Interconnection Queue Intelligence SaaS for Clean Energy Developers
At the end of 2024, 10,303 power projects totaling 2,290 gigawatts were waiting in America's interconnection queues, more than the country's entire installed generating fleet. About 90% of projects that enter a queue never come out the other side as an operating plant. For years, entering was nearly free, so developers sprayed requests across every promising substation and let the queue sort it out. That era ended with FERC Order 2023: queue entry now demands a $5,000 nonrefundable fee, study deposits up to $250,000, 90% site control, and withdrawal penalties for backing out. Picking a queue position is now a six-figure bet placed with almost no information. The developers placing those bets, roughly 1,400 smaller shops, community solar builders, and data center developers new to the grid process, do their diligence in spreadsheets and ISO PDFs. The intelligence layer that prices queue risk before the deposit is wired does not exist at their price point.
The Problem
Before a power plant can sell a single electron, it must survive the interconnection queue: the study process run by grid operators (ISOs, RTOs, and utilities) that determines what transmission upgrades a project needs and who pays for them. The queue is where clean energy goes to wait. Lawrence Berkeley National Laboratory's annual Queued Up review found 2,598 GW seeking connection at the end of 2023, up 27% in a single year, with solar (1,080 GW), storage (1,028 GW), and wind (366 GW) making up 95% of it. Only 311 GW, about 12%, had even reached an executed interconnection agreement, typically the last step before construction.
The queue is not a line. It is a filter with a 90% rejection rate. LBNL found that only about 19% of projects requesting interconnection between 2000 and 2018 had reached commercial operation by the end of 2023; the rest withdrew or are still stuck. Enverus, which scores every queued project with a machine learning model, estimates roughly 90% of renewable projects never progress beyond the queue, and that only about 10% will come online in the next three years. Timelines keep stretching: the median project took under two years from interconnection request to operation for plants built in 2000-2007, and over four years for plants built in 2018-2023. In CAISO, Enverus puts the average development timeline near eight years.
For a decade, the rational strategy was to enter early and often. Queue positions were cheap, first-come-first-served ordering rewarded speculation, and a developer could hold a dozen positions for the price of paperwork. FERC Order 2023, issued July 2023 with compliance rolling through 2024 and 2025, killed that strategy on purpose. The rule replaced serial first-come studies with first-ready, first-served cluster studies and attached real money to every step: a $5,000 nonrefundable application fee, study deposits of $35,000 plus $1,000/MW for projects under 80 MW, $150,000 for 80-200 MW, and $250,000 above 200 MW, plus 90% site control demonstrated at the time of the request and 100% by the facilities study. Withdraw, and withdrawal penalties apply, capped at the deposits collected. A 150 MW solar project now posts $155,000 before any engineer studies whether the local substation can even take it.
Here is the information asymmetry. The grid operator knows the queue: every active request, every study result, every upgrade cost assigned, every withdrawal. The developer knows their own project and whatever they can scrape from seven ISO websites, each with different formats, different update cadences, and different definitions of "active." A mid-size developer choosing between three substations for a new project is placing a $155,000 bet with a PDF of last quarter's queue report and a phone call to a consultant who bills $400 an hour. The data to price that bet exists. It is just scattered across forty-plus sources in formats designed for compliance, not decisions.
The 2024 Withdrawal Bill: A Calculation Nobody Runs
2024 was the first year queue capacity ever declined, from 2,598 GW to 2,290 GW across 10,303 active requests, according to ClearView Energy Partners' analysis of LBNL data. The decline was not a triumph of throughput. A record 31 GW of large solar and 11 GW of storage did complete interconnection, but new entries collapsed and a record 112 GW of solar and storage withdrew, driven by political uncertainty, high interest rates, tariffs, and local permitting fights. The queue stopped growing because projects started dying faster than they were born.
Nobody totals what those withdrawals cost. Let us do it, with the assumptions stated plainly.
