The comment period for comments regarding the safety valve provision in New York Public Service Law (PSL) § 66-p for the Renewable Energy Program in the Climate Leadership & Community Protection Act (Climate Act or CLCPA) recently ended. I published a series of posts describing this process that was summarized in my most recent post. Independent Intervenors Richard Ellenbogen, Constantine Kontogiannis, Francis Menton, and myself explained why we thought it would be appropriate to hold a hearing addressing the safety valve provisions in our filing. We agreed that we would not submit technical arguments from all of us but we are all driven by a common motivator that when we see something outlandish, we have to respond. Richard Ellenbogen’s filing describing an Artificial Intelligence (AI) analysis impact of the Climate Act on utility rates was initiated by such a trigger.
Richard Ellenbogen has been speaking to lawmakers, policy makers and regulators since 2019 regarding the deficiencies inherent in New York State (NYS) energy policy. He has a proven record implementing carbon reduction programs at his own manufacturing business in Westchester County where it has reduced its electric utility load by 80% while reducing its carbon footprint by 30% – 40% below that of the downstate system. I have previously published other articles by Ellenbogen including a summary description of his issues with the Climate Act.
Background
New York Public Service Law (PSL) § 66-p establishes a renewable energy program for the Climate Act. It provides that the Commission “may temporarily suspend or modify the obligations under such program provided that the commission, after conducting a hearing as provided in section twenty of this chapter, makes a finding that the program impedes the provision of safe and adequate electric service; the program is likely to impair existing obligations and agreements; and/or that there is a significant increase in arrears or service disconnections that the commission determines is related to the program”. This safety valve was included because even the lawmakers realized that it may not be possible to transition the electric system to “zero-emissions” by relying on renewable energy.
In response to the PSL 66-P safety valve, Ellenbogen filed a description of his analysis of the results of other jurisdiction’s attempts to implement energy policies like the Climate Act. His filing was titled: “An AI Analysis of the CLCPA that Shows it will raise Utility rates between 50% and 150% over the Next 40 years with no rate relief appearing for between 100 – 150 years, IF EVER” and was described:
“Prejudices Inherent in the drafting of the CLCPA Introduced a rigid framework, unsupported by scientific facts, that preclude better long-term solutions from both a carbon emissions standpoint and an economic frame of reference”. He also submitted a spreadsheet that was developed by Perplexity in response to his queries.
This post is based on the filing and his email description circulated to his mailing list. I have lightly edited his work for consistency.
AI Analysis Background
Both Ellenbogen and I have been using Perplexity AI more as a research tool and we are just starting to realize the potential of its use. He explains how he used it for this analysis:
For the past three weeks, i have been sitting and asking energy related questions of Perplexity AI. The questions were asked without bias and conditions were only added based upon responses from prior questions, although as I learned over time, the AI algorithm remembered all of the conditions without my asking and applied those to subsequent questions.
The paper covers everything that I could think of. It has AI answers to the major problems facing NY State. I added context to the answers to help people understand them and explain how we got here.
The questions were asked in plain English. Some of the AI responses can be technical in places.
The first 23 pages tell most of the story with the 50 pages of appendices adding context. AI has become a frighteningly powerful tool. Watching it generate code to answer some of the questions was comparable to what an incredibly fast human would do that could access information at a pace that no human could possibly do, let alone think of all the parameters needed. When I was building the spreadsheet, the second query was to add BESS to the calculation to make the wind dispatchable so that it would be more of an “apples to apples” comparison to the retooled gas plants. I didn’t specify how much but the AI algorithm automatically researched it and found NYISO requirements and pulled those into the calculation. It also was telling me what it was doing as it did it.
An experienced utility analyst looked at the spreadsheet and said, “That’s exactly what I used to do for the utility companies, except that would take at least a week.” It took me an hour. It was built from three queries and the AI wrote code, produced results, checked conformity, found its errors, redid the code and checked conformity again until the results made numerical sense. It took about 20 minutes for each question. It then added the additional questions to the same spreadsheet and suggested more related questions to research. It could turn into an infinitely long research rabbit hole. When I asked about natural gas prices, it had remembered my questions about the LI gas plants and automatically related the pricing questions to those without my asking. It inferred that was why I was asking.
I spent time checking the answers which accounted for much of the time writing the document. I knew the answers to many of the questions before I asked them based upon extensive reading and almost fifty years of experience dealing with energy systems. I am hoping that AI confirmation of the answers will make people realize that this is not ideological.
On a final note, I’m a tech geek and this technology is very useful but a little creepy. When I was doing this, I felt like I was dealing with “Skynet”. The rate of evolution of the technology is breathtaking, and I don’t think that I could have generated this report six months ago.
Ellenbogen’s Filing
Consistent with Ellenbogen’s use of AI, I submitted the following question to Perplexity: “Provide an overview summary of Ellenbogen’s filing that I can incorporate into a blog post. “ The following is the response:
Ellenbogen’s filing argues that New York’s CLCPA will drive very large, long‑lasting rate increases and will not deliver genuine rate relief for at least 100–150 years, if ever, under anything resembling realistic assumptions. He frames CLCPA as a capital‑heavy, prejudiced policy design that blocks cheaper and faster emission‑reduction options (notably repowered gas and nuclear), locking customers into a multi‑generation cost burden for modest climate benefits.
