The race for grid capacity is reshaping where data centers get built, who builds them, and how fast they can get power. This week: PJM's fragmentation risk hits the mainstream press, Google's Texas co-location model signals a structural shift in how hyperscalers think about generation, and two gigawatt-scale campuses in Ohio and Illinois surface with unmet power origination needs. Queue data across PJM and WECC puts supply context behind each of these signals.
Power market intelligence from a practitioner with two decades in origination and interconnection — 5+ GW delivered to NTP across CAISO, PJM, MISO, ERCOT, and WECC. The analysis connects public news to what's actually moving in the queue.
Feature
Connector Markets — Load & Supply in the Same Geography
A "connector market" exists where a known data center load announcement overlaps with viable generation in the same ISO/state footprint. These are the highest-value origination windows — one conversation can structure both sides of a deal. Three markets stand out this week.
| Market |
Load Signal |
MW |
Viable Supply in Queue |
Strategic Note |
| OH · PJM |
QTS Van Wert mega-site · $10B campus |
~2,000+ |
23 projects ≥ Score 6 Top: PJM-AE1-092 (207 MW Solar, Score 7)
PJM-AG2-329 (200 MW Battery, Score 7) |
AEP Ohio territory. Strong Columbus-area queue density. Multiple Feasibility-stage projects with motivated sellers. |
| IL · PJM |
Eagle Rock Christian County · No in-house power team |
1,000 |
5 projects Score 7 PJM-AD1-100 (850 MW Wind, IA Executed)
PJM-AI1-149 (250 MW Battery, Cook) |
Property developer with no apparent origination capability. 2029–2030 construction start = early engagement window now. |
| VA · PJM |
Northern Virginia Corridor Multiple hyperscaler campuses active |
Multi-GW |
12 projects Score 7+ PJM-AG2-518 (80 MW Battery, Score 8, Lightshift)
PJM-AI1-018 (300 MW Battery, Spotsylvania) |
Dominion territory. Most congested PJM queue. Premium location, high network upgrade costs. Acquisition plays require deep queue knowledge. |
Queue Spotlight — Highest-Score Acquisition Targets This Week
PJM-AG2-518
80 MW Battery · Montgomery, VA · Score 8
Lightshift Energy · System Impact Study
West-G0998
1,120 MW Nuclear · Benton, WA · Score 7
BPA · Early study · BTM/co-location candidate
PJM-AD1-100
850 MW Wind · Kankakee, IL · Score 7
IA Executed · 7+ years · Motivated seller signal
West-Q278
250 MW Battery · Maricopa, AZ · Score 7
APS territory · Phoenix load pocket
Interconnection
PJM Fragmentation Risk — What It Means for Power Procurement
AI load growth is straining the country's largest grid operator to a potential breaking point — and that creates both risk and arbitrage for developers who understand queue dynamics.
Reports this week flagged the possibility of PJM — which serves 13 states and 67 million customers — fragmenting under the weight of data center load growth. The mechanism: state-by-state energy policy divergence, transmission cost allocation disputes, and a reformed queue that drew 220 GW of new requests in Cycle 1 alone against a grid that took decades to build. PJM expects electricity demand to increase by more than 30 GW between 2024 and 2030, driven largely by data centers.
For power procurement teams, fragmentation risk cuts two ways. On the supply side, queue projects in contested transmission zones face higher network upgrade cost exposure — a 500 kV POI in a congested corridor may look attractive until the SIS comes back. On the demand side, any ISO restructuring resets the tariff and interconnection rules mid-cycle, potentially invalidating PPA assumptions baked into financial models years earlier.
The practical read: projects currently in System Impact Study in PJM's most congested zones (Northern Virginia, ComEd, AEP Ohio) should be stress-tested against both current tariff rules and a fragmentation scenario. Greenfield data center siting decisions made today will be executing against whatever grid structure exists in 2028–2030.
PJM Queue — Current Snapshot (Source: PJM, May 2026)
220 GW
Cycle 1 requests — 811 projects, first reformed queue cycle
30+ GW
Expected demand growth 2024–2030, driven by data centers (PJM forecast)
160+ GW
Studied since 2023 reforms; ~46 GW transition projects remaining through 2026
Development Activity
Campus Announcements — Scale & Power Implications
Google's Texas Co-Location Model Changes the Conversation
1 GW+ generation paired directly with a data center campus in Gray County, TX
Google and Intersect — a clean energy developer Google acquired for $4.75B in March 2026 — broke ground this week on the Meitner Energy Center, a co-located data center and generation complex in Gray and Roberts Counties, Texas Panhandle — wind, solar, storage, and gas firming in one package. This is the "power-first" siting model in practice: find the generation, then build the campus around it.
