Posts

On the drive to do something that matters and roads not (yet) taken Challenges to widespread deployment: Stuff that’s cheaper to operate after you buy it is usually more expensive to buy in the first place. Clean, efficient cooling equipment is no different. Properly aligning incentives and allocating risk to avoid resistance to higher upfront costs is critical (CaaS), as is quickly iterating and scaling production of innovative technological solutions to drive down those costs over time until mass adoption can be achieved simply by offering consumers and businesses a better product for cheaper (Gradient, Merino). Final word: More people, higher incomes, hotter weather -> blowout cooling demand. Huge opportunity to develop and deploy better, cleaner solutions; huge risk to the climate and to daily quality of life in hot places if we don’t get it right. ----- [5] The UN’s sustainable energy program. [6] I.e. the cost of electricity used to run the unit plus any maintenance costs required for servicing/repairs. [7] No, your dog is not a secret Einstein. Although some dogs are smart in other ways. Stay thirsty, my friends. Long-Duration Energy Storage (LDES)This pick could be considered cheating, to the extent that you consider any form of energy storage as fitting in the “utility-scale renewables” bucket that I focus on at Cypress Creek. But it’s really not – it’s more of a renewables-enabling technology – and anyway, economically viable LDES has not yet been deployed at utility scale the way that 2- and 4-hour lithium-ion batteries have [8]. To get electrons on demand, you need to convert electrical current to a different form of energy that’s, let's say, "shelf-stable." For example, true batteries (in the electrochemical sense) literally charge when you send electrical current into them, with electrons flowing to the anode as electrical energy is converted to chemical energy. The main drawbacks to this technology as currently implemented in grid-scale lithium-ion batteries are: 1.     You only get so many charge-discharge cycles before stuff breaks down and gunk builds up on the battery terminals (can you tell I’m not an engineer?). This compromises roundtrip efficiency as electrons shuttle back and forth over and over between the anode and the cathode. 2.      Lithium-ion batteries are prone to thermal runaway, when the battery temperature gets uncontrollably high due to an internal short circuit or an overcharge, the risk of which is also increased by charge-discharge “chemical wear and tear” over time (see gunk buildup item above). 3.      Lithium-ion batteries are fairly expensive to produce because they contain cobalt, nickel, and copper, all of which are currently scarce relative to demand. One way to get past this is to make an electrochemical battery from different materials. Sodium-ion batteries like what Peak Energy develops don’t need scarce, expensive metals like cobalt, nickel, and lithium, and are therefore expected to carry a substantially lower capital cost than lithium-ion once firms can stand up more robust sodium-ion supply chains and manufacturing capacity to enable exploit economies of scale. Shoutout to my friend and business school classmate Yutong, who's advancing Peak's efforts to achieve this. The cheaper your materials, the easier it is to build a really big battery relative to the size of your electrical connection, which means the battery has a long duration (the LD in LDES) – it can discharge at max output for a long time because, to use a fossil-fuel analogy, it’s like a car with a huge gas tank and a small, efficient engine. Another way to do LDES is to drop the whole electrochemical thing altogether and instead store heat, which is what the folks at Antora Energy (including my good buddy Raghavendra!) are doing: You use electricity to superheat a big lump of something, then recapture that heat when you want to use it. This approach’s key advantage over chemical batteries is that the materials are cheaper: The graphite that Antora uses is cheap and abundant, even relative to the less-scarce metals used for sodium-ion batteries. Again, that means less spend per unit duration of battery, unlocking batteries that can discharge for double-digit hours vs. today’s 2-4-hour grid-scale lithium-ion batteries. It’s also simpler to store energy in the form of heat (just build really good insulation around your big-ass superheated graphite block!) than to store it in the form of an electrically charged chemical solution (better keep all those ions contained!). The thermal battery comes with less capex, no moving parts, and generally better operational safety. The trade-off is that it’s really hard to efficiently convert heat back into electricity. One possible solution that Antora is exploring, which I saw when I visited their HQ in 2023: Thermophotovoltaics (TPV), which are like solar panels but use heat instead of light to make electricity. TPV is a technology that should work in theory but requires a lot of R&D to get right. The other answer, perhaps the obvious one: Just hook it up to a steam turbine, as with natural gas, coal, and nuclear power plants. Sure, now we have moving parts back in the equation, but steam turbines are still safer and more reliable than ion batteries because they are exciting electrons by using hot water to spin something around rather than by facilitating an electrochemical process. Market(s): Industrial heat consumers (near-term, for thermal batteries only); wholesale electricity markets (long-term with policy/market design support). Why it’s exciting: Cheap materials storing zero-marginal-cost electricity means we can meet more energy end users’ needs with clean electrons from solar and wind, bridging the gap between consumers who need energy on demand and renewable generation resources whose supply of energy varies with the weather and/or the time of day. Business models: Behind-the-meter/co-located LDES can reduce C&I facilities’ demand charge exposure by reducing or eliminating net grid imports during high-demand (therefore high-priced) periods and can enhance resilience by unlocking day-plus islanding capabilities that dramatically reduce the damage done to both facility throughout and critical machines and equipment by sudden grid outages or voltage fluctuations. In order to expand beyond this onsite demand management and resilience use case and capture the technology’s full value in the form of grid-connected LDES, the wholesale power markets (ISOs and RTOs) will need to wrap their heads around the value that a shorter-term dispatchable capacity resource could bring to the grid (see the next section). Once such a short-term capacity or week-ahead energy product exists, LDES will be the natural choice to supply that product to the market at the lowest possible all-in cost. Challenges to widespread deployment: Where’s the revenue mechanism? Wholesale power markets typically consist of two separate power markets: A short-term market for day-ahead and real-time energy, and a long-term market for capacity [9]. Let’s imagine a 72-hour LDES battery (chemical or otherwise) with 67% roundtrip efficiency: We’d need to charge it for three days in order to provide two days of discharge. Does this operational profile make economic sense in either of the existing power markets? Not really: • To make money from energy markets, storage assets must charge when energy is cheap and discharge when it’s expensive (“buy low, sell high”). But if your battery takes three days to fully charge, as a 72-hour LDES would, you need to receive a discharge price signal three days in advance to optimize charge-discharge decision-making. The existing day-ahead and real-time market constructs don’t provide this kind of multi-day visibility to LDES assets, and no one is going to underwrite an energy arbitrage revenue strategy that relies on guesswork. • To make money from capacity markets, storage assets must bid into forward capacity auctions, committing to provide capacity in peak demand hours during a future period. Again, this timeline is misaligned with the LDES operational envelope, as an unexpected peak event requiring capacity resource performance could catch LDES assets in a compromised position if they have not received sufficient advance notice to charge up. This is a critical weakness of storage technologies relative to dispatchable thermal resources like natural gas or coal: Unlike storage, thermal generation can switch on