The post-July 4 solar frame gets described as an inventory story. Wood Mackenzie’s post-deadline tally put utility-scale solar safe-harbored between mid-2024 and July 4, 2026 at 216 to 240 GW DC, and the intuitive read is that the bottleneck moves from origination to modules, cells, and transformers. That read is half right. The other half is that all of that inventory has to land on a physical grid point, and the number of large-block interconnection agreements that can absorb a modern solar-plus-storage complex is not growing in step with the safe-harbored bank sitting on top of them.
Enlight Renewable Energy’s CO Bar close on June 25 is the clearest picture of what this looks like at execution. Clēnera Holdings signed roughly $2.6 billion in debt across a seven-bank syndicate ($1.70 billion of term debt plus $1.45 to $1.52 billion of expected tax equity) against 1.2 GW of solar and 4.0 GWh of storage in northern Arizona. The lender list (BNP Paribas, Crédit Agricole CIB, MUFG, Natixis, Nord/LB, Société Générale, Wells Fargo) is a full syndicate of global infrastructure banks. Revenue is fully subscribed across five 20-year offtakes with Salt River Project and Arizona Public Service. It is a large deal by any 2026 standard.
What makes the deal unusual is not the size. It is that all five sub-projects, all five offtakes, and the entire debt package sit under a single 1 GW AC interconnection agreement.
The anchor is the interconnection point, not the megawatts
Every element of the CO Bar capital stack points at the same asset. The syndicate underwrote against contracted revenue that terminates at the SRP and APS meters. The tax equity underwrote against Section 48E qualification and the domestic content adder mechanics. The 20-year PPAs are structured around a specific delivery point. The debt tenor is set against a physical interconnection agreement that will not exist twice in the same substation.
A different way to read the same deal: subtract the interconnection agreement and CO Bar is five separate projects that have to run five separate queue applications, five separate site-control processes, and five separate rounds of interconnection facilities studies and system impact studies. Add the interconnection agreement back in and CO Bar is one project that clusters five deployment slots inside a pre-cleared 1 GW AC envelope. The interconnection right is not one of several inputs. It is the single asset that lets the rest of the stack close on the same date.
This is not new in principle. Anchor-tenant structures have existed in industrial parks for decades and appear in wind repowering under a different name. What is new is the scale and the market timing. A 1 GW AC connection point is a large asset by any US utility standard, and it is the kind of asset that Arizona, Texas, and California can produce roughly on the same cadence they have been producing them for the past decade. It is not the kind of asset that scales with the safe-harbored solar inventory sitting behind it.
The scorecard says reform, not throughput
Advanced Energy United’s June 23 progress report on generator interconnection reform is the clean read on what is available for the next Enlight to buy. SPP and CAISO earned “promising improvements.” PJM, ISO New England, and NYISO drew “expected progress.” ERCOT was rated “incremental.” MISO came in “incomplete.” Every RTO now runs cluster studies. Every RTO now uses gating deposits and readiness milestones to cull speculative applications. Every RTO has, on paper, implemented FERC Order 2023.
The paragraph in the AEU report that the framing here rests on is the one that says none of that has actually raised the rate at which queued megawatts reach commercial operation. AEU describes what has changed as “shifting more viable projects up in the queue.” That is a real gain if you already have financing and site control. It is a smaller gain if the question is grid-connected megawatts.
If throughput at the RTO layer is flat, the supply of large-block interconnection agreements is also flat, or close to it. The demand side does not have the same problem. Safe-harbored inventory of 216 to 240 GW DC is not sitting still. It has 18 months from the OBBBA July 4, 2026 deadline to hit commercial operation for the paired placed-in-service window (December 31, 2027) or it defaults into a tail that Wood Mackenzie’s own read describes as rarely feasible on typical three-year timelines. Every quarter that passes narrows the window. The demand for interconnection agreements steepens; the supply does not.
What that does to the valuation stack
Three consequences follow, and they are consequences developers with active portfolios are already pricing in.
