The July 4, 2026 begin-construction deadline closes in 15 days, and the headline number coming out of it (Wood Mackenzie’s 216 to 240 GWdc of safe-harbored US utility-scale solar between mid-2024 and the deadline, reported through SolarQuarter) is being read as proof that the OBBBA cliff is, for solar, already discounted into the asset side of the equation. The reading is half right. The credit eligibility on that pool is, in practice, locked. What is not locked is the four-year continuity window’s other leg: a safe-harbored project still has to energize to monetize, and the interconnection queue is the bottleneck that decides whether the stockpile becomes deliverable capacity or shelf inventory. The right way to read the post-July 4 market is as a five-year forward contract written against the queue, and the queue’s clearing rate is the variable that determines what fraction of 240 GW lands on the grid by 2030.
What the safe-harbor stockpile actually locks
Begin with what the credit clock secures and what it does not. A solar project that establishes begin-of-construction by July 4, 2026 under the Section 48E or 45Y framework preserves full credit eligibility plus a four-year continuity-of-construction window to reach commercial operation. The June 6, 2026 District Court ruling that vacated IRS Notice 2025-42 restored the 5 percent cost safe-harbor as an alternative pathway to physical-work, widening the qualifying funnel for late-stage developers. Treasury has not appealed yet, and practitioners are still recommending continuous-work documentation as a belt-and-suspenders backstop. (See: news/2026-06-16-court-vacates-irs-notice-2025-42 and news/2026-06-11-storage-itc-safe-harbor.) Projects that miss July 4 fall back to a December 31, 2027 placed-in-service deadline, which is a much tighter bar given current EPC and interconnection lead times.
The continuity window is the part that matters for the deliverability question. A safe-harbored project does not have to be in service in 2026 or 2027. It has to be in service by mid-2030 (a four-year window from the begin-construction date, with Treasury continuity rules allowing extensions in narrow circumstances). That window is what gives 240 GW of locked-credit inventory the room to wait out interconnection studies, transmission upgrades, and supply-chain timing. The window is also what makes the queue the binding constraint, because the credit clock does not require energization by 2027 the way the non-safe-harbored pool does. The queue is what decides whether the inventory ships.
What the queue can actually clear
Stack the safe-harbored stockpile against the RTO-by-RTO interconnection throughput data and the deliverability picture sharpens.
In PJM, the queue reform under FERC’s June 13 expedited interconnection order (see: news/2026-06-13-ferc-pjm-expedited-interconnection-track) creates a fast lane for projects that can demonstrate readiness, but the underlying cluster-study capacity remains the constraint. PJM cleared roughly 11 GW of new generation in the 2024 and 2025 study cycles combined, against a queue backlog that exceeded 200 GW entering 2026. The 2026 cluster opened in February and is structured to complete system-impact and facility-study work on a 24-month timeline. Realistically, PJM can clear 15 to 25 GW per year of new capacity through 2030 if the expedited track operates as designed, with storage and load-flexible resources taking a meaningful share of that throughput. Solar’s share of PJM cluster awards has historically been in the 40 to 55 percent range, which puts the solar-specific PJM clearing capacity in roughly the 6 to 14 GW per year band through 2030.
ERCOT is structurally different because it does not gate generation on the same cluster-study pace, but the binding constraint has rotated to large-load interconnection. The June 17 reporting on the ERCOT large-load queue at 226 GW (see: news/2026-06-17-ercot-large-load-queue-226gw) is the other side of the same coin. ERCOT’s generator-interconnection throughput is closer to 25 to 35 GW per year of new capacity, with solar historically taking 40 to 60 percent of new connections. The 2026 picture is more constrained because the transmission-planning queue under the May 2026 ERCOT Regional Transmission Plan update is absorbing the large-load surge alongside generation, and the network upgrade timing for both can stretch to 36 to 48 months. ERCOT can probably move 10 to 18 GW per year of new solar through the queue through 2030, with the upside band requiring the transmission-build assumptions to hold.
MISO is the third major pool. The MISO Order 2023 cluster studies are running on the two-year cycle, and the regional-transmission planning under the Order 1920 first-cycle filings (see: news/2026-06-06-miso-order-1920-june-12-data-center-load) is the gating factor on how much of MISO’s roughly 180 GW queue can clear. MISO’s recent throughput has been 8 to 12 GW per year of new capacity awards, with solar taking 35 to 50 percent. A reasonable forward read is 4 to 8 GW per year of solar clearance through 2030.
CAISO, SPP, and the non-RTO West each clear smaller volumes. Adding them in conservatively, the combined non-PJM/ERCOT/MISO US clearance rate is in the 5 to 10 GW per year band for solar.
Stack these together. The aggregate US interconnection clearance rate for solar through 2030 is roughly 25 to 50 GW per year, with the upside band requiring transmission upgrades to execute on schedule and the lower bound being closer to where the historical run-rate sits. Across four years (mid-2026 through mid-2030), the cumulative deliverable solar pool is approximately 100 to 200 GW. The safe-harbored stockpile is 216 to 240 GWdc.
That is the math worth holding. The credit-locked inventory exceeds the deliverable capacity over the same window by 20 to 100 GW, depending on which RTO scenarios hold.
The binding constraint has rotated
Pre-OBBBA, the developer’s question was sequencing: when do I begin construction, how do I document continuity, can I lock the credit. Post-OBBBA, with the stockpile in hand, that question is settled for 216 to 240 GW of pipeline. The new question is queue position. The developers that get paid in 2027 to 2030 are the ones whose safe-harbored projects are sitting at the front of cluster studies, have signed interconnection agreements (or large generator interconnection agreements in ERCOT parlance), and have started network-upgrade work. Developers that safe-harbored cell and module inventory but did not advance the corresponding interconnection paperwork are holding credit eligibility on assets that cannot reach commercial operation inside the four-year window.