to fulfill its capacity commitments with very little notice, so long as it has sufficient fuel available (usually via long-term, fixed-price supply agreements). By contrast, unpredictable system supply-demand conditions over the course of a yearlong capacity commitment period means LDES assets risk being caught without sufficient charge to meet their peak demand obligations, resulting in forfeiture of capacity payments and potentially additional financial penalties. And yet, it’s apparent that LDES does offer something valuable to the grid: Flexibility and resilience. Let’s play this out: On Sunday, the ten-day forecast shows that from next Thursday through the following Monday, a slow-moving snowstorm will settle down in Texas and the Great Plains states, and stay there. Winds will be minimal and clouds will block the sun. ERCOT, the grid serving most of Texas’s electricity demand, gets almost a quarter of its electricity supply from wind and another 14% from solar, while SPP, which serves the western Great Plains, has barely any solar but leans heavily on wind (38%). Without generation from variable renewables, these regions will pay much more for stopgap electricity generated by high-marginal-cost gas peaker plants – and may even be subject to blackouts if those peakers can’t perform in the cold, as occurred during Winter Storm Uri in 2021. But if these markets are forward-thinking and implement a rolling five- or ten-day-ahead energy market, LDES systems that are cheaper to build and operate than thermal units can respond to a longer-lead price signal and begin charging with cheap renewable electricity that’s available from now until the storm rolls in, knowing they’ve pre-sold that energy via the advance market at a premium to their charging cost while still providing a cheaper source of supply than natural gas could have. And, as of 2019, 5-day forecasts have become as accurate as 1-day forecasts were in 1980. With medium-term weather forecast quality continuing to improve, a market for energy on the scale of days to weeks in advance is now both achievable and a great idea. Such a market would incentivize deployment of LDES resources that could: 1.      Backstop a growing base of cheap, variable renewable generation against extended periods of low insolation and/or low wind speeds. 2.      Proactively charge in advance of extreme weather events to ensure grid resilience on a local basis in the event that regional generation or transmission resources go offline due to lightning strikes, flooding, downed power lines, frozen fuel supplies, or other weather-driven damage or obstruction. 3.      Enable long-term generation-shifting to ensure that the output of the lowest-cost generation resources can be matched with the periods of highest demand, even if these do not occur within a few hours or even days of each other. Final word: Right now, you can have energy that’s cheap or energy that’s dispatchable, with few options to electrons that are both cheap and on-demand. LDES unlocks huge economic value and emissions reduction potential by taking already-cheap renewable energy and making it dispatchable. Big things ahead. ----- [8] …Although Form (using iron-air batteries, another chemistry that's cheaper but less energy-dense than lithium-ion) does seem to be close to achieving fully commercialized utility-scale LDES deployment. Good for them! [9] Here’s a brief refresher on the difference between capacity and energy, paraphrasing a previous blog: Energy consumption is the area under the curve, in kilo/mega/giga/terawatt-hours, while load or capacity is instantaneous power draw, in kilo/mega/giga/terawatts. For example, a gigawatt is a rate of power consumption at any given point in time, and a terawatt-hour is a quantity of energy consumed, specifically the amount consumed when you power one gigawatt of load for one thousand hours. I’ll publish a short addendum post next week that provides a more detailed description of how the energy and capacity markets work in most US ISO/RTOs, so stay tuned if you’re interested in that whole shebang. Middle-Market Infrastructure Private EquityOne thing I love about my current seat at Cypress Creek is that I get to go deep and build pattern recognition for what “good” looks like with respect to utility-scale renewables projects. However, deal flow and sentiment in the industry have historically been pretty tightly indexed to federal policy developments, and trying to maintain a fundamentals-driven investment philosophy is difficult when the ground under your feet is constantly shifting [10]. I knew that’s what I was signing up for when I joined CCR, and it’s been a great experience so far, learning from some of the smartest, most experienced renewables minds in the business – but at times I do envy folks at infra funds, whose broader investment mandates allow for diversification across multiple verticals. A little level-setting here: For infrastructure PE, I consider a fund “middle-market” if it’s under $5B, with $5-10B being upper-middle-market and $10B+ being megafunds. This is for individual funds, not for total manager infra AUM, which will likely be 2-10x the individual fund size depending on how long the GP has been raising and deploying infra funds. And figure each infra fund makes perhaps six to ten discrete investments over an eight to twelve-year fund life before “exiting” those investments by selling them to another owner or taking them public via IPO. These funds are largely similar to standard private equity buyout funds, except that they seek to acquire assets (or platforms owning assets) with infrastructure characteristics. This means stable, contracted long-term cash flows generated by capital-intensive assets with high barriers to entry. A utility that operates as a regulated monopoly, with governmental fiat preventing competition from new entrants, is a great example of a typical infrastructure PE acquisition target. Other assets and firms in the infra PE sweet spot include independent power producers, toll roads, airports, passenger and freight railroads, parking lots and garages, and wastewater treatment plants. However, all of these types of investments would likely fall under a “core” or “core-plus” investment mandate, which means relatively low risk coupled with lower target returns of around 8-13% (levered IRR). Core and core-plus infra funds usually, though not always, are larger than higher-upside “value-add” and “opportunistic” funds. This is because smaller funds can find more high-growth opportunities at their target check size, often rolling up multiple assets into a platform, investing in development-stage projects, or purchasing uncontracted assets and putting long-term offtake on them (therefore taking on execution, development, construction, and/or commercial risk), rather than buying stuff that’s already operating with long-term contracted revenue agreements. By contrast, there are only so many large, de-risked operating infrastructure assets out there that are big enough to be in the strike zone for the $10B+ megafunds, since infra funds tends to make around eight investments per fund, regardless of fund size (above I assumed six to ten; occasionally it will be more but very rarely fewer). Therefore, bigger funds must write bigger individual checks. This drives fiercee competition when big-ticket, low-risk assets do come onto the market, resulting in higher entry prices that suppress returns. What this means practically is that middle-market infra tends to underwrite growth stories – a solar project developer whose first tranche of projects is 18 months from being ready to build but who needs capital now to fund interconnection deposits and equipment procurement; an EV charging station operator planning to go from 50 locations to 500 while tightening up the operational performance of its existing assets; a proven lithium-ion battery manufacturer seeking capital to expand its production facility and meet booming customer demand. These investments may look a bit more like non-infra growth or buyout private equity, perhaps possessing “infra-like” characteristics without having the locked-in cash flows of core or core-plus assets like, for example, a municipal landfill operator with a 50-year fixed-price tipping rate and predictable operations costs via long-term labor and equipment maintenance contracts. Some firms doing interesting work in the energy transition corner of this space are NOVA, SER, Spring