by David Riester and Paul Hildebrand
The H.R.1 reconciliation bill passed by the House on Thursday, May 22, would directly cause a staggering number of planned power plants to be canceled. Projects that would otherwise turn into power plants will die as a direct result of this legislation, if passed, causing severe and lasting damage to i) the electrical grid, ii) communities, and iii) industry.
Recognizing that the window to correct this remains wide open, the Segue team spent some time the past couple days analyzing and estimating this damage. We sought objective methods/sources wherever possible and aspired to set aside our biases (perhaps to the point of overcorrection). The second half of this document lays bare the methodology, for skeptics to decide for themselves how we did.
The Headline Numbers
- We estimate that 48% of planned solar, battery, and wind projects on a trajectory to turn on between now and 2033 will be canceled as a direct result of the amendment imposed by the Rules Committee with regard to sections 48E and 45Y[1]
- We estimate that 245 GW of currently planned power capacity would be canceled, representing $371 billion less investment in the US power grid. To put this into context, ~84% of the 60 GW total planned additions to the American electrical grid in 2025 are solar, wind, or energy storage power plants.[2] Those same energy sources represent 66% of the forecasted new generation added to the grid between 2026 and 2033, according to the EIA; canceling 48% of those power plants reduces growth of the US generating fleet by a third.
- At the same time, between now and the end of 2030, most sources[3] predict that load will increase by 125 – 150 GW, without accounting for replacing the capacity of existing power plants slated for retirement. Utilities and grid operators are already struggling to reconcile the two competing priorities i) facilitating this load growth (which is essential for regional and national economic growth) and ii) keeping consumer electricity costs from spiraling out of control. The only real solution is to build more generators as quickly as possible[4]; canceling one-third of planned generators does not help. The inescapable reality: to maintain a reliable, functional, affordable electricity grid and support the economic growth expected from the data center and advanced manufacturing boom, America requires many hundreds of GWs of newly built power plants; as is already the case, the vast majority of that planned added capacity would come from solar, wind, and storage.
- New natural gas builds can’t save us – they take 5-10 years to develop and construct and cost 2.5x what they used to. There are not nearly enough of them in queue for development and construction. Advancing in the interconnection queue to the stage where a turbine order is appropriate takes 2 – 4 years. Turbine lead times are currently 4 – 5 years[5]. Construction takes, on average, 3 years.[6] The latter two stages can overlap, but not the first two. Gas turbine lead times are an enormous issue; even projects with interconnection rights and state incentives secured are being canceled due to an inability to timely secure equipment.[7]
- And even if gas were a feasible replacement – which, again, it is not – replacing 245 GW of wind/solar/storage with natural gas would result in electricity prices from added generation over the next 10 years that is ~25 – 30% higher[8] than electricity from the currently planned fleet, by our rough estimates (which are generally in line with other analysis).[9]
- The canceled projects would represent a ~$371 billion reduction in economic activity in the form of construction and manufacturing work, at a time when hard-won growth in domestic manufacturing of both PV modules and grid-scale batteries is skyrocketing[10].
- The canceled projects would represent a loss of ~$79 billion in land lease payments to farmers and other landowners hosting the power plants.
- The canceled projects would represent a loss of ~$23 billion in property tax revenue to cities, counties, and school districts.
- Geographically, the effects of HR1 would disproportionately harm states that voted Trump in 2024: $234b (63%) of canceled investment in our model is in Trump-voting states; SEIA analysis finds the same pattern, and also captures the heavy red-state concentration of advanced solar and storage manufacturing.[11]
- H.R.1 is estimated to directly cause the loss of several hundred thousand American jobs. SEIA projects 330,000 lost jobs (solar and storage, but excluding wind)[12], and Energy Innovations projects 830,000[13], taking into account the entire bill. Our model doesn’t extend to employment effects, but these appear to be the most credible estimates available as of this post.
- Firms developing data center and advanced manufacturing plants in the US will be markedly set back by the proposed changes. The five highlighted rows above show five of the top ten states for forecasted datacenter buildout, according to Orennia. Many proposed datacenters would see load agreements canceled for lack of available power, which will result in canceled datacenters, lost jobs, and lost county/city revenue. You simply can not add an energy intensive data center to an area without either i) added power capacity, or ii) existing electricity users experiencing blackouts and price spikes. If there are no economically viable power plants built, then plans will be canceled, or residents and businesses will suffer. For those with an eye on the “AI Arms Race” with China, the impacts multiply rapidly to include existential threats and national competitiveness.
