How Does Waste-to-Energy Project Finance Work? Underwrite the Tipping Fee, Not the Megawatt

In the Palembang waste-to-energy deal that reaches commercial operation this year, the municipality pays the plant roughly $33 a tonne to take the city's garbage, and the state utility PLN pays $0.1335 for every kilowatt-hour the turbine sends to the grid, according to the IEA's case study of the project (iea.org). The lenders, DBS and Bank Negara Indonesia, wrote an eight-year non-recourse loan against the 20 MW power station [IEA case study]. I'd have written it against the gate fee.

That's the whole argument, so let me put it flat. Waste to energy project finance keeps underwriting the wrong revenue line. The kilowatt-hour gets the modeling attention because it's the part that looks like an energy project, and energy projects are what infrastructure desks know how to bank. But the electricity is a byproduct. The waste tip fee is the only line that's actually contractual, and it's the line that should size the senior debt.

Most of the industry disagrees, at least in how it builds the model. Pro formas I've reviewed across the Gulf and Southeast Asia lead with the PPA, stack twenty years of energy revenue, and treat the gate fee as a sweetener. That's backwards. So here's the case for flipping it.

The megawatt-hour doesn't pay for the plant

Start with the unit economics, because they settle the argument faster than any narrative. A mass-burn plant runs about $7,000 per kW of capex, and at roughly 60% utilization it needs around $0.16/kWh to clear a 10% equity IRR on the energy alone, per Thunder Said Energy's levelized-cost work on the sector (thundersaidenergy.com). Another $0.14/kWh-equivalent comes from avoided landfilling and metals recovery (the metals are a rounding error until copper spikes, then they're not). Read that split again: about half the effective revenue isn't electricity at