A central plank is a detailed Australia comparison, which he uses as a “best possible” renewable case study to bound what New York can realistically expect. Australia has six times New York’s solar capacity, far higher solar capacity factors, far more land at a fraction of New York’s cost, and far lower heating loads, yet wholesale prices only clearly began trending down in 2024–2025 after roughly 20 years of policy implementation, and even then retail bills are only expected to fall about 5 percent over the next five years.
He then scales this experience to New York’s much worse fundamentals: four times the per‑capita heating load, roughly half the solar capacity factor, far less developable land, and much higher land and interconnection costs. On that basis, he argues New York would need 80–150 years before seeing durable rate relief, with residential rates rising on the order of 50–150 percent over the next 30–40 years and upstate households facing bill increases of 2,000–5,000 dollars per year at the peak of the transition.
Grid, labor, and supply‑chain constraints
The filing emphasizes that official NYSERDA/CLCPA planning assumes “ideal” supply chains and grid conditions that do not exist. He cites transformer shortages with multi‑year lead times, insufficient trained labor to electrify buildings and upgrade panels and substations, and local grid stress already appearing as “blue sky blackouts” in electrified pockets like the Albany area.
Because load from electrification is ramping faster than new dispatchable capacity and network reinforcements, he argues New York is replicating Victoria’s mistakes in Australia: voltage problems, constrained substations, and reliability risks driven by “electrify first, supply later” policy sequencing. His bottom line is that these constraints stretch the transition into a 40‑year or longer construction effort, add huge cost premiums, and push any hypothetical “payoff” far into the next century.
Cost comparisons: offshore wind, storage, and gas repowering
A major section uses an Excel LCOE framework, fed by AI‑generated parameter sets, to compare offshore wind plus batteries against repowered combined‑cycle gas at Long Island sites like Northport, Barrett, and Port Jefferson. Across a wide range of gas prices (current Henry Hub and 2–4× that level), the model shows that offshore wind plus 4–8 hours of battery storage remains substantially more expensive than repowered gas over 20‑ and 40‑year horizons, even after layering in a state carbon tax on gas.
He also notes that the wind‑plus‑storage cost stack in his base case does not include synchronous condensers and other stability equipment required in a heavily inverter‑based system, which would drive the renewable system cost even higher relative to combined‑cycle plants. In his framing, the oft‑repeated “renewables are cheaper” line collapses once you include the full system cost of firming, transmission, and stability equipment in a New York context, as opposed to Texas or Australia with far better wind and solar resource.
Nuclear and statute‑driven “prejudice”
Ellenbogen then asks, via AI prompts, what an unconstrained least‑cost, low‑carbon system for New York would look like over a 60‑year horizon. When the CLCPA’s statutory renewable carve‑outs and timelines are removed, the response (which he quotes extensively) favors a nuclear‑heavy system, with long‑life reactors providing 60–100 years of output at high capacity factors and lower firm system LCOE than “firmed renewables” once overbuild, storage cycling, replacements, and grid expansions are included.
He uses that to argue that the CLCPA embeds an explicit prejudice against nuclear and gas repowering by design, because it hard‑codes high penetrations of wind and solar irrespective of system cost, reliability, or practical feasibility. In his conclusions, he characterizes the CLCPA as an ideologically driven statute that overrides physics and engineering judgment, and he urges the PSC to invoke its PSL‑66 powers to pivot toward a combination of repowered gas as an interim 40–60‑year bridge and nuclear as the primary long‑term zero‑carbon backbone, with hydrogen‑fueled peakers using excess nuclear output as a complement.
Discussion
Ellenbogen’s email concludes that the analysis shows a very grim future for NY State if there is not a course correction. He notes:
Hopefully, an impartial question and answer session with the world’s computer systems will alter the discussion and make people realize that if we proceed down this path, we will be increasing emissions while needlessly increasing costs in search of a holy grail that is not feasibly attainable in NY State’s climate. As I have written on several occasions, the CLCPA was a college thought experiment about what could be done if there were no limitations on money, labor, supply chains, land, and if there was no public resistance to having large renewable installations located near their homes. Unfortunately, our state government, and particularly Andrew Cuomo, turned that thought experiment into public policy. While the authors of the thought experiment considered it attainable, engineers in the energy industry understood that the concept was preposterous in NYS.
Conclusion
I have a great deal of respect for Ellenbogen because he has personally approached the challenge of decarbonization as an experiment. His home and manufacturing facility were the subjects and included data monitoring systems for the test. He knows what will work and his opposition to the Climate Act technology transition plan is based on what his experiment shows will not work.
His use of AI is based on his experience and the results are consistent with what he has learned. I recommend his entire analysis for the unconvinced. This analysis is another great example why New York should conduct a PSL 66-P hearing to determine if renewable energy can provide safe, adequate, and affordable electricity in New York.
Pragmatic Environmentalist of New York 