Lane, GDEV, Khasma, and Wollemi. Each takes a unique approach to middle-market infra and energy transition investments, not only in terms of their target firms and sectors, but also the financial structures they use to fund their portfolio companies’ growth while satisfying LPs’ desire for aligned incentives and downside protection. For example, the Spring Lane model is to invest in their portfolio companies’ first few assets at the project level, injecting expensive but necessary equity to help developers without a balance sheet scale the next X projects after the company pilots its first-of-a-kind asset, while also taking a minority stake in the portco’s corporate entity. This way, Spring Lane retains upside as the portco scales up deployment beyond those next X projects required for commercial proof and finds cheaper capital to finance additional future projects. Market(s): The energy, transport, waste, and water industries, mostly in the US, Western Europe, and East Asia, although emerging markets infra funds also continue to become more common. Why it’s exciting: Broadly, investors outperform when they have the skills, expertise, and conviction to better understand and underwrite certain risks than the rest of the market. For middle-market infra funds, this means building a team and an investment process that combines operational experience with financial sophistication, and values deep thinking about what makes a real asset business successful. I deeply enjoy the process of grappling with the risks and uncertainties associated with bringing renewable energy projects to fruition, and I’m also interested in applying that knowledge to a broader set of decarbonization infrastructure businesses facing similar challenges as they scale up to meet basic human needs more cost-effectively and sustainably. Of course, as an investor, you’re by definition always “buying” something when you deploy capital to an opportunity, but middle-market infra, more often than its upper-middle-market or megafund siblings, adds a meaningful “build” component into the “buy,” and that scratches an itch for me. Business model(s): As a PE investor, you typically want to raise a dedicated fund with a 2-and-20 structure: 2% fee charged to limited partners on assets under management, and 20% carry (profit sharing) on all investment returns above a minimum threshold, usually an 8% IRR. Infra funds may charge lower management fees to LPs because their returns tend to be lower, and they tend to manage more money per head, than venture, growth equity, or standard buyout PE shops, but the structure historically has looked similar across these asset classes even if the numbers vary a bit. However, we’re in a weird fundraising environment at the moment: It’s very difficult to raise a first-time infra fund right now, especially if you’re pursuing a non-obvious mandate that involves platform-building and/or taking positions in pre-COD assets. The alternatives are to raise a smaller 2-and-20 fund than you wanted and augment this with co-invest from your LPs when you want to make a larger investment than your dedicated fund size allows, or to operate as a “fundless sponsor,” raising capital from LPs on a fully deal-by-deal basis. Regardless, because MM infra means smaller deal sizes, the main way to make money in the sector is from carry, not from management fees, which means there’s a strong incentive to perform. That incentive is weaker for megafunds, which can miss their carry returns hurdles and still print money for the GP because, well, 1.5% of fund AUM is a lot of money when fund AUM is $10 billion! The bigger your dedicated pool of capital, the more your incentives shift from “outperform to win” to “market perform and sit on a fee stream.” That might make some people rich, but it’s fundamentally uninteresting and also bad for LPs. All this to say that the intended incentive structure behind the 2-and-20 business model seems to be a better fit for MM infra PE than for larger funds. Challenges to widespread deployment: N/A; MM infra PE is already a well-understood market segment and there’s nothing novel about GPs raising money from LPs to invest in assets or platforms. Unlike larger fund sizes, MM infra funds are typically the first institutional owners of the businesses they acquire, so it’s less likely that these funds need to raise a continuation vehicle or sell off a business for parts because it’s too big for another fund to buy – a problem that is increasingly affecting larger infra fund managers as they scramble to figure out who the next buyer is for massive platforms that they have already owned for six or seven years and now need to divest in order to return capital to their LPs. So while PE in general, including infra PE, is going through a bit of an exit squeeze, MM funds should be less affected by this than the big boys. ----- [10] Last summer’s renewables tax credit rug-pull being the latest example of this. Honorable Mentions: Other cool stuff I’m excited about!• Energy Efficiency and Demand Response: The ESCO model is time-tested for commercial facilities, but the sheer costs associated with connecting and serving large-load customers like data centers and factories means these large loads have an unprecedented economic incentive to manage and self-curtail their operations via wholesale market DR program participation, and to invest in equipment and operational upgrades that increase facility energy efficiency. As the old saying goes, the cheapest unit of electricity is the one you never use. Ameresco is the 800-pound gorilla in the legacy ESCO jungle, but new entrants to the DR/EE space also show promise -- although next-gen players like Kraken and Renew Home appear more narrowly focused on DG asset optimization (i.e. VPPs) rather than providing a full suite of EE/DR offerings to larger industrial customers. • Critical Metals Supply/Waste Recovery/Recycling: As discussed above, I expect we’ll soon reach full commercial LDES deployment for stationary, grid-scale use cases. However, existing lithium-ion technology’s high energy density makes it the clear front-runner in passenger vehicle electrification. That means a lot of valuable metals will be needed to produce EV batteries in the coming years, and will later need to be disposed of when those EVs reach end of life. Same thing with solar panels: Need a ton of them now, and need a way to safely get rid of them after about 30 to 40 years of pumping out clean power. Two interesting firms in this space are Nth Cycle, where my good friend Allie and her colleagues are creating a one-stop shop for domestic critical metals refining from both mined and recycled raw materials, and Comstock Metals, which is in the process of commissioning its first-of-a-kind solar panel recycling facility in Nevada, with a second Nevada facility and a third site in Ohio also planned [11]. • Modern Power Electronics: Producing main power transformers (MPTs), inverters, and high-voltage breakers that can help us deploy renewables assets more quickly, optimize their operations, and establish tighter, more resilient manufacturing and supply chains for these typies of compontents than legacy suppliers like Siemens, Hitachi, and GE can offer. Heron and DG Matrix, among others, are all over this challenge (Thanks Drew for putting DG Matrix on my radar!). I admit I'm both excited and undereducated about this space. Basically the pitch is to take pieces of power electronics that previously were "dumb," with just a few possible operating configurations and limited ability to swap between them, and making them "smart," with more possible configurations and more flexibility to select the optimal one in real time as asset output and transmission system conditions change. [11] Disclosure: I have a position in Comstock Inc, the parent company of Comstock Metals. Any discussion of Comstock Inc, Comstock Metals, and their affiliates herein is not investment advice. Do your own research. ---------------------- That’s all for now, folks! May your spring weather be mild and your Q2 be prosperous. Thanks for reading, Coop Standard Disclaimers• This blog post is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored by, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. Claude Pro was used to assemble certain sources and summarize some research findings, but no AI was used to draft or edit the text of this post. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.  

