The hidden bottleneck in clean energy [partner content]

Episode Summary

Executive Summary: Rich Deming explains how clean energy project finance is slowed by fragmented diligence, late-breaking contract issues, and duplicated review across many parties. He presents CertScore as an AI-assisted platform that scores project risk early by combining curated domain knowledge, document analysis, and external signals to help developers, investors, and insurers make faster, better decisions.

Main Topics: Transaction friction in clean energy finance (Priority: 5/5): Deming describes how renewable projects involve many documents, counterparties, and review cycles, creating delays and surprise issues close to closing. Why AI needs structured domain data (Priority: 5/5): He argues that raw large language models alone are unreliable for diligence; CertScore was built with curated questions, expert input, and decision-grade data to avoid hallucinations. How CertScore works (Priority: 5/5): The platform ingests a data room, analyzes legal, financial, permitting, and counterparty risk, crawls public information, and outputs a 1–100 risk score with a summary report. Changing risk environment in clean energy (Priority: 4/5): Deming says the sector is now more mature but also more unforgiving due to tighter margins for error, tougher regulation, tax credit complexity, and interconnection volatility. Primary users and use cases (Priority: 4/5): The strongest demand is from funds and acquirers, but developers and insurers also benefit by identifying landmines earlier and avoiding wasted diligence capacity. Building across sectors (Priority: 3/5): Although CertScore began in solar, batteries, and bioenergy, the platform is designed to extend into additional energy domains using agentic AI and new curated inputs.

Key Arguments: Clean energy deals fail less because of the grid or supply chains than because the transaction process itself is inefficient and opaque. Early diligence matters more than late diligence; many project risks should be identified at term sheet stage, not 30 days before close. Generic LLMs are not sufficient for project finance diligence because the work requires structured, curated, discipline-specific data. CertScore combines human expertise from lawyers and engineers with agentic AI to create decision-grade analysis, not just a summary tool. A 1–100 risk score helps prioritize projects, surface landmines, and separate funding-ready deals from those needing more work. Market volatility and tighter policy conditions mean investors are asking harder questions, so avoiding mistakes matters more than ever. Insurers now function as critical gatekeepers because tax credits and project finance often depend on what can be insured. The platform can save time and capital by preventing wasted diligence budgets, repeated reviews, and late-stage deal failure.

Data Points: Platform development time: About 4 years - Deming says CertScore has been built over several years before launch/evolution. Initial question set: 300 questions - He and legal/technical experts mapped the core diligence questions for the discipline. Data room upload time: About 20 minutes - A developer can upload a project data room quickly, depending on file organization. Analysis runtime: About 3 hours - After upload, the platform runs multiple analysis processes and subprocesses. Risk score range: 1 to 100 - CertScore outputs a numeric project risk score. Typical diligence budget: $400,000–$600,000+ - He cites the capital a fund may deploy on diligence at term sheet stage. Team size: About 7 folks - Deming describes the CertScore office team in New York. Project scan cadence: 10 minutes per project - At East Energy, they used search scores to quickly thumbs up or down acquisition targets. Cap raise example: $1.2 billion - He references a fund needing to deploy raised capital under time pressure. Current timeline pressure: 30 days before close - He describes how documents or clauses often surface very late in diligence.

Pivotal Quotes: "if these problems are so common, like, why hasn't a better system emerged?" — Rich Deming: He frames the motivation for building CertScore after years of seeing recurring diligence failures. "If you just started taking a data room and throwing it into a large language model, you're going to get nonsense." — Rich Deming: He warns that generic AI without curated data is unsafe for high-stakes project finance. "the hidden bottleneck in clean energy. It's not the grid, not permitting, not supply chains, but the transaction itself." — Stephen Lacey: Host summary of the episode’s central thesis.

Implications: The industry may speed up if diligence becomes earlier, more standardized, and AI-assisted. Tools like CertScore could reduce wasted capital, help investors choose better projects, and let developers fix or abandon risky deals sooner.