For developers without Google's balance sheet, this model is instructive. The interconnection queue strategy, PPA structure, and site control sequence that makes a 1 GW co-location work are exactly the skills that transfer to smaller campuses — and most data center developers don't have them in-house.
Eagle Rock: 1 GW in Illinois, No Power Team
$8.8B campus proposed in Christian County — construction 2029–2030, community opposition emerging
Eagle Rock's Christian County, Illinois announcement is a textbook early-stage origination opportunity. A property developer — not an energy company — is proposing 1 GW of data center load in MISO with a 3–4 year construction horizon and no apparent in-house power origination capability. The $8.8B campus would span 475 acres with ten 100 MW buildings.
The Illinois queue has viable supply: five Score 7 projects within the state including an 850 MW wind project with IA Executed status. One watch item: community opposition is active, with a petition for an 18-month moratorium gathering ~700 signatures. Project approval is not guaranteed — engagement now, while the project is still in early discussions with no formal application filed, is the right timing.
QTS Van Wert: $10B in AEP Ohio Territory
Mega-site acquisition signals multi-GW demand in Columbus corridor
QTS's acquisition of the Van Wert, Ohio mega-site — with a $10 billion campus announcement — puts gigawatt-scale load directly in AEP Ohio territory. The Columbus load pocket already has 23 queue projects scoring ≥6 within reach, including multiple Feasibility-stage solar and battery projects in Marion, Logan, and Madison counties.
The interconnection path in this corridor runs through the Marysville-Flatlick 765 kV hub — a known congestion point. Network upgrade cost exposure is real, but project selection at the right POI can thread that needle.
Stargate Expands to Michigan — Massive Load on DTE Grid
Oracle/OpenAI break ground in Saline Township; Governor Whitmer attends
The Oracle/OpenAI Stargate campus in Saline Township, Michigan — "The Barn" — marks the first major hyperscale data center build in Michigan in years. DTE Energy's transmission system will be the interconnection path, and the state's relatively clean queue (compared to Northern Virginia or ComEd) means earlier-stage projects may still have viable interconnection economics.
Michigan's queue is thin on supply-side candidates relative to load growth implied by this campus, which creates scarcity value for any project with a credible IA path in DTE territory.
Regulatory & Policy
Signals Worth Watching
This Week's Policy Flags
Eagle Rock · Christian Co. IL
Community opposition petition (~700 signatures) seeks 18-month moratorium on data center development. No formal application filed yet — project approval not guaranteed. Watch county board proceedings.
PJM Fragmentation
Mainstream press coverage elevates political risk of ISO restructuring mid-cycle. Monitor FERC docket for any state withdrawal filings or tariff revision proposals.
Google–Xcel PPA · MN
$1.5B projected customer savings validates utility-hyperscaler PPA model. Sets pricing precedent for MISO-zone load-serving deals in 2026–2027 procurement cycles.
SoftBank · France · 5 GW
EDF partnership on former power plant sites — repurposing existing interconnection rights. Model directly applicable to US coal retirement sites in PJM and MISO.
Compass · Ellis County TX
$15M community investment to resolve opposition to 830-acre campus. Community benefit agreements increasingly required for large campuses in constrained markets.
PPA & Financing
Market Pricing Context
The Google–Xcel deal in Minnesota — projected at $1.5B in customer savings over 15 years on a data center PPA — provides a useful public benchmark for utility-scale PPA economics tied to hyperscaler load. The deal structure (utility-administered, load-serving agreement) differs from a direct corporate PPA but reflects the pricing dynamic: hyperscalers are willing to pay above-market rates for certainty and deliverability, and utilities are willing to structure creative vehicles to land the load.
The Crusoe–Bergen Engines deal (750 MW, gas reciprocating engines) represents the other end of the spectrum — behind-the-meter, firm capacity, no ISO queue exposure. For campuses that can't wait for queue projects to clear study phases, BTM gas remains the fastest path to firm power, at the cost of carbon exposure and regulatory scrutiny.
About This Brief
NexPhase Advisory
The Interconnect is produced by NexPhase Advisory — a power origination and interconnection strategy practice serving data center developers, IPPs, and investors navigating the US power market. Service lines include interconnection queue strategy, deliverability analysis (CAISO TPD, PJM NLTP, MISO GIA), PPA origination, and market entry advisory.
Questions, feedback, or to discuss a specific market or project: info@nexphaseadvisory.net