This will not be a “normal” edition of Coop’s Energy Transition Blog. Why? Well, I had one of those about 80% drafted and ready to go, and in the shuffle of rushing out the door to catch a flight to see family in the San Francisco Bay Area for Thanksgiving, I failed to pack my personal laptop for the trip. Then I was back home in New York again briefly before heading to India for a friend’s wedding. Also left the personal laptop at home during that time. And of course I keep all of my drafts exactly where they belong: In the Downloads folder on my desktop, no online backup. As an investment professional, I’m aware this does not speak well of my ability to assess and mitigate risk. Lesson learned? We’ll see. Now, here I am in Heathrow Terminal 3 on the way home from Delhi, trying to drop something at least mildly interesting into your inbox before 2026. Have also been sprinting to close a project M&A deal by end of year, along with (it seems) just about everyone else in the industry. That means it’s time to lower the bar, shoot from the hip, and The findings from the report’s modeling of the combined flex-connect/BYOC approach sound promising. Across the six modeled PJM data center sites, “Grid power remained available for more than 99% of all hours in the year, with on-site or co-located resources dispatched for only 40 to 70 hours annually to stay within transmission or generation limits. In total, this combined approach enables utilities to connect up to 3x as much data center capacity within two years as they could in 5-7 years under conventional methods, while maintaining system reliability.” I’m not smart enough to fully understand the details of their modeling methodology, although I expect it’s some form of security-constrained economic dispatch (SCED) on the generation side along with a capacity pricing auction simulation driven by expected PJM capacity builds and retirements over the next decade or so – at least, that’s how we did it at ICF when I worked in energy consulting for a few years right after college. One methodological detail that did jump out to me, though, was an assumptions slide in Appendix B titled “Flexible Demand Modeling.” [Image]  I would have loved to get the details behind the decision to use these 20/40/60% demand flexibility assumptions. 20% is assumed as the base case, but why? Is it common knowledge among data center operators that curtailing a facility’s utilization by 20% is relatively easy and painless? Are there new technologies available that allow data center operators to shift compute execution across a geographically diverse set of data centers in real time to enable 20% or greater curtailment of certain flex-connected facilities when required, while making up the difference by increasing utilization of data centers elsewhere that are not being curtailed at that time? And why use 40% and 60% demand flexibility as the more aggressive cases, rather than, say, 25% and 30%? What I’m getting at is a desire to better understand the operational implications of data center utilization curtailment, both at the individual facility and portfolio levels. Others more familiar with data center operational patterns and procedures may already get this intuitively, but it's a new topic for me, so I want to understand the intellectual underpinning behind these assumptions. Another comment: Even if each data center that implements the flex-connect/BYOC approach only experiences 40 or 50 hours of curtailment per year, it seems important to understand whether a) that curtailment occurs within the same 40 hours for all sites in the portfolio (indicating a systematic risk of inadequate compute availability at certain times when everything is being curtailed due to grid supply limitations) or b) there is low overlap between curtailment hours for each individual data center site (such that full compute needs can be met at all times as simultaneous curtailment across multiple sites is rare). Data center flexibility looks much more attractive and achievable without compromising compute capacity uptime if the operational reality reflects b) rather than a). Disclaimer: I need to sit down with the report in greater depth to see if they’ve addressed these issues and I just missed this on my first reading. A Brief Holiday Thank You 🎄🍾🙏I started publishing my blog back in June because I had too many thoughts about the renewable energy industry bouncing around my head and nowhere to put them. At a bare minimum, writing helps me crystallize my thinking and have something to refer back to as my understanding of the world changes and my knowledge base grows over time. It’s even better when that growth is accelerated by conversations sparked by something I’ve written about – whether in person, on the phone, or via a written message. With this in mind, I’m very grateful to you, the reader, for taking your scarce time and energy to absorb and respond to my writing. I look forward to continuing to share my thoughts here in 2026, and I’m excited to keep our dialogue going, whether you’re a close friend whom I bugged to subscribe back in June or you’re just happening upon the blog for the first time (in which case, please read my old posts; they’re better than this one). As a reminder, you can scroll down to the bottom of this page and enter your email to subscribe. This will send my future posts to your inbox a few days before I share them with the world on LinkedIn. Have a wonderful holiday season, enjoy some sweet treats, stay hydrated, and I’ll see you next year! Energetically, Coop Standard Disclaimers• This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored by, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.    

Further examining challenges to the narrative of rapid and sustained AI electricity demand growthTom Cruise gave us Mission: Impossible -- Dead Reckoning Part One last year, and followed it up rather confusingly with The Final Reckoning this year. No such titling-and-numbering chicanery will be tolerated on this blog. This is just straight-up Part Two, the sequel to Part One (which you can find here). Quick Recap of Part OneMany, many market observers are projecting rapid, sustained growth in electricity demand through the next decade, driven by the growth of AI data centers. In order for these projections to come true: 1. Paying demand for AI products and services must skyrocket. 2. The compute efficiency of AI training and inference must only increase slowly, if at all. 3. The energy efficiency of compute hardware must only increase slowly, if at all. There are several pretty good reasons to suspect that one or more of the three prerequisites above will not actually occur. Even if all three do occur, it’s not clear that the resulting AI-driven electricity demand growth would form an outsize share of total electricity demand growth. In This Part Two We'll:• Hear from an industry expert on a few additional reasons to doubt that aggressive AI electricity demand growth will occur. • Understand what we do (and don't) know about how Goldman Sachs created their AI electricity demand forecast. • Discuss how AI-driven demand, if it does spike, figures into topline electricity demand growth through 2030. • Explore what energy project developers can do to limit the risks of building assets to meet AI demand. Let’s get into it. Someone Way Smarter and More Sciencey Than Me Lays Out the AI Bear CaseRocky Mountain Institute cofounder and Stanford lecturer Amory Lovins, a near-mythical figure in the world of energy efficiency, renewables, and environmental policy, recently published a very well-written and comprehensive piece of analysis with a Hall-of-Fame title: “Artificial Intelligence Meets Natural Stupidity.” In the piece, which I encourage you to read in full, Lovins makes several key observations that cast doubt on hyper-bullish AI electricity demand growth forecasts. My takeaways from his writing: A. Future electricity demand growth tied to AI is highly uncertain, driven by countervailing trends of increasing AI training and inference usage, and increasing per-query energy efficiency, both of which have rates of change that may prove to be exponential, linear, or something else (asymptotic)? B. The revenue model to drive continued growth and eventual profitability for many large AI players is still unclear, as consumer and business willingness-to-pay for AI solutions will eventually depend on the extent to which AI applications meaningfully enhance productivity beyond the existing novelties of LLM text and image editing/generation services. C. The hype around AI for energy markets is reminiscent of similar demand growth optimism during the dot-com boom, which drove the construction of many coal and gas plants that ended up being underutilized when these rosy growth forecasts did not materialize. D. Firm/dispatchable power for data centers is less critical than popularly held, as many data centers have significant untapped load flexibility potential that can be optimized to better match the generation profile of renewables. E. AI and data centers more broadly are just one piece of the electricity demand picture: AI only accounted for 8% of global data center electricity use in 2023 and is projected to reach 11% in 2025, and data centers' share of total electricity use was just 4% in the US and 1.5% globally in those years. Items A and B map pretty well onto points that I made in Part One regarding the future trajectory of the three key unit conversions required to calculate AI electricity demand: AI Tasks/Year * Compute/AI Task * Energy/Compute = Annual AI Electricity Consumed.