Start with the nonrefundable piece. At an assumed average withdrawn project size of 120 MW, 112 GW implies roughly 930 withdrawn requests. Each paid a $5,000 nonrefundable application fee under the Order 2023 framework: about $4.7 million in fees alone, gone regardless of outcome.
Now the sunk development spend. Before withdrawing, a developer typically pays for land options or site control work, permitting, legal, and the study deposits themselves. Industry development budgets for utility-scale solar commonly run $25,000 to $45,000 per MW through the interconnection study phase. Take a central $30,000/MW: 112,000 MW × $30,000 = $3.4 billion in sunk capital on capacity that will never generate. The range across $20,000-$45,000/MW is $2.2 to $5.0 billion. Some of that spend is recoverable, land options can be redeployed, studies inform the next site, but even discounting a third as salvageable leaves over $2 billion destroyed in a single year by queue positions that should never have been taken.
That is the measurable price of entering queues blind: a $2-5 billion annual triage failure. Software that moves even a fraction of doomed entries from "deposit posted" to "never entered" pays for itself hundreds of times over. A developer who avoids one bad 150 MW queue entry saves $155,000 in deposits and fees plus roughly $4.5 million in sunk development spend. At $30,000 a year for the top software tier, the payback math is not close.
The Gap in the Market
Queue intelligence exists, but it is built for buyers with enterprise budgets and analyst teams.
| Company | What They Do | What's Missing |
|---|---|---|
| Enverus (+ Pearl Street Interconnect) | Energy SaaS with ML probability-of-success scoring on every queued project, annual Interconnection Queue Outlook, grid analytics monitoring 1,300+ generators and 43,000 constraints. Acquired Pearl Street Technologies for in-house shadow studies and injection capacity maps. | Enterprise contracts priced for majors and investors. The outlook is a report, not a workflow. Success scores are a black box tuned for portfolio valuation, not for a 5-person developer deciding where to site next quarter's project. |
| LandGate (PowerData) | Standardized national queue database joined with parcel and land data, LLC unmasking for M&A, county-level queue pressure maps. | Built for land transactions and acquirers, not developer workflow. No withdrawal prediction, no network upgrade cost forecasting, no cluster-study scenario modeling. |
| Nira Energy | Denver startup (backed by Energize Capital) pairing ex-transmission engineers with software talent to shave interconnection costs through siting analysis, for generators and data center load alike. | Services-heavy and bespoke. A consulting engagement, not self-serve SaaS a small developer can buy on a credit card. |
| Wood Mackenzie / S&P Global | Power and renewables market research, asset-level data, analyst access. | Five-to-six-figure subscriptions, analyst-driven cadence. Nobody's project siting tool. |
| LBNL Queued Up (free) | The base truth: annual public dataset on every queue, the source everyone else builds on. | Annual, backward-looking, no predictions, no alerts, no workflow. By the time it publishes, the queue has moved. |
The structural hole: every paid option assumes the buyer is a large developer, an investor, or a land company. The roughly 1,400 smaller developers, community solar builders, and first-time data center entrants who now must post real deposits get the free LBNL annual report and whatever they can download themselves. Nobody sells them a self-serve tool that answers the three questions that matter before the deposit is wired: will this queue position survive, what will the upgrades cost, and how long will it take.
The Solution
A self-serve queue intelligence platform that normalizes every US interconnection queue into one database and layers prediction on top:
1. Queue survival scoring ($800/mo team plan): Every active request in every ISO and major utility queue, deduplicated and entity-resolved (the same developer often files under five different LLCs; the platform unmasks that). Each position gets a survival score trained on 15 years of LBNL outcomes: given the ISO, the substation, the cluster, the developer's history, and the project's stage, what share of comparable requests reached operation? A developer sees not just "you are #47 in the MISO queue" but "positions like yours reach operation 23% of the time, and the median survivor took 5.1 years." Track up to 20 projects with change alerts when a competitor withdraws, a cluster restudy is announced, or an upgrade cost revision hits your queue.