Conclusion
Following the passage of H.R.1 on Thursday morning, we set out to create a “ballpark estimate” of the damage America and the energy industry should expect if the Senate fails to avert this catastrophe. Given the sense of urgency, we view this as an evolving analysis, further refinements of which will be shared if materially different. We welcome suggestions for methodological improvements. Yet… even if key inputs are adjusted such that the impacts are softened, we are still looking at a severe power shortfall, economic impairment, spiking electricity prices, and a shackled manufacturing and artificial intelligence sector. We may never really understand why or how House members see this as a positive step for America, but we will find out soon enough if the Senate can avert a genuine power sector disaster that will set our nation back for decades.
Methodology
Queue Data
- Methodology: Aggregated actual real time data from ISOs and RTOs
- Sources: Data published by ISO/RTO and utility websites, aggregated and chance-of-success weighted by Orennia
Value of GW Canceled
- Methodology (PV): $1.64/Wac ($1.10/Wdc * 1.4 DC/AC ratio, plus $0.10/Wac interconnection upgrades)
- Methodology (Wind): $2.60/Wac ($2.50/Wac, plus $0.10/Wac interconnection upgrades).
- Methodology (BESS): $0.85/Wac ($250/kWh * 3-hr average duration, plus $0.10/Wac interconnection upgrades)
- Sources: Field data, Segue contracts, industry contacts. These numbers vary across geographies and technology but provide a reasonable ballpark number for turning thousands of canceled MWs into a rough sense of foregone investment
- Note: Capex does not include recurring revenues to host communities, such as property tax revenue, lease payments to landowners, and other community benefits, plus ancillary benefits such as employment from construction and operations/maintenance jobs and increased grid reliability.
Generator additions
- Source: EIA Energy Outlook 2025
- Note: There are several other sources available here; in addition these forecasts are also subject to large swings based on other present policy concerns such as Federal support/opposition to offshore wind and a highly uncertain tariff environment affecting imported power equipment
Near-term Load Growth
- Source: “Infrastructurereportcard.org” (see earlier footnote)
- Note: There are dozens of journals and studies that report expected load growth. Many of them are part of articles and reports centered on AI and data center growth. Many estimates track back to the same two estimates, referenced herein.
Natural Gas Timelines and CapEx
- Source: We gathered about 10 datapoints here with all ranging between 5 – 10 years and 7 years being the most common. Two of the sources are cited in footnotes above.
- Note: We use a basic assumption of $2.50/W. This is a commonly referenced “back of envelope” estimate, which is also in line with Lazard LCOE (v17.0) estimates.
Electricity Price Deltas
- Source: We used a composite of the Lazard LCOE (v17.0) and the EIA LCOE and then adjusted further for tax credits
Lease Revenue
- Note: All lease revenues represent the sum total across the operating life of the system
- Methodology (PV): $1,000/acre/year, 2% escalator, 6 acres per MWac, 40 year useful life = $362k of landowner revenue per MWac built.
- Methodology (Wind): Similar assumptions to PV above except assuming 15 acres per MWac, = $906k of landowner revenue per MWac built.
- Methodology (BESS): $2,000/acre/year, 2% escalator, 0.1 acres per MW (assuming 3-hr average duration), 25 year useful life = $6k of landowner revenue per MWac built.
- Source: Segue portfolio + field data, in very round numbers. Lease rates vary widely across geographies, as do acres required for power plant
Property Tax Revenue
- Methodology: We assume personal property taxes at 0.30% of system capex per year, and real property taxes at 0.05% of system capex per year. 5% annual decrease in assessed value over time and 40–year useful life.
- Source: Composite estimates taken from a cross-section of the ~275 projects in Segue’s two fund portfolios. Note that these vary widely from region to region, township to township, and technology to technology.
Geography of Expected Cancellations
- Note: We did not adjust our model for higher or lower cancellation rates by states; we applied similar cancellation rates nationwide by project interconnection status and PIS year
Estimated Canceled GWs:
- We started with raw interconnection queue data, as aggregated by Orennia and downloaded on May 24, 2025. Interconnection queue data (i.e. projects that have applied to the relevant ISO/RTO to secure the right to interconnect to the grid, triggering a multi-year process) is the best public data set available for planned generators.
