprenticeship partnership.

Dominion 450 MW VPP Pilot (Virginia, referenced Tue, via SEPA Q1 tracker). Legislatively directed.

Xcel Colorado 125 MW VPP (referenced Tue, via SEPA Q1 tracker). PUC-approved, first Xcel Colorado VPP.

🔬 EPRI Research Spotlight

Nature Energy: “AI Data Centres as Grid-Interactive Assets” (Volume 11, Pages 254–261, February 2026). Peer-reviewed EPRI-led field demonstration described above. The first rigorous validation that AI workloads contain inherent flexibility dispatchable via software-only intervention. Cross-platform applicability validated through the companion Nebius / NVIDIA / National Grid / EPRI demonstration at Nebius’s London AI Factory (March 2026). Direct relevance to DERMS architecture (no new hardware integration required, dispatchable via existing ISO/RTO and utility DR signals) and to the entire data center capacity debate — the very loads driving the grid crisis may be a substantial portion of the solution.

🚩 Utility-Sector Relevance Flags

⚑ Google / Voltus BYOC Sets Hyperscaler-Funded VPP Precedent
Topic: VPP / Large Load Integration
Relevance: First market-based mechanism where a hyperscaler funds DER capacity that serves the broader grid, with payments flowing to participating ratepayers. Establishes a replicable framework for monetizing the Brattle-estimated $100B in consumer savings from VPP optimization while directly addressing PJM’s cost-allocation problem. Utilities in hyperscaler-heavy territories should evaluate whether BYOC-style structures can be embedded in large load interconnection tariffs.
Action Signal: Engage — Begin tariff and program design conversations; coordinate with state regulators on enabling frameworks before competing models lock in.

⚑ Nature Energy / EPRI Demonstrates 25% Software-Only DR from AI Workloads
Topic: Demand Flexibility / EPRI Research
Relevance: Peer-reviewed proof that AI data center load is dispatchable as a grid resource without hardware investment. If extrapolated, the U.S. fleet contains ~19 GW of theoretical flexible capacity — equivalent to PJM’s full backstop auction target. Reframes hyperscalers from problem to potential solution.
Action Signal: Implement — Utilities with data center load should begin program design conversations with hyperscaler customers; integrate AI workload flexibility into IRP demand-side resource portfolios.

⚑ PJM Backstop Auction Accelerated; Cost Allocation Crisis Unresolved
Topic: Regulatory / Capacity Markets
Relevance: ~9 GW capacity procurement now arriving September 2026. PJM board has explicitly warned that absent state cost-allocation frameworks, residential ratepayers across 13 states could shoulder costs built primarily for hyperscalers. The “connect and manage” rules under development effectively mandate DR participation as an interconnection condition — a regulatory pathway that converts DERMS investment from discretionary to compliance-driven.
Action Signal: Engage — Participate in PJM stakeholder process; coordinate with state PUC on cost-allocation frameworks and “connect and manage” implementation. Engage with FERC technical conference (July 2026).

⚑ FERC Rejection of MISO Order 2222 Model Creates 4-Year RTO Implementation Gap
Topic: Regulatory / DER Market Access
Relevance: Identical DER assets will have vastly different wholesale market access depending on RTO until 2030 (ISO-NE Nov 2026 → MISO June 2029 → SPP Q2 2030). DERMS vendors and VPP operators building multi-regional platforms face fragmented market design. For DSM/DR programs, wholesale market revenue cannot be assumed as a near-term value stream in MISO and SPP territories; state regulatory pathways must carry the business case.
Action Signal: Watch — Monitor FERC Chair Christie’s signaling on potential Order 2222 sunset; revise IRP modeling to exclude wholesale market revenue in MISO/SPP timeframes through 2029.

⚑ Data Center Pipeline Attrition (30–50%) Reshapes Capacity Planning Risk
Topic: Load Forecasting / Large Load Integration
Relevance: Convergent finding from Sightline Climate and Grid Strategies: NERC’s 160 GW load forecast and PJM’s $1.4T CapEx through 2030 are sized to data center demand that may materialize at half the projected rate. Lumpy supply-side commitments face stranded-asset risk; modular demand-side resources (DR, VPPs, DERMS-orchestrated portfolios) can scale with actual rather than speculative load growth. Distribution planning models must adopt scenario-based load forecasts that account for high attrition rates in large load projections.
Action Signal: Implement — Update IRP and distribution planning models to incorporate data center attrition scenarios; weight demand-side resources higher in resource adequacy portfolios.

⚑ State-Level VPP Mandates Shift from Permissive to Prescriptive
Topic: Regulatory / VPP Policy
Relevance: SEPA Q1 2026 tracker confirms 24+ states with VPP/DER policy action in a single quarter — Massachusetts 3.5 GW load-management target, Virginia 450 MW Dominion pilot, New Jersey BPU 6-month VPP mandate, Colorado Xcel 125 MW. Utilities are increasingly under legislative mandate rather than encouragement. Strengthens avoided-cost business case by transforming DR/VPP investment from discretionary optimization into compliance obligation with regulatory cost recovery.
Action Signal: Implement — Utilities operating in active-policy states should accelerate program filings; DERMS procurement requirements must be architected for regulatory flexibility across market structures.

⚑ Long-Duration Energy Storage Faces “Strategic Squeeze” — Reinforces Near-Term DR Value
Topic: Storage / Resource Planning
Relevance: Wood Mackenzie finds lithium-ion will hold 85% of the storage market through 2034; LDES funding fell 30% in 2025. The 2.5-hour to 20-hour duration gap WoodMac identifies for grid reliability will not close on a near-term timeline. DR and VPPs remain the lowest-cost, fastest-to-deploy tools for the 4–8 hour peak windows that drive most capacity market procurement.
Action Signal: Watch — Track Hydrostor Willow Rock and Form Energy / Google / Xcel iron-air commercialization milestones; LDES will not be a planning-grade resource in the 2026–2030 IRP cycle.

⚑ Hyperscaler Vertical Integration Erodes DR Enrollment Base
Topic: Large Load Integration / DR Program Design
Relevance: 19+ GW of hyperscaler-controlled generation through nuclear, SMR, and long-term PPA commitments. The most flexible loads are simultaneously the most likely to leave the grid — reducing addressable DR base while increasing stress on remaining customers. The March 2026 Ratepayer Protection Pledge signals hyperscaler willingness to internalize infrastructure costs, but also confirms they view themselves as independent market participants.
Action Signal: Watch — Monitor hyperscaler behind-the-meter generation commitments in service territories; adjust DR enrollment projections accordingly.

⚑ DOE 202(c) Emergency Orders Now Institutionalized
Topic: Regulatory / Grid Reliability
Relevance: Year-specific DOE tracking pages and CRS report R48568 reflect a permanent expansion of federal emergency intervention. Every order is a quantitative argument for expanded DR — PJM May 18 event saw <5,800 MW reserves with PJM DR flat at 7,864 MW. The data center backup generation provisions expand the universe of DERMS-addressable assets into industrial/commercial backup generation.
Action Signal: Engage — Utilities with backup generation–capable large loads should evaluate DERMS integration of these assets; participate in DOE 202(c) implementation discussions and Congressional oversight.

DER Weekly Digest — Week Ending Friday, June 5, 2026