2. Upgrade cost estimator ($2,500/mo pro plan): The number that kills projects is the network upgrade bill, which arrives as a surprise in the system impact study, often 18 months after the deposit. The estimator trains on historical assigned upgrade costs by substation, voltage class, and cluster composition to produce a pre-study range: "comparable 150 MW solar at this substation drew $8-22M in assigned upgrades; your cluster has 4.2 GW competing for the same headroom." Unlimited projects, cluster scenario modeling (what happens to your costs if the two projects ahead of you withdraw), and API access for in-house tooling.
3. Siting copilot (pro plan): Given a target market and project size, rank substations by expected all-in interconnection cost and timeline, blending the survival model, the upgrade estimator, and published available-transfer-capability data. This is the pre-entry diligence product: the $155,000 deposit decision, made with data instead of a consultant's gut.
4. Diligence reports ($15,000 one-off, enterprise $60K+/yr): For investors, lenders, and acquirers: portfolio-level risk scoring on queued assets under LOI, benchmarked against the survival model. The enterprise tier adds custom model tuning, data feeds, and white-labeled reporting for funds doing programmatic development-stage M&A.
Revenue Model
| Revenue Stream | Amount | Notes |
|---|---|---|
| Scout (self-serve team) | $800/mo | Up to 20 tracked projects, survival scores, change alerts. Credit-card purchase, the wedge into small developers. |
| Developer Pro | $2,500/mo | Unlimited projects, upgrade cost estimator, siting copilot, cluster scenarios, API. Annual contract. |
| Enterprise | $60K+/yr | Investors, lenders, large IPPs. Custom models, data feeds, white-label diligence reports. |
| Diligence reports (one-off) | $15,000 | Per-portfolio risk scoring for acquisitions. Feeds the enterprise pipeline. |
Unit economics on a 40-person solar developer: The developer runs 12 active queue positions across MISO and SPP. Developer Pro at $30,000/year. The platform flags two positions with survival scores under 15% and upgrade estimates 3x the developer's budget; the developer withdraws before the facilities study, saving roughly $300,000 in deposits and an estimated $7M in sunk development spend across the two projects. CAC for a mid-size developer (industry conferences, targeted outbound, content): $8,000-12,000. Payback under six months. Gross margin on software revenue above 85% once the data pipeline is built; the pipeline is the fixed cost, each new ISO a marginal one.
Market Size
TAM: ~$50M/year. Build it bottom-up. Active requests at end of 2024: 10,303. At an assumed 7 requests per developer, that is roughly 1,470 developers with queue positions. Add adjacent buyers of queue intelligence: ~300 infrastructure investors and lenders active in US clean energy, ~150 utilities and transmission owners buying planning data, and ~250 corporate offtakers and data center developers entering the process from the load side. Buyer universe: ~2,170 organizations. Blended ARPU of ~$23,000/year (70% self-serve at ~$12K, 30% enterprise at ~$48K) yields $50M. The 7-requests-per-developer and ARPU-mix assumptions are the soft spots; halve them and TAM is $25M, double the enterprise share and it is $75M.
SAM: ~$28M/year. The startup does not chase Enverus's enterprise accounts on day one. The serviceable segment is the long tail: ~1,400 smaller developers, community solar builders, data center entrants, and land originators at ~$20,000 average contract = $28M. This is the segment with no purpose-built tool today.
SOM (year 3): ~$3.2M ARR. 180 customers at ~$18,000 average (mix of Scout and Pro, a handful of enterprise): 120 Scout at $9,600 = $1.15M, 50 Pro at $30,000 = $1.5M, 10 enterprise at $60,000 = $600K. Total $3.25M ARR, roughly 12% of SAM. The constraint is sales capacity, not demand: each enterprise diligence report converts to a $60K subscription at a plausible 30% rate.
Why Now
Deposits made diligence valuable. Before Order 2023, queue intelligence was a nice-to-have because entry was cheap and positions were disposable. Now a single entry risks $35,000 to $250,000 in deposits plus withdrawal penalties, and site control ties up land capital. The ROI of pre-entry diligence went from theoretical to arithmetic in one rulemaking. Every developer doing the math for the first time under the new rules is a prospect.