- Then parsed this data by i) stage (Pre-study, Studies Ongoing, IA Executed, In Construction) and ii) PIS date, using Orennia’s classification methodology. Of note: based on observed historical patterns between ISO data on expected PIS year and actual PIS years, Orennia adjusts PIS dates generally “to the right” of the raw ISO reported data. Even with that calibration, Segue believes the PIS dates are systemically optimistic, as most project delays are difficult for ISO/RTOs and utilities to foresee and capture in their queue data.
- Next, we culled down the raw queue data by applying Orennia’s “Chance of Success” metric, which uses historical patterns and go-forward supply-demand forecasts to estimate project attrition arising from permitting, interconnection costs/issues, economic viability, environmental issues, etc. In other words, this helps cull the queue down to something that meets the overall needs of the grid, rather than taking every single project at face value.
- After doing so, we have the “Business as Usual” scenario – presently-queued projects, discounted at a reasonable rate. Nothing related to H.R.1 is in the analysis yet.
- Then, we applied “H.R.1 Survival” multipliers to each project based on (a) PIS year and (b) project interconnection status, shown here:
These are meant to capture a few things:
Timeline impacts on possible project cancellation:
- The new ITC requirement that a project commence construction within 60 days of the bill becoming law causes a sharp dropoff in projects coming online after 2026. The closer to 2028 PIS, the higher the likelihood it’s canceled. This may seem counterintuitive, but the further a project is from its PIS, the more development work (risk) that remains (on average), reducing the likelihood that a developer will incur the costs/risks necessary to start construction in the 60–day period. Moreover, the longer a developer must stretch out – and carry the costs of – construction, the less likely it is that the project will remain economically viable. Finally, there are basic operational constraints when one considers an entire industry looking to enter into the same contracts (purchase orders, master supply agreements) for the same products (transformers, modules, breakers, switchgear) within a 60–day window. Our analysis assumes the power sector can process about 5x the normal volume, but not more.
- For the years falling after the proposed ITC sunset (2029+), there are three factors to consider:
- The first is the simple fact that a 30% subsidy falls away.
- Working in the other direction, the further into the future, the more likely it is that the added costs from the loss of the ITC can be “pushed through” to the power plant’s customers.
- Finally, solar and BESS cost curves come down over time, so we would expect less attrition as we move further into the future, though on an absolute basis we would still expect less capacity built than in the “business as usual” scenario.
Interconnection status
- H.R.1–related project attrition will be slightly worse for projects that are currently undergoing interconnection studies, or are pre-study. There are a couple reasons for this:
- Projects with signed interconnection agreements are more likely to procure (or have procured) equipment to safe harbor “start of construction”, due to project maturity and remaining risk.
- Developers will face a capital shortage and need to recommit limited capital to safe harboring and ensuring the success of later-stage, de-risked projects, abandoning earlier-stage projects. We expect large withdrawals from queue cycles given large interconnection costs and the need to triage liquidity.
[1] Of note, this analysis ignores the additional impacts we could expect from other proposed changes to existing energy policy, of which the elimination of transferability would be most pronounced.
[2] EIA: https://www.eia.gov/todayinenergy/detail.php?id=64586#:~:text=Solar%2C%20battery%20storage%20to%20lead,U.S.%20Energy%20Information%20Administration%20(EIA).
[3] E.g. https://infrastructurereportcard.org/cat-item/energy-infrastructure/#:~:text=Energy%20Generation&Data%20centers%20alone%20are%20expected,growing%20roughly%2010%25%20per%20year.
[4] With all due respect and optimism to grid enhancement technologies and sophisticated demand management, they are palliatives, not cures.
[5] https://rbnenergy.com/i-will-wait-backlog-for-natural-gas-turbines-expands-on-surging-demand-supply-constraints
[7] A deep dive into the status of the Texas Energy Fund projects is a sobering look at how goals of near-term gas deployment are being stymied by the realities of complex supply chains and lead times. https://www.douglewin.com/p/two-more-texas-energy-fund-projects?utm_source=publication-search
[8] 15-35% depending on the source and methodology. Lazard’s LCOE is generally considered the gold standard. Using LCOE and then adjusting for the availability of the tax credits (on both sides) returns a nationwide, annualized estimate of 30%. Lazard. (2024). Levelized Cost of Energy Analysis — Version 17.0. U.S. Energy Information Administration. (2024). Annual Energy Outlook 2025, Levelized Costs of New Generation Resources.
[9] Rocky Mountain Institute’s "The Economics of Clean Energy Portfolios" and Bloomberg NEF’s summary report of their 2025 LCOE model are the best two examples