(See Part One Recap above or click back to the full Part One blog here for more on this.) But there's far more to explore here, as Items A and B also provide a good jumping-off point to dig into how market observers like Goldman Sachs develop their load growth forecasts. Then, in Items C and D, Lovins raises additional points that I didn't consider at all in Part One. Finally, Item E returns to the idea that all this AI data center demand growth could just be a drop in the bucket of overall electricity demand – a topic that I alluded to in Part One but didn’t have space to cover in detail. Let’s lump Items A and B together, then take on Items C-E one-by-one. Items A and B: Growth Assumptions and Forecasting UncertaintiesLovins has this to say about upward bias in AI power demand forecasting: “Even as the world’s largest tech companies invest hundreds of billions in new data centers amid an AI-led arms race, there is broad consensus that most AI projects requesting power will never come to fruition:” • “Developers often consider multiple sites for a single project.” • “Utilities are bombarded with speculative power requests from landowners hoping to flip properties to data center firms. (Requesting power is usually free.)” • “Big tech firms “spray” power requests across regions to hedge power-supply lags and their own planning uncertainties.” • “Developers may reserve more capacity for initial model training than they’ll need later for inference, to reduce competitively damaging delays in time-to-market.” The speculation and duplication issues Lovins identifies here indicate savvy forecasters should understand that when they see 10 GW (let’s say) of data center load requests in utility queues, they should expect only perhaps 1-2 GW of those requests to become reality: ““Conservatively, you’re seeing five to 10 times more interconnection requests than data centers actually being built,” said Astrid Atkinson, a former Google senior director of software engineering and now co-founder and CEO of grid optimization software provider Camus Energy.” (Quick aside: I worked with Astrid and the Camus team as a pre-MBA intern in Summer 2021 – they are doing great work to help utilities and project developers build a cleaner, more resilient, and more flexible grid.) With this in mind, let’s revisit Goldman’s projections. They see total US electricity demand (all sectors) growing at a 2.4% CAGR through 2030, and they expect 90 bps of that growth to come from data centers. They also say that data centers will account for 8% of US power demand by 2030, vs. 3% in 2022. How are they arriving at these projections? First off, they helpfully note: “A key question impacting compute demand is whether that demand is:” • “Pent-up (i.e., available new servers will be bought regardless of budget),” • “Not pent-up and constrained by demand itself, or” • “Constrained by customer budgets.” “In other words, What’s an energy developer to do?If you’re an energy project developer, and if you think, based on what I’ve laid out here, that there’s even a 10% chance (or 1%, Snyder-verse fans?) that electricity demand growth from the AI infrastructure buildout is substantially lower than all of these very smart bankers and consultants are predicting... What do you do about it? I see two straightforward answers for developers at risk of “putting all their eggs in one basket” by prioritizing AI demand: 1) Build a stronger, better-cushioned basketPrioritize serving data center customers for as long as the current boom lasts, but make sure you get sufficient commercial protection built in. What does this mean? If you’re an energy project developer who’s entered into a 20-year power purchase agreement to sell electricity to a data center, and all of a sudden that data center is underutilized such that it’s buying barely any electricity, or even shuts down entirely, you’re screwed – especially if you’ve built your asset behind the meter, without a grid connection that allows you the option to sell excess merchant energy into the wholesale markets as a backup plan. No matter how creditworthy OpenAI or Anthropic or even Google looks now, their massive spending to construct new data centers doesn’t guarantee that these facilities will be fully utilized, generate sufficient revenue, and therefore demand as much energy as anticipated. And if these companies own and finance their data centers through subsidiaries, it won't be their corporate balance sheet collateralizing your PPA with them. Plus, data center operators are obsessed with speed-to-power, meaning a behind-the-meter asset (or combination of assets) you’re building to serve them is likely only valuable in the very near term. Once they can get a full grid connection in year 4 or 5, they’ll likely be able to get cheaper 24/7 power from a front-of-the-meter PPA or from liquid wholesale markets than from your collocated generation assets. How can energy projects mitigate these risks? A few possibilities: 1. Sign higher-priced PPAs for a shorter period of time, reducing the risk of PPA default or termination in the later years if data center utilization rate drops off over time as more compute capacity enters the market. 2. Insist on a take-or-pay contract structure in which the offtaker is obligated to pay for energy that the project generates or the capacity that the offtaker is entitled to call on command, even if the offtaker does not actually need or use this energy or capacity. 3. Use a two-tiered contract structure consisting of a PPA and a tolling agreement, under which the PPA provides behind-the-meter energy supply for a shorter period of time until the data center receives its large load interconnection rights, along with a longer-duration tolling agreement to provide callable backup power/peak-hour overage even after the data center gets its grid connection. 4. Push for high PPA termination fees and long notice periods, so that you have sufficient time to re-contract the asset if the offtaker does decide to exit the PPA early. 5. Insist on a ParentCo guarantee from your off-taker, so that your PPA is collateralized by their balance sheet (or at least the balance sheet of a reasonable large subsidiary/division) rather than by an AssetCo SPV that contains only the data center itself. 6. Ensure that your project obtains its own grid connection rather than staying fully behind-the-meter, even if this means filing a separate generator interconnection request after COD, to enable commercial optionality if the initial PPA is later terminated, downsized, or otherwise modified by the offtaker. Ideally you’d pair two or more of the first five options along with the sixth to combine strong contractual protections from the off-taker with a backup plan for commercialization in the wholesale markets enabled by the physical configuration of the asset, should those contractual protections prove insufficient. 2) Put some of your eggs in a different basketWhile it’s reasonable to pursue some projects aimed at serving data center customers, developers should diversify with projects that serve other customer profiles, like: • EV charging and fleet electrification • Heavy industry and advanced manufacturing • Fulfillment centers, warehouses, cold storage facilities, and other logistics real estate • Campuses and multi-building complexes that value energy resilience, like universities and hospitals Here’s what I like about these four categories of electricity buyers: They all serve needs that already exist and must continue to exist for vast portions of society to function. Putting it in cavemannishly simple terms, we need to get where we’re going, and we also need to extract, make, handle, fix, and transport physical stuff – mine copper, can peaches, assemble airplanes, set broken bones, sort and deliver packages. While the new AI overlords are intent on taking as much of human lived experience out of the physical realm as possible, we are still creatures of the flesh, with all of the physical needs that this simple fact implies. Even if we consider a hypothetical techno-utopia (dystopia)? in which universal multipurpose 3-D printer ownership allows you to make whatever food, furnishings, toiletries, clothes, medical supplies, and electronics you want or need using a single machine without ever leaving your home, the raw material to feed into such a miraculous omni-printer still needs to be sourced, packaged, and distributed through the physical world to arrive at your door. Whether the nutrients in your 3D-printed and auto-cooked meals come from strawberries, steak, or genetically engineered protein-and-vitamin slush, it’s just not very efficient for each household to try to grow or gather those things on its own. Ditto for durable goods and electronics: Good luck mining cobalt or copper in your backyard. Unironically: Yay for factory farms and open-pit mines! Please feel free to take this line out of context and cancel me. Based on the idea that these material needs aren’t going anywhere, fully electrifying as much of the physical world’s energy consumption as possible strikes me as a good bet for investors and project developers – certainly less risky than betting that the AI behemoths manage to close the revenue gap for their products quickly enough to pay in full the massive hardware and energy bills they plan to rack up over the next few decades. Whatever you think of AI and its implications for energy use, GHG emissions, and the future of work, we already have a whole physical-world economy for which the transition to cheaper, cleaner power is equally urgent and likely a better investment. Standard Disclaimers• This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored by, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.