Three signals from the last sixty days confirm the rotation. First, the secondary-market pricing for late-stage interconnection positions has firmed. Trade-press reports through May and June show queue-position transfers (where they are permitted) trading at 3 to 7 percent premiums above the underlying project NPV, and that premium has roughly doubled since Q4 2025. Second, the developer M&A flow has shifted from cell and panel supply targets to project-portfolio targets with documented interconnection progress. The T1 Energy-KORE Power acquisition (see: news/2026-06-07-t1-energy-kore-power-bess-acquisition) is the early-stage version of the same pattern. Third, the hyperscaler procurement teams are signing 7- to 10-year PPAs against named projects with disclosed interconnection milestones, not against generic pipelines. That contract structure is the buyer side voting that the queue is the binding constraint.
What this means for H2 2026 deal flow
Four reads for developers, lenders, and offtakers.
First, the equity premium for queue-cleared safe-harbored projects widens through 2026 and 2027. A project with a signed interconnection agreement, a safe-harbored credit position, and network-upgrade scope locked in is the cleanest asset on the market, and the bid-to-asks on those projects already reflect that. Lenders that want low-risk solar exposure are concentrating into this pool, which compresses the equity yield. Developers without queue-position progress on their safe-harbored inventory are facing a structurally less liquid bid.
Second, the 2027 placed-in-service backstop pool (projects that did not safe-harbor by July 4) is going to underdeliver in absolute GW terms. The headline cliff is the credit-eligibility loss for solar above 1.5 MW, but the operational cliff is that almost no project that missed July 4 has a realistic path to a December 31, 2027 in-service date through the current cluster-study cadence. That pool’s contribution to 2027 and 2028 installations is going to round to single-digit gigawatts, regardless of how aggressive the developer financing is. The safe-harbored stockpile is the real 2027 to 2030 pipeline.
Third, the 240 GW stockpile is a forward demand contract for FEOC-compliant modules, inverters, trackers, and racking. Domestic and allied manufacturers with capacity coming online in 2026 to 2028 are selling into a pool of projects that already have their tax treatment settled. That offtake setup is structurally cleaner than the pre-OBBBA environment, but it is also concentrated, because the deliverable subset of the stockpile (the 100 to 200 GW that actually energizes) is the part that turns into module orders. The 40 to 100 GW gap between credit-locked inventory and deliverable inventory is the cushion that, if the queue clears at the upside, becomes incremental demand, and if the queue clears at the downside, becomes write-down risk for the safe-harbored holders.
Fourth, the storage attachment math gets more aggressive. With solar interconnection capacity as the binding constraint, the marginal MWh on the grid increasingly comes from storage attached to existing or queue-advanced solar projects, not from new solar interconnections. The H2 2026 BESS deal flow (see: posts/domestic-content-bess-cell-supply-curve-h2-2026) is the supply-side mirror image of this constraint. Developers are pulling forward storage scope on solar projects that have queue position because the storage augments revenue without requiring new generator-interconnection capacity.
What to watch into 2027
Five datapoints over the next eighteen months will calibrate the deliverability curve.
The July Q2 10-Q cycle from the major utility-scale developers. The disclosures worth reading are not the safe-harbor inventory totals (those will be at or near the 216 to 240 GW industry estimate, allocated across the public-developer cohort) but the percentage of safe-harbored inventory with executed interconnection agreements. That percentage is the deliverability number.
The FERC Order 2023 cluster-study completion rates by RTO through the 2026 and 2027 study cycles. The Order 2023 framework was designed to move queues faster by clustering applications, requiring more financial commitment, and reducing late-stage withdrawals. Two years in, the throughput data has been mixed. (See: posts/ferc-order-2023-cluster-studies-two-years-in.) If the 2026 cluster studies complete on schedule, the upside-band clearance numbers above are achievable. If they slip, the deliverable pool compresses toward the lower band.
The PJM 2028/2029 BRA cleared composition (closing July 7). The capacity-payment number is largely priced, but the megawatt count of solar that cleared with documented interconnection-agreement status is the read on how much of the PJM safe-harbored solar pool is positioned to energize within the four-year window. (See: posts/pjm-2028-2029-bra-price-collar-watchlist.)
The ERCOT transmission-build progress through 2026 and 2027. The 226 GW large-load queue creates demand for the same transmission infrastructure that new solar generation needs. ERCOT’s ability to permit, finance, and energize the planned 345 kV upgrades on the timelines disclosed in the May 2026 RTP update is the deliverability constraint for both pools. Slippage in the transmission build is the dominant downside risk to the ERCOT clearance numbers above.
The first announced safe-harbor inventory transfer or sale-leaseback. The 40 to 100 GW gap between credit-locked and deliverable inventory creates a structural incentive for safe-harbored holders without queue positions to monetize the credit value through sale to holders with queue positions. The first deal of this kind will set the discount rate, and the discount rate is a direct read on what the market thinks the queue can actually clear.
The thesis stays simple. The 240 GW number is real, the credit eligibility on it is real, and the four-year continuity window is real. The deliverability is the part that is not yet settled, and the queue is what settles it. Pre-OBBBA the credit clock was the binding constraint. Post-OBBBA, for the safe-harbored pool, it is the interconnection queue. The H2 2026 deal flow is going to reward whoever read the rotation early.