First, the marginal capex dollar in US utility-scale solar shifts from origination to interconnection acquisition. Greenfield acreage in decent solar states can still be bought at commodity land prices. A 500 MW to 1 GW interconnection agreement in an RTO with anything better than “incomplete” progress cannot. The valuation gap between the two is now widening in real time because the origination pipeline (safe-harbored modules, financed capital, negotiated offtakes) has capacity to absorb the acreage far in excess of what the transmission network can absorb.
Second, developers with existing interconnection rights on brownfield, retired-thermal, or repowered wind sites get a durable premium that does not compress if solar module ASPs drop. That premium sits above pure resource quality, which is already how site scoring has been framed since Order 2023 landed. What the AEU scorecard implies is that the premium is not going to compress even if module prices halve, because the throughput constraint is not module-supply-side.
Third, cluster-under-anchor structures like CO Bar become a template that only developers with pre-existing large-block interconnection rights can execute. Enlight had the queue slot before it had the debt. A new entrant working a 300 MW project in the standard MISO queue cannot copy the structure. The template exists but not everyone can print from it, and the count of developers who can is a specific and finite number.
The FEOC layer sits on top
Post-July 4 the interconnection question stacks against a second filter. IRS Notice 2026-15’s foreign-influenced entity provision hardened on the same date, and any license agreement a US storage developer enters into with a specified foreign entity after July 4 satisfies the effective-control prong of the FEOC test for Section 48E. Storage projects that begin construction in 2026 also have to hold a 55 percent material-assistance cost ratio, stepping up to 75 percent for 2029 starts.
Both filters apply to the same deal at the same time. CO Bar is a 4 GWh storage buyer inside a 1 GW AC interconnection envelope, and the cells that Enlight orders against that envelope have to clear FEOC and material-assistance. Cell orders that were assumed available a month ago are narrower now. Cell orders that clear the filter and can also be delivered against a 1 GW AC anchor are narrower still.
Which cell suppliers Enlight discloses on CO Bar 1, 2, and 3 (in construction) and on 4 and 5 (mobilizing 2H 2026) will be a read on the intersection of the two filters. The 45X production credit and 48E investment credit math for the tax equity slice depend on it. Every comparable project still working its capital stack against a comparable interconnection anchor will use the CO Bar disclosure as the comparable.
Signals to watch on the interconnection axis
Three things to track through H2 2026 and into 2027.
Large-block interconnection agreements issued in Arizona, Texas, and California, at 500 MW AC and above. Public agreements at that size are a finite list. The rate at which the list grows across CAISO, ERCOT, and SPP (the three RTOs closest to Arizona from a project-financing frame) is the physical proxy for the strength of the anchor-asset premium.
FERC action on the large-load track. Grid operators are negotiating a FERC track on large-load interconnection driven by data-center peak-load applications, which reached 198 GW in ERCOT alone in Q1 2026. The unresolved question is whether a 500 MW hyperscaler campus is treated as a generator, a load, or a hybrid, and whether large-load queue positions consume the same substation capacity as large-generator queue positions. The answer changes the supply of anchor-scale interconnection agreements available to solar-plus-storage developers.
Whether cluster-under-anchor structures show up outside the Southwest. CO Bar’s model works in Arizona because SRP and APS are willing counterparties, the solar resource is durable, and the transmission network into the load pocket is defined. The next place the model shows up (Texas, Nevada, New Mexico, and eventually the SPP tier of the Great Plains) is the read on whether Enlight found a local template or wrote a national one. Portfolio developers with US pipelines have every reason to run the same experiment. The frequency with which the experiment prints in the second half of 2026 is the tell.
The frame
Against the grid vertical thesis, this is thesis-confirm on the interconnection-as-binding-constraint frame first laid out in May, and thesis-extend on the safe-harbor-stockpile piece. Interconnection queue times were named as a constraint. Interconnection agreements at large-block scale are now visible as an asset class inside that constraint. The safe-harbored solar bank rests on that asset class, and the AEU scorecard says the asset class is not scaling with the bank on top of it. Portfolio-depth developers with pre-existing queue positions compound. Everyone else runs the standard queue.
The pricing signal is already in the ground at CO Bar. The question is how quickly the rest of the market rerates against it.