Examining challenges to the narrative of rapid and sustained AI electricity demand growthOver the course of human history, there have been a lot of can’t-miss trends that did, in fact, miss. And the surest sign you’re in a bubble is that you hear people talking about the trend in question using certain phrases: • “As we all know” • “It’s obvious that” • “McKinsey says” You get the idea. From Dutch tulips to mortgage-backed securities to skinny jeans, generally, when we reach a point at which everyone agrees an economic or business phenomenon is a no-brainer money-printing machine or an irreversible societal trend that will go on forever... it doesn't. Hockey-stick growth is rarely, if ever, sustainable over years and decades, with the unfortunate exception of atmospheric CO2 concentrations since 1958 (yikes). I’m not going to try to convince you that the growth of AI is a scam. Nor am I going to try to convince you that the near-term expansion of data center capacity to support the near-term growth of AI is a myth. Heck, I won’t even try to convince you that the projected rapid growth in electricity demand required to power that new data center capacity is a fantasy. Nor will I be so bold as to tell you that we’ll see reach the “peak of inflated expectations” and see the bubble burst this year, or next, or the following. To definitively assert any of the above would convey a level of certainty that I absolutely lack. But here’s what I do think is likely: Over the next ten years, we will see significantly lower electricity demand growth from AI-driven data center buildout than what most market observers are currently projecting. What the Experts* SayWhat are those growth projections? To keep it brief, here are a few big headline numbers folks are throwing around (all for the US market): • Goldman Sachs has data center electricity demand increasing at a 16% compound annual growth rate (CAGR) from 2023 through 2030, contributing 90 bps toward an overall 2.4% CAGR during that period for total US electricity demand. • McKinsey says (see above 😉) that data center electricity demand will increase from 178 terawatt-hours in 2024 to over 600 TWh in 2030, a 22.6% CAGR, meaning data centers’ contribution to total electricity demand would almost triple from 4.3% to 11.7% in that time. • Deloitte states that while topline data center demand is expected to grow more than fivefold in the next decade, AI’s share of this will grow much faster: AI accounted for about 12% of data centers’ 33 GW of power demanded in 2024, but they claim this will explode to 70% of a total 176 GW demanded in 2035. *Meaning the bankers and consultants. Up to you whether or not to read any sarcasm into this word choice. Energy industry veterans, feel free to skip the following two paragraphs on energy vs. load. Before we go further: These estimates are not apples-to-apples in that the Goldman and McKinsey figures predict energy consumption, or the area under the curve, in terawatt-hours, while the Deloitte figures predict load, or instantaneous power draw, in gigawatts. For data centers, whose utilization rate is fairly constant, i.e. their instantaneous power draw is relatively steady over time, we can make some basic assumptions that load translates fairly linearly to energy consumption over a certain period of time, but just know there is technically a distinction between a gigawatt, which is a rate of power consumption, and a terawatt-hour, which is a quantity of energy consumed, specifically the amount consumed when you power a gigawatt of load for a thousand hours. All of this to say that it’s possible for load, specifically peak load (the highest recorded instantaneous power draw of a system) to increase while energy consumption stays the same or even drops, or vice versa – but in the case of data centers, we can pretty safely assume that load and energy consumption move together. To be clear, when I refer to “electricity demand” or “power demand” throughout this blog, I am using that as an umbrella term to capture both load and energy consumption, because, again, given the operational patterns of data center, growth in one equates to growth in the other. Folks skipping the energy vs. load refresher, keep reading from here. Need a simple visual on what these demand growth projections look like? Here’s a handy chart from NRUCFC, a financial services co-op that lends to rural electric utilities: Click here to read Part Two of this blog.   Standard Disclaimers• This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored by, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.

A Book Review of The Great Texas Wind Rush Coop’s Commentary: Three Key Takeaways1. For emerging technologies, policy support starts as a “power button” and evolves into a “volume control.”According to the book, Kenetech’s Texas Wind Power Project, the first utility-scale wind farm in the state, happened because: a)     The LCRA, a public utility, was willing to pay $60/megawatt-hour, above market rates at the time, to procure clean energy and fund required transmission upgrades to deliver it, and b)     The federal Production Tax Credit (PTC) effectively juiced revenues by an additional $15/MWh. In other words, a combination of forward-looking utility executives (those in the industry know this phrase is almost as self-evidently paradoxical as, say, “rude Canadian” or “gourmet dry cat food”) and supportive federal policy incentives flipped the switch from “off” to “on” for the Texas Wind Power Project and others like it in the late 1990s. As the industry grew, turbines became cheaper and more durable, and best practices for development, construction, financing, and operations emerged, policy support transitioned from being a binary determinant of wind deployment to merely an accelerator. Today, following the repeal of the Inflation Reduction Act via the OBBB, and with it the elimination of the longstanding PTC for wind projects (along with the elimination of the newer solar PTC, and the ITC for both technologies), it’s important to understand this. Wind projects certainly benefited from the IRA’s $27.50/MWh, 10-year PTC incentive: By my calculations, the IRR uplift from PTC for the equity sponsor was about 300-450 bps, depending on your tax equity/ tax credit transfer assumptions and other project-specific inputs. However, some projects still can and will get built if the tax credits go away: With onshore wind capital costs having settled at around $1,700/kW (including transmission upgrades) plus another $20MM or so in other project delivery costs, a 300 MW project with a 40% net capacity factor, levered at 65-70% (based on a 1.3x debt service coverage ratio, or DSCR) with debt priced at SOFR + 200 bps (or about 6.5% as of today), assuming merchant revenues of around $40/MWh (real 2025$) upon PPA expiry, needs to hit a target 20-year PPA price of about $55/MWh to reach something like a 10% equity return. Whew. Long sentence there. All of those numbers are back-of-the-envelope, but they’ll do for our purposes here. That 10% is a fairly skinny return figure, and in a post-PTC world it largely depends on achieving higher merchant capture prices and PPA rates than what we are seeing today. But higher electricity prices are very likely to materialize post-OBBB given rapid recent and forecasted demand growth, widespread retirements of older thermal capacity, and a 5-year order backlog for new gas turbines. In this context, some amount of renewable projects can still offer a compelling contracted price to the big tech firms and other corporate electricity buyers who account for much of the demand for renewable energy, and therefore still find a way to get built. Of course it would have been much better for the industry and for general decarbonization efforts to preserve PTC entirely or phase it down over several years, but at this point the effect of eliminating tax credits on renewables deployment is more like turning down the volume dial (to be clear, WAY down with the passage of OBBB) than flipping the off switch. There will be serious consequences to this policy change: Many (most?) of today’s early-stage renewables projects targeting late-2020s CODs won’t be financially viable if they mature after ITC and PTC expiration (SOC by July 4, 2026 and PIS within 4 years of SOC, or PIS by the end of 2027), unless we see a massive increase in off-takers’ willingness to pay for PPAs. A lot of people in the industry will lose their jobs. A lot of investors will lose their money. This is all bad if your goal is to deploy as much clean energy as quickly as possible. However, this downturn won’t kill the industry outright: Once the tax credits are gone, the strongest projects, with advantages on items like capex, wind resource (i.e. projected capacity factor), nodal economics, and/or network upgrade costs, will still get built and make good money for their owners. Capital cost learning curves, project development expertise, and creative financial structuring solutions all flourished in the years following initial government support back when said support was a requirement, not merely an accelerant, for renewable energy deployment to occur, and