The Door Marked ‘Refinancing’ Is Narrower Than It Looks

Sponsors often treat commercial operation as the finish line for refinancing risk. The cases from Bahrain to Chile show that COD removes construction risk, but it does not create the cash-flow headroom or legal flexibility a new lender needs.

A $550 million debt balloon can become a $120 million equity problem without a single construction failure.

That is the trap hidden inside the reassuring phrase “commercial operation date”. Imagine an asset is running, selling power and meeting its original model. Its sponsors expect a new lender to refinance the remaining debt. Then interest rates rise, a lender trims expected output, and six months of debt payments must be reserved in cash. The asset still works. The refinancing maths no longer does.

Al Dur 1, a power-and-water project in Bahrain, came close to living that lesson in real time.

In 2009, ENGIE, the French energy company, and Gulf Investment Corporation had expected to arrange long-term financing for Al Dur. Lehman Brothers’ failure disrupted that plan and the wider financing market. The sponsors instead closed an eight-year hard mini-perm, lifted equity from $300 million to $500 million, extended the power-and-water purchase agreement from 20 years to 25 years, and accepted an 80% debt balloon at maturity, according to the World Bank’s Al Dur case study.

A hard mini-perm is temporary debt with a very real deadline. If it is not refinanced at maturity, default follows. Al Dur’s lenders also had an earlier alarm bell: if refinancing had not occurred by year five, the loan margin rose by 50 basis points and all surplus cash had to be swept into debt repayment.

That sounds punishing. It was also a survival mechanism.

Al Dur refinanced $1.3 billion in 2018 after roughly seven years of operations, then refinanced again for $1.2 billion in May 2024. The 2018 refinancing was hardly effortless. The World Bank records that rising US rates ended an intended bond tranche, while the negative value of old interest-rate swaps complicated the transaction. Regional bank liquidity stepped in instead.

The question is not whether Al Dur reached COD. It did. The question is why it still had a route out when many operational projects discover that the door marked “refinancing” is much narrower than expected.

Most sponsors know the conventional script. Construction ends. The plant begins operating. Construction debt becomes infrastructure debt. A bank, private-credit provider or bond investor takes out the remaining balance.

Plausible story. Wrong test.

The lender does not refinance your history

A new lender does not lend against the money already spent, the sponsor’s original spreadsheet or the asset’s nameplate capacity. It asks a colder question: how much debt can this asset safely repay from future cash flow?

That is why lenders use DSCR, the debt-service coverage ratio: cash available for debt payments divided by the payments due. A DSCR of 1.30x means the lender expects $1.30 of qualifying cash for every $1 owed in interest and principal.

Take the illustrative refinancing case in the research. An asset has $100 million of annual lender-case cash flow, a 1.30x DSCR and ten-year repayment. At a 5% interest rate, it can support about $594 million of new debt. At 8%, its capacity falls to $516 million.

Nothing has gone wrong operationally.

Apply a 10% haircut for merchant prices, availability or weaker contract pricing and capacity falls to about $465 million. Require six months of debt payments to be held in reserve and usable refinancing proceeds shrink to roughly $430 million.

That is how a $550 million balloon becomes a $120 million equity injection or amendment request.

S&P Global Ratings makes the point more formally in its project-finance methodology. It forecasts the balance still outstanding at maturity, applies its own cash-flow and repayment assumptions, and tests the DSCR after refinancing. Merchant exposure, contracts nearing expiry and weak cash-sweep performance are all refinancing risks.

Put that approach beside Al Dur and a pattern appears that none of the deal announcements says plainly: a mini-perm is not one loan with one exit. It is a sequence of fresh tests. The asset must earn cash, retain enough of it, keep its contracts alive long enough and legally rearrange its financing for the next capital provider.

A long contract cannot repair a broken plant

Cerro Dominador in Chile is the uncomfortable counterexample.

The EIG-backed 210MW photovoltaic and concentrated-solar complex raised $758 million in May 2018, including a $638 million seven-year mini-perm. Most output was covered by 15-year power purchase agreements priced at $114 per megawatt hour. The concentrated-solar plant was inaugurated in June 2021, according to infrastructure publication UXOLO.

On paper, refinancing risk should have been fading.

Instead, Cerro Dominador says mass failures in its hot-salt tanks were identified during 2023. The company states that Brookfield entered its ownership structure in June 2023 to support a capital increase and take control, while negotiations with senior lenders followed. Its published company information also reports no net generation from the concentrated-solar plant in 2024, compared with 115.4GWh in 2023, as repairs, refinancing and control changes proceeded.

Cerro Dominador showed that contracted power prices cannot support debt when technical failures halt generation. Photo: Ministerio Bienes Nacionales / Wikimedia Commons, CC BY 2.0.

A PPA protects a price. It does not repair equipment.

This is where most people stop looking. Contracted revenue is often discussed as if it were a physical asset. It is not. It is a promise conditional on delivery. Technical availability comes first. Then revenue. Then debt service. Then refinancing.

Cerro Dominador had contractual demand. It did not have net CSP generation in 2024.

The sponsor response was to add capital and negotiate with lenders before the maturity wall. That may preserve an asset. But it is not the easy takeout implied by the phrase “operational infrastructure”. For assets facing grid constraints as well as equipment risk, the warning is similar: a PPA can protect price without protecting revenue.

The cautious model that bought flexibility

London Array, the 630MW offshore wind project in the United Kingdom, took the opposite route.

After COD in June 2013, Masdar, the Abu Dhabi renewable-energy company with a 20% interest in London Array, financed that holding with up to £266 million of 12-year limited-recourse debt. Masdar’s financing materials sized the debt using P90 generation, a conservative production estimate, a 5% operating-cost uplift, a minimum DSCR of 1.20x and an average DSCR of 1.30x.