2024 broke the old playbook. The first-ever decline in queue capacity, the record 112 GW of withdrawals, and the collapse in new solar entries mean the market shifted from "get in line anywhere" to "pick the surviving line." PJM flushed over 1,100 requests in its backlog clearing (3,065 to 1,942); MISO's queue grew by 544 requests in the same year. Regional divergence is now the story, and regional divergence is exactly what a model prices.
Data center developers are grid novices with money. Large-load interconnection requests are surging alongside data center buildout, and the buyers, hyperscalers, developers, and their capital partners, have no institutional memory of interconnection risk. LandGate already markets queue data to "energy and data center developers" as one segment. A self-serve tool that speaks their language, timelines and dollar risk rather than tariff appendices, meets a buyer with budget and urgency.
The IRA wave is hitting the cost shock. Over 1,200 GW entered queues after the IRA's 2022 passage. Those projects are now receiving cluster study results, and the upgrade bills are the moment developers learn what their queue position actually costs. Demand for upgrade-cost forecasting peaks exactly when the first big cluster results land, which is now.
Startup Costs
| Category | Cost | Notes |
|---|---|---|
| Data engineering (12 months) | $320K | 2 engineers. Scrape and normalize 7 ISO/RTO queues plus 40+ utility queues, each with different formats and cadences. Entity resolution across LLCs. Historical backfill to 2010 from LBNL datasets. This pipeline is the moat. |
| ML and modeling (9 months) | $140K | 1 ML engineer. Survival model trained on 15 years of queue outcomes; upgrade-cost estimator by substation and cluster; backtesting framework. Validated against held-out cohorts, not vibes. |
| Product and frontend (9 months) | $120K | 1 engineer. Self-serve app, alerting, API, billing. Credit-card signup must work on day one; the long tail will not sit through enterprise sales. |
| Grid-domain advisors (part-time) | $90K | 2 ex-ISO/RTO interconnection engineers. Model validation, credibility with developer customers, cluster-study reality checks. Non-negotiable: developers smell purely-software teams instantly. |
| Pilot program (15 developers, 6 months) | $40K | Free access for design partners across solar, storage, community solar, and one data center developer. Goal: 3 paid conversions and a public case study with measured avoided deposits. |
| Sales and marketing (year 1) | $60K | RE+, ACP conferences, targeted outbound to developers with recent withdrawals (public data), content marketing on cluster study results. |
| Infrastructure (12 months) | $36K | Scraping infrastructure, hosting, data licensing, monitoring. Queue pages change without notice; the scrapers need babysitting. |
| Legal and buffer | $44K | Terms of service for a product whose outputs influence six-figure decisions, plus operating reserve. |
| Total | $850K |
Limitations
The $2-5 billion withdrawal-bill calculation rests on the $30,000/MW sunk-cost assumption, which blends land options, permitting, legal, and deposits across technologies and markets that vary enormously. A community solar project in Illinois and a 300 MW solar-plus-storage project in Texas do not burn the same dollars per MW before withdrawal. The range is honest but wide, and the true number could sit outside it if withdrawn projects skew smaller and cheaper or larger and pricier than assumed.
The TAM build-up assumes 7 requests per developer and a 70/30 self-serve/enterprise mix, both unvalidated. If the long tail consolidates, if developers share logins, or if the real buyer universe is half the estimate, the SAM halves with it. The SOM assumes a 30% conversion from diligence report to enterprise subscription with no evidence yet.
The survival model trains on historical outcomes in a regime that just changed. FERC Order 2023's cluster process, withdrawal penalties, and site control requirements mean the 2010-2023 training data describes a game with different rules. Backtesting against the transition period is essential, and the model's early predictions deserve wide confidence intervals until the first full cluster cycles complete.
Queue data quality is the operational risk. ISO queue postings are compliance documents: fields go missing, statuses lag reality by weeks, and "withdrawn" sometimes means "moved to a different queue." The entity-resolution problem, matching "Bluebird Solar LLC" to its parent developer across seven ISOs, is genuinely hard and never finished.
Strongest Counterargument
Enverus already built this. They score every queued project with machine learning, publish the industry's standard outlook, monitor 1,300 generators and 43,000 constraints in real time, and acquired Pearl Street Technologies specifically to do in-house shadow studies and injection capacity maps. They have the data, the models, the enterprise relationships, and a multi-hundred-million-dollar SaaS business funding further development. LandGate owns the land-plus-queue join. Nira Energy has the ex-transmission engineers and venture backing. A startup selling "queue intelligence, but cheaper" is bringing a spreadsheet to a gunfight: the incumbents can cut a downmarket tier the moment the segment proves real, and they start with better data.
The counter has three answers, none decisive. First, Enverus's product is tuned for portfolio valuation by investors, not for siting decisions by developers; the workflows differ enough that "cheaper Enverus" misdescribes the product, which is closer to a Bloomberg terminal for queue positions than a discounted outlook report. Second, incumbents face the classic innovator's dilemma on price: a $9,600/year self-serve tier cannibalizes the enterprise contracts that fund their analyst teams, so they will cede the long tail longer than expected. Third, and most honestly, the deepest threat is not Enverus but FERC itself: Order 2023 requires transmission providers to publish heatmaps of available headroom, which gives away part of the siting value for free, and any ISO that ships a decent public API collapses the scraping moat overnight. The defense is that heatmaps show capacity, not outcomes: the survival model, the upgrade-cost estimator, and the entity-resolved history are the product, and none of them come from a heatmap. But if ISOs ever publish standardized, machine-readable, real-time queue data with full history, the data moat evaporates and only the models remain. That is a regulatory risk no pitch deck can hedge, only monitor.
The Bottom Line
America's clean energy transition has a $2-5 billion annual triage problem hiding in plain sight: developers post six-figure deposits on queue positions with a 90% historical failure rate, using diligence tools built for an era when entry was free. FERC Order 2023 changed the economics overnight, making every queue entry a priced bet, while the intelligence to price it stayed locked inside enterprise contracts and annual PDF reports. The 1,400 smaller developers, community solar builders, and data center entrants now placing those bets have no self-serve tool that tells them which positions survive, what the upgrades will cost, or how long the wait runs. Whoever normalizes forty-plus queues into one database and sells survival scoring at $800 a month owns the workflow that decides where the next trillion dollars of generation gets built. The incumbents are watching portfolios. The developers are picking substations. That gap is the business.
What You Can Do
If you are a clean energy developer: Before your next interconnection request, pull the last three years of withdrawal and completion data for your target ISO and substation from the public queue postings. Compute the survival rate for projects at your stage and size. If it is under 25%, get a second opinion on the site before wiring the deposit, whether from a consultant or, eventually, from software like this. And track your own queue positions against the LBNL historicals: most developers have never benchmarked their portfolio's survival rate, which means they cannot tell skill from luck.
If you are an investor or lender: Ask every development-stage deal how the interconnection risk was underwritten. If the answer is "our consultant knows the ISO," you are underwriting on relationships. Survival scoring on the specific queue position, benchmarked against 15 years of outcomes, is the difference between diligence and storytelling. The 112 GW that withdrew in 2024 had investors too.
If you are building this: Start with two ISOs, not seven. MISO and SPP have public queue data, active developer communities, and enough regional contrast to prove the model. The entity-resolution pipeline is the moat, so build it before the models: a survival score on misattributed projects is worse than no score. Hire the ex-ISO engineer before the second ML engineer; credibility with developers is the distribution channel, and it cannot be scraped. Price the Scout tier to be a credit-card decision, because the 5-person developer will never survive your enterprise sales process, and they are the customer.
Related
📰 Second-Life Solar Brokerage: the secondary market for aging solar assets that survived the queue
📰 Data Center Water Compliance SaaS: the other infrastructure headache facing data center developers
📰 Transformer Fleet Health Intelligence: grid hardware analytics for the equipment the queue is waiting on