now the industry has the scale, capabilities, and institutional knowledge to survive in a post-tax credit era – even if it’s harder to thrive than it was when the IRA and its precursor tax credit regimes were still in full effect. With wind and solar power now mature industries, attention is starting to shift toward deployment of the next tranche of clean energy technologies. As the industry looks to supplement rising solar and wind grid penetration with clean-firm solutions to keep the lights on (and the data centers humming) at all hours, what kind of policy support mechanisms are required to help emerging technologies like long-duration energy storage (“LDES”: Form, Antora, Noon, Malta, Rondo) and geothermal (Fervo, Quaise, Sage) deploy and come down the cost curve more quickly? To be clear, “policy support” need not be tax credits or subsidies; certain adjustments to power market design could be even more impactful. Almost everyone I’ve spoken to about LDES says the main barrier to grid-scale projects is that today’s ISO/RTO market structures don’t recognize the value of a 48-hour battery to the grid. And yet, in the abstract, we can all say with a certain amount of confidence that there’s a substantial gap between the very near-term incentives set by day-ahead and real-time energy markets, and the longer-term incentives set by year-ahead capacity markets. The creation of new multi-day reserve products in markets like PJM and MISO could bridge this temporal planning gap and help keep the lights on during extreme weather events, which thanks to meteorological advances we can now forecast with pretty good accuracy several days in advance – enough time to charge massive LDES systems that can then backfill any gaps in generation from either renewable or thermal resources until the heat wave or hurricane has passed. So, wholesale power markets (and vertically integrated utilities, too, if you dare): Get on this! 2. Transmission has become the main non-policy bottleneck to wind deployment.A lot of this book is devoted to showing us the multi-decade growing pains of the Texas wind industry in three main areas: • Manufacturing cheap, reliable, durable turbines. • Getting someone to buy the power they produced. • Finding policy support to make the economics work out. Until wind developers figured out these three issues, nothing else was relevant. Once they did, the focus shifted to tying up suitable land for project development, which meant identifying areas with good wind speeds and access to sufficient transmission to deliver the power to demand centers. Before OBBB’s passage a few weeks ago, this last challenge, transmission, had become the gating factor in the US preventing the deployment of renewable energy (both wind and solar) at a sufficiently rapid pace to slash power-sector carbon emissions while meeting fast-growing electricity demand from AI, EVs, and industrial electrification. Now, as the phase-out of tax credits means developers can no longer justify developing at anything other than a very strong node with minimal upgrade costs (again, assuming limited off-taker appetite to pay more for a PPA), this is even more true. The Great Texas Wind Rush previews this with its discussion of the Competitive Renewable Energy Zones (CREZ) program, a $6.8 billion expansion of ERCOT’s high-voltage transmission network in West Texas and the Panhandle, but the book was written before CREZ was completed, so allow me to fill in the rest of the story: CREZ worked for a while, until all of that additional east-west capacity was claimed by new projects, and now we’re back in the same place we were fifteen years ago, imploring the grid operator to beef up the high-voltage transmission network. ERCOT has generally been unresponsive to these requests until very recently, when its Regional Transmission Plant proposed building a few new 765-kV lines in response to demand growth driven by both the buildout of AI data centers and the electrification of oil and gas drilling operations in the Permian Basin. In a transmission-constrained environment, successful project siting and development depend on developers’ ability to identify uncongested points of interconnection (POIs) on the grid and and tie up land nearby. The idea is to target areas where sufficient injection capacity exists to ensure the project can sell its energy into the market without suffering from negative pricing events when local supply far outstrips demand. Of course, expanding transmission is (with apologies to Kevin Costner and James Earl Jones) an “if you build it, they will come” situation, or, as economists and highway engineers call it, induced demand: The more transmission you build, the easier it is to interconnect a new project, and the more projects get built, eventually gobbling up all of that additional capacity until the lines are once again fully utilized. But here, since we expect rapid electricity demand growth that can’t be met unless we build a lot more projects, this is a good thing! So – Texas and everywhere else – if you want abundant, affordable electricity, build more big long-distance transmission lines ASAP. 3. Show us the development grind!For all of its encyclopedic coverage of the personalities, policies, and locations that shaped the growth of Texas’s wind energy industry, the book is largely lacking in coverage of one key area: The multi-year development process that occurs between site control (signing of a lease or lease option at the proposed project location) and construction start. During this time, the project developer must do several phases of engineering and design work, refresh its outlook on local injection capacity and transmission constraints, advance various permitting and legal processes, navigate any potential community opposition from people opposed to the project, and raise financing for construction. As the project achieves key milestones over time, its probability of completion increases, while if it fails to achieve any of these milestones or suffers significant delays, especially early on in the process, the developer will likely cut their losses and discontinue development. As I mentioned, the book covers the R&D, manufacturing, site acquisition, and energy offtake aspects of the wind industry. However, it doesn’t really get into the nuts and bolts of project development after site control is obtained. Perhaps my work buying, selling, and developing pre-construction renewables projects has made me a glutton for punishment, but I wanted to hear more about the title insurance fumbles, mineral rights mishaps, and interconnection queue dramas that no doubt plagued the first wave of Texas wind projects before the shovels went into the ground. My college baseball coach loved to say, “Life’s too short to learn from your own mistakes, so you have to learn from other people’s.” In that spirit, I would have appreciated the opportunity to learn from the mistakes of the Texas wind pioneers and understand how their early experiences contributed to the emergence of certain best practices that we may take for granted today. To be fair, this book was written for a general audience, not for industry professionals, and it is a well-researched, entertaining read for folks less familiar with (or less interested in) some of the nitty-gritty details of project development. Final ThoughtsTwelve years on from its initial publication, The Great Texas Wind Rush holds up quite well as a deeply researched and painstakingly crafted narrative spanning the history of wind energy in Texas. While at times the book focuses too much on a few individual personalities at the expense of a greater level of insight into equally relevant processes and institutions, it is entertaining, informative, and a great entry point for general audiences seeking to understand Texas’s role in the global growth of wind power. I can only hope that one day, someone will publish an equally insightful work covering Clint Dempsey’s influence on the world of American soccer. Until then, it’s back to the YouTube highlight videos. ★★★★☆ (4/5 stars) If you’re interested in reading the book, check out your local library or click here to buy a copy. Standard Disclaimers• This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.

I’m sure I’ve missed plenty of other ways a renewables developer can achieve differentiation in today’s market, and I invite you to throw in your two cents – how else can developers set themselves apart? My overriding takeaway: With project economics set to take a nosedive and plenty of zombie projects clogging the interconnection queues, you’ll need some kind of special sauce to come out on the other side with cash in your pocket and your projects achieving COD as the entire US renewables industry sails into gale-force policy headwinds over the next few years. So figure out where your advantage is (or could be) and start executing on it, right away. Standard Disclaimers • This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way. • Want to weigh in with your thoughts? Propose a correction or improvement? Find me on LinkedIn or shoot me an email: cwetherbee at alumni dot gsb dot stanford dot edu.

Webster’s Dictionary defines “risk” as… Just kidding. This isn’t a 1990s commencement speech at a Northeast liberal arts college where the students wanted David Foster Wallace but got the president of the local credit union instead. Nope, this is a renewables and cleantech investing blog that nobody reads yet (if you're here in June/July 2025, you're one of the first!). Let’s talk about risk.  As a renewable energy investor, I haven’t seen a lot of good sources on risk – specifically the kinds of risks that should be evaluated and understood in order to confidently underwrite investment in developing greenfield renewables projects. So I decided to take a shot at it myself. SummaryHere are the basics if you’re pressed for time or just lazy:  • Severity * probability = magnitude. The two components of risk are severity and probability. The product of these is magnitude, or “expected loss” (i.e. a negative expected value). • It’s useful to specify which component of the risk equation is being discussed to enable clearer communication and better decision-making. People often speak broadly about risk without specifying whether they are speaking about severity, probability, or magnitude, which leaves room for misinterpretation and therefore bad decisions and bad outcomes.  • Some project risks are binary and others are non-binary. Binary means that the feared outcome would kill the project if it came to pass. Non-binary means that there is a range of possible bad outcomes, each of which would do a certain amount of harm to the project’s returns, timeline, or probability of completion, without (necessarily) killing the project outright. • I identify three “functional categories” of risk: Physical, permitting, and commercial. Physical: Can you build and operate the project? Permitting: Are you allowed to build and operate it? Commercial: Will doing so actually make you any money?  • Policy risk might deserve to be a fourth “functional category.” But the blog was getting long and there’s a lot to unpack on policy, so I’m leaving it out for now. Plus, this is more of a macro risk than anything project-specific. Read on for the whole shebang. Of course, the distinction between binary and non-binary risk can be arbitrary. For example let’s take an extreme case with an extreme proposed intervention: If buildability constraints due to steep slopes on the site are going to double my solar capex because I need to literally move tons of earth or build a massive platforms on stilts with concrete anchors to hold the panels, I could find a construction firm able to build this – it’s just that, in almost all cases, a doubling of capex is an automatic project killer, all else equal.  If all else is not equal – let’s say, for example, I’m in a very sunny climate and my energy off-taker is a mine operator willing to pay me $250 per megawatt-hour for a behind-the-meter power purchase agreement (PPA) with no interconnection process to navigate – then maybe the capex bump is not a project-killer. But for your standard grid-connected utility-scale solar project with a 20-30% first year AC net capacity factor, claiming 30% investment tax credit (ITC), with debt at about 2% above the risk-free rate, and contracted revenues via hub-settled virtual PPA at ~$40-$60/MWh, a 100% increase in capex puts the project well below the equity hurdle rate of ~10-12%. This cookie-cutter hypothetical project is what I’m using as a reference point when I say something is a project-killer, “all else equal.”  Risks By Function There’s one more dimension to add to the puzzle, which I’m choosing to call “function.” The functions I have in mind are Physical, Permitting, and Commercial.  Physical Physical risks are risks to the ability to physically construct, operate, or maintain the project. For example, if the project is in an area with unstable topsoil and steep slopes, buildability on the proposed project site could be a significant source of physical risk. Similarly, if the project is in a flood zone, and it is not economically viable to raise the height of the panels and power electronics onsite (assuming we’re talking about solar; not a lot of onshore wind projects are sited in flood zones!), panel inundation during a flood event is a physical risk. Basically, if you have doubts about your ability to install the thing, operate it, or get insurance for it given the site's combination of topography, geology, weather conditions, and natural disaster exposure, you're probably chewing on physical risk. Permitting Permitting risks are risks to the ability to obtain, and to manage the costs of obtaining, the necessary approvals and permits to legally build and operate the project. I see three subtypes here:  • Construction and land use • Environmental • Interconnection  Construction and land use permitting risks are possible threats to the ability to prove that your land is where you say it is, that it contains what you think it does, and that you can do what you want with it. To limit and eventually remove these risks, developers have to complete tasks like title commitment, ALTA surveys, mineral rights confirmation, DoD and FAA clearances, and local zoning and building code approvals processes.  Environmental permitting risks have to do with a project’s potential proximity to or overlap with endangered species habitats, migratory bird flight paths, waterways, and other protected areas designated and enforced by state and federal environmental authorities like EPA, the Department of the Interior, and (to pick out just one state-level agency among many) the Arizona Department of Environmental Quality.  The extent to which these two types of permitting risks overlap with physical risks depends on the jurisdiction. Some counties and states will pretty much let developers build whatever they want, where they want, and make it the developer’s problem if what they build breaks, catches on fire, washes away in a flood, etc. Others have strict building and zoning codes, with many layers of approval to navigate and very specific planning and environmental requirements that, if not followed to a T, will result in rejected permitting applications.  Interconnection permitting risks are risks to the project’s ability to secure connection to the grid – the right to “turn it on and plug it in.” Interconnection permitting processes vary from place to place and depend more on who operates the grid near your project than on what state your project is in. Different ISO/RTOs (in restructured wholesale markets) and utilities (in vertically integrated markets) have different rules, costs, timelines, and study processes that developers must navigate to execute an interconnection agreement that allows them to connect to the grid.  For example, the ERCOT interconnection process, for the grid that covers most of Texas, is fairly quick and cheap, partially because ERCOT (the grid operator) farms out the interconnection study grunt work to the local transmission owner (TO) into whose line a project will eventually interconnect. By contrast, many other ISO/RTOs have slow, opaque queues that can take several years and many thousands of dollars to complete, although recent FERC rulemakings have pushed grid operators to streamline their interconnection processes. We’ll see how this shakes out in 2025 and beyond, but I’m not particularly optimistic these efforts will succeed in clearing the backlog of existing projects in the queue, much less accelerating the process for new applicants. That’s too long of a discussion to get into here – stay tuned. One final point of discussion: You could make a solid argument that interconnection risks are best treated as a separate category, rather than going in the permitting bucket, but at the end of the day, you're paying money and doing studies in order to get permission from a quasi-governmental entity to turn on and plug in your project. For me, that's a permitting risk, even if it is a very specific type of permitting that also carries some impact on capex via interconnection upgrade cost assessments (which I would separate into the commercial risks category, for what it's worth). CommercialCommercial risks are risks to the project’s capital costs, operating costs, and revenues as compared to the commercial base case underwritten by the developer or investor. Two basic examples: First, on the cost side, the imposition of import tariffs, as we saw recently in December 2024 with Commerce’s updated anti-dumping rates imposed on certain Southeast Asian nations, could increase solar module pricing by 20% or more, raising project capex. Then, on the revenue side, expected PPA pricing in a project’s ISO/RTO (or in its specific load zone within that ISO/RTO) might decline, reducing the project’s expected revenues, if big-name off-takers like Google, Walmart, and Coca-Cola signal that they have already met their clean electricity procurement targets to power their data centers, warehouses, factories, etc. in that market region. These commercial risks don’t say anything about our ability to obtain the necessary approvals (permitting) and then actually build and operate (physical) the project – rather, they raise the possibility that the project may be less profitable than desired due to changes in the prices at which the project’s inputs are bought and/or its outputs are sold.  Closing ThoughtsTo name it is to know it. When I have trouble wrapping my head around a big, hairy topic like development risks, the natural thing for me to do is expend a little time thinking about how to categorize smaller buckets – to chunk it out until the chunks are of a more comprehensible size and scope. I’m going to iterate on how to use this very basic taxonomy in my work, but I already have a couple of ideas on how to use this framework to more effectively navigate conversations with development staff and with counterparties in potential project sale or financing processes.  Policy risk is a topic for another blog. I struggled with where to put it in this framework: As a subset of commercial risk, in its own category, or somewhere else? This also could depend on the specific policy at risk of repeal or modification. Right now, the main policy risk on renewables folks’ radars is a potential repeal or restructuring of the Inflation Reduction Act of 2022, specifically removing or reducing the renewables tax credits provided for in that law. I think we can consider ITC and PTC repeal or reduction to be a commercial risk, since this would negatively impact project capex requirements (ITC) and operational revenues (PTC). Monetization of these tax credits is technically mediated through tax credit transfer and/or tax equity partnership transactions rather than direct changes to costs and revenues, but the effect is nonetheless to reduce investment returns on renewable energy projects. Your feedback is welcomed. This is a first shot at putting some thoughts to paper in the hope that they might help me, and others, be more systematic in identifying, planning for, and mitigating key sources of risk in renewable energy project investment and development. If I’ve overgeneralized, omitted key considerations, or just flat-out missed the point, or if you have anything else to add to the conversation, drop me a line: cwetherbee@alumni.gsb.stanford.edu.  Standard Disclaimers• This blog is written in my personal capacity and reflects only my own thinking, research, experience, and opinions. • This blog post is not sponsored, endorsed by, or affiliated with my employer, although its content may be informed by some non-confidential aspects of my work. • This blog post is not investment advice and does not constitute any offer, solicitation to offer, or recommendation of any investment product or security. • This blog post is 100% human-drafted and edited. AI tools were not used in any way.

Coop's semi-regular thoughts on energy policy, projects, markets, and investment strategies to follow. Sign up below to join the mailing list and receive my writing in your inbox.