In April 2020, Masdar refinanced with approximately £466 million of bank debt maturing in December 2032. Masdar’s offering circular records 20 years of Renewable Obligation Certificates and an offtake agreement covering both electricity and certificate benefits.

London Array’s conservative debt sizing gave future lenders room to underwrite a downside production case. Photo: Bodgesoc / Wikimedia Commons, CC BY-SA 4.0.

At first glance, conservative debt sizing looks like money left on the table. Less borrowing means more equity is needed early. Yet it creates room for the next lender to test weaker output and higher costs without finding a structure that fails at the first unpleasant assumption.

Al Dur achieved something similar through repayment. Its loan amortised as though it ran for 20 years, despite lasting only eight. The 25-year purchase agreement outlasted the debt. The cash sweeps reduced the remaining balance if refinancing did not arrive quickly.

Both projects were buying optionality. They were not merely repaying debt.

The clause nobody reads until it matters

South Africa offers a different surprise. The refinancing of operating renewable assets there was not made possible solely by better operations.

In September 2021, Globeleq, the power-company owner, and Absa refinanced the 138MW Jeffreys Bay wind farm and the 50MW De Aar and Droogfontein solar plants with approximately R5.2 billion of senior debt. The assets had 12 years left on their PPAs. Absa said the refinancing was expected to save Eskom, South Africa’s state-owned power buyer, more than R1 billion through lower tariffs.

South Africa’s refinancing rules made lender changes and tariff-sharing a documented feature, not a last-minute negotiation. Photo: Caracal Rooikat / Wikimedia Commons, CC BY-SA 4.0.

The important decision had occurred before the deal. South Africa’s 2020 REIPPPP Refinancing Protocol expressly permitted changes to interest rates, maturity and debt quantum, while requiring refinancing gains to be split 50/50 between government and project companies. The World Bank notes that standardised 20-year local-currency PPAs and government support for Eskom’s obligations gave incoming lenders a known revenue and consent framework.

That is the twist. Refinancing optionality can be a feature of procurement policy, not an eleventh-hour sponsor negotiation.

The next lender may come from elsewhere

ReNew, the Indian renewable-energy company, offers another useful break with the usual script.

In July 2022, ReNew announced that it had replaced $525 million of dollar bonds due in 2024 with amortising project debt from an Indian non-bank lender. The company said the transaction cut its rupee-equivalent interest cost by 200 basis points, fixed rates for three years and extended maturity to fiscal 2027.

The available July 2022 ReNew release establishes that transaction. It does not establish later completion claims or separate repayments, so those should not be treated as part of the documented refinancing record here.

The lesson remains significant. ReNew moved two years before the offshore maturity and moved into domestic, amortising debt rather than assuming the original bond market would be available when needed. Currency-matched domestic liquidity can act like a refinancing reserve, even though it is not cash sitting in an account.

For founders who mistake announced financing for certainty, the broader warning will sound familiar: the term sheet is not cash. Neither is a planned refinancing.

When COD really does help

COD is not meaningless. Al Dur and London Array show that operating history can support longer-dated debt when performance is demonstrated and the remaining structure is sensible.

The evidence does not prove that every project with a short contract tail will fail to refinance. Nor does it mean every mini-perm needs identical reserves, hedges or cash sweeps.

It says something more useful. Completion improves the odds only if it leaves the next lender with a credible downside case. A project that is over-borrowed, technically unreliable or contractually constrained does not become safe because a ribbon was cut.

GI Network's view: The question before signing a mini-perm is not “Who will refinance us at COD?” It is “What would make a cautious new lender refuse us three years later?”

What sponsors should build before the clock starts

Sponsors should model the refinancing date as a separate credit event. Start with the remaining debt balance, not the original loan amount. Test it against lower output, higher rates and weaker pricing than the sponsor’s base case assumes.

Then inspect the documents as closely as the financial model. Can the project retain cash? Can hedges be transferred, unwound or replaced? Do the PPA and concession allow a new lender, revised repayment schedule or changed debt amount? These clauses decide whether a workable refinancing can actually close.

Map more than one capital pool well before maturity. Al Dur shifted from an intended bond component to regional banks. ReNew used domestic project debt instead of waiting on the offshore bond market. The lender that funds construction may not be the lender that suits operations.

What investors should ask for

Investors should be wary when a sponsor says an asset is “operational” as if that ends the analysis. Ask for the maturity balance. Ask how much contract life remains after the proposed new debt matures. Ask which assumptions belong to the sponsor and which ones a new lender will replace.

Cash sweeps, reserve accounts and distribution lock-ups can frustrate equity holders. Their purpose is not decorative lender caution. They prevent a future refinancing from becoming hostage to an avoidable debt balance or liquidity gap.

The gap between reported performance and a lender’s verdict is often the whole deal, particularly where future cash flow must carry old debt. It is the same distinction explored in bankable revenue rather than reported revenue.

GI Network would rebuild the refinancing case around the future lender rather than the original sponsor model: stress the maturity balance against downside cash flow and rates, examine contract and hedge-transfer rights, test reserve requirements, identify the structural gap and rehearse the precise objections an investment committee will raise before capital outreach begins.

The Four-Gate Takeout Test

Carry four questions into every mini-perm discussion: Balance, Cash, Contract, Permission.

Balance: Will repayment and cash sweeps leave a debt balance that a new lender can actually fund?

Cash: Does downside cash flow still clear the lender’s DSCR after higher rates, lower availability or price haircuts?

Contract: Do revenue agreements and hedge protections outlast the new debt, and can the asset physically deliver what it has promised?

Permission: Do the financing, PPA and concession documents allow cash retention, new lenders, revised maturity and hedge changes?

Fail one gate and COD may still arrive. The refinancing may not.