Capital
Liberation

Something has to give.^¹⁻⁴

• Cash Deposit · 12% — cost of taking cash out of a development loan to post as deposit.
• Letter of Credit: 12% — whether cash-collateralized via expensive debt, or reflecting the opportunity cost of a corporate revolver. If you have a strong corporate revolver, your direct LC fee might only be 2%. But that LC eats your borrowing base one-for-one. Every dollar tied up in an LC is a dollar of credit you cannot deploy to lay steel or acquire a new site. The true cost isn't just the bank fee; it's the opportunity cost of lost leverage.
• Parental Guaranty · 0% — no explicit carrying cost if the parent absorbs the risk on its own balance sheet.
• Surety Bond · 2% — annual premium. No principal posted; the surety's balance sheet stands behind the obligation.

$36M − $6M = $30M of capital you could be deploying to build.

Thirty million dollars is not a rounding error. It could be a mid-sized community solar portfolio. It could be an interconnection upgrade at a second site. It could be two years of payroll for a 125-person team.

This is what I mean by capital liberation.

Cash is two parties. I give you my money; you hold it; you return it. Letters of credit are two parties plus a bank mechanism — essentially bilateral. Surety is three parties. The principal, usually the developer. The obligee, the utility or Independent System Operator (ISO) or offtaker. And the surety, backed by a highly rated balance sheet.

And what holds those three parties together is not collateral. It's trust — and not trust as a feeling. Trust with collateral behind it. Trust with signed indemnity. Trust in the balance sheet. Trust with consequence.

At the center of this is the General Agreement of Indemnity — the GIA. Signed by the principal who stands behind the obligation. The GIA is a trust-building document. When a principal signs a GIA, they are signaling something concrete to the surety: As the principal, I am standing behind this obligation. When a bond is placed, the obligee receives that trust signal indirectly, through the surety's willingness to underwrite.

Trust in surety is not an aspirational word. It's an operational one. Every term in the canonical trust equation maps to a specific, observable behavior — the way each of the three parties proves itself to the others. What looks like credit underwriting is also trust underwriting. What looks like claims handling is also trust maintenance. The bond is only ever as strong as the trust it encodes.

There's an equation for this.

1. 01
   Credibility

   Balance sheet, A.M. Best rating, T-listing, capacity limits, claims-paying history.
2. 02
   Reliability

   Ten-business-day pay, no mid-term cancellation, consistent underwriting year over year.
3. 03
   Intimacy

   The surety broker–underwriter–principal relationship, deep file knowledge, pre-qualification built across projects.
4. 04
   Self-Orientation

   Commission-chasing, carrier opportunism in hard markets, cancellation threats when the wind shifts.

“Trust is a confident relationship with the unknown.”

The unknowns stack high in the energy sector. Future cash flows. Interconnection timing. Offtaker solvency. Decommissioning twenty years out. Natural catastrophe exposure. The bond is the instrument that lets the three parties act confidently in the face of all of that.

This is why surety is not a commodity. It's why you can't buy it purely on price. And it's why, if we want to scale this — and we do — we have to treat trust as the product we're building. Not just the math we're marketing.

Every project has at least two.
Most would have three or more of these bonds.

Many professionals have either tried a chatbot on their work, or been told they should. Most walked away thinking: "this is great for drafting emails, but it doesn't understand my business."

That gap has a name. It's called context engineering. The industry consensus in 2025 shifted from prompt engineering to context engineering — Anthropic's engineering team calls it "the delicate art of filling the context window with just the right information."¹¹ The difference is simple. A prompt is what you ask the model. Context is what the model already knows when you ask.

Role. Input data. Task. Output format.

Before the question. The answer changes completely. For example: here are three prompts, each anchored to a workflow you run today.

You are a senior surety underwriter evaluating a new energy-sector
principal for a contract bond.

## Input Data
- Principal entity: [name, structure, state, years in operation]
- Project obligation: [bond type, face value, duration, obligee]
- Financials (last 3 years): [revenue, EBITDA, working capital,
  net worth, bank line utilization]
- Backlog: [$M, margin, concentration by obligee]
- Character signals: [claims history, liens, litigation, leadership
  tenure, references]
- Corporate indemnitors: [affiliated entities, parent structure,
  consolidated net worth, cross-guarantees]

## Task
1. Score the principal on the Three Cs (Character, Capacity, Capital)
   — Red / Yellow / Green on each, with the evidence cited.
2. Assess corporate indemnitor depth: consolidated net worth,
   cross-guarantee coverage, any affiliate gaps.
3. Flag the single biggest underwriting concern and the mitigation
   that would change your answer.
4. Identify the 3 diligence items you would require before issuance.

## Output Format
- 3Cs Scorecard: Character [R/Y/G] · Capacity [R/Y/G] · Capital [R/Y/G]
- Indemnitor Depth: Consolidated NW $[M] · Gap: [narrative]
- Biggest Concern: [one line] → Mitigation: [one line]
- Diligence List: [3 items]
- Preliminary Capacity Indication: [$M single / $M aggregate]

You are a surety claims counsel reviewing a proposed bond form
for enforceability and alignment with industry practice.

## Input Data
- Obligee: [utility / ISO / municipality / corporate]
- Bond type: [interconnection / PPA / decommissioning /
  performance / payment]
- Face value and term: [$M, years]
- Bond form text: [paste full bond language]

## Task
1. Score the form against the five workable-form features:
   (a) Defined pay timeline (<= 10 business days)
   (b) Default as a payable claim
   (c) Clear claim process (notice, documents, payment trigger)
   (d) Forfeiture on non-replacement
   (e) No mid-term cancellation
2. Identify any language that converts an enforceable payable claim
   into a discretionary investigation.
3. Flag on-demand or unconditional language that mimics letter-of-credit
   risk triggers (autonomous payment, no defense rights) — which
   increases underwriting friction and may impact claims-experience
   assumptions.
4. Recommend the three edits that most improve the form for both
   obligee protection and surety enforceability.

## Output Format
- Feature Scorecard: (a)–(e) [Pass / Fail / Partial] with 1-line evidence
- Discretion Risks: [list with line references]
- On-Demand / LC-Mimicking Flags: [yes/no · citations]
- Top 3 Edits: [edit · rationale · impact]

You are a surety claims manager walking through a default scenario
on a live energy obligation.

## Input Data
- Bond in force: [type, face value, duration, obligee]
- Principal: [name, current financial status, any distress signals]
- Claimed default: [narrative of what happened]
- Notice received: [date, form, completeness]
- Corporate indemnitor structure: [affiliates, consolidated NW]

## Task
1. Walk through the claim pathway in order:
   (a) Notice sufficiency — is this a proper claim?
   (b) Investigation scope — what facts must the surety establish?
   (c) Investigate-vs-pay decision window — estimate the days.
   (d) Likely outcome — pay, defend, or negotiate settlement.
2. Model three subrogation scenarios:
   - Principal solvent and cooperative
   - Principal distressed, corporate indemnitors solvent
   - Principal and corporate indemnitors distressed
   For each, estimate recovery % and timeline.
3. Identify where the GIA gives the surety leverage
   (audit, collateral, takeover rights, etc.) and where it doesn't help.
4. Flag any bad-faith exposure if the surety delays or denies.

## Output Format
- Claim Pathway: (a)–(d) with 1-line conclusions
- Subrogation Scenarios: 3 rows · Recovery [%] · Timeline [months]
- GIA Leverage Map: [where strong / where weak]
- Bad-Faith Watchlist: [3 items]

These are analysis. Not verdicts. The underwriter still underwrites. Claims still require claims counsels. Treat the output as a smart first-pass analyst — one who never sleeps and occasionally makes errors. Check the math. Always.

T Ngo (ARM, RPLU) is the founder of Clean Power Whisperer and a Strategic Risk & Insurance Advisor specializing in AI, energy infrastructure, and climate risk. Over two decades, T has underwritten, structured, and advised on complex risk and capital decisions across organizations including AIG, The Hanover, Essex Property Trust, and Pine Gate Renewables. T has completed executive training through Carnegie Mellon's Chief Risk Officer program and Stanford GSB's Sustainability Strategies program.

This briefing is the product of structured collaboration between practitioner judgment and AI systems. The thesis, the frameworks, the 2026 market reading, the surety mechanics, and the practitioner perspective are T's based on her experience in building surety programs and research. The drafting and iterative refinement were conducted in partnership with AI tools (Claude Code, Gemini CLI), with rendering and design developed on Lovable.

Capital Liberation is a public-facing argument, deliberately transparent in its citations and reasoning. A reader can follow the logic, audit the sources, and test the conclusions independently. That transparency is intentional: credibility is built by showing the work, not by obscuring it.

Clean Power Whisperer operates at the intersection of what AI can analyze and what experienced practitioners must decide. This briefing is an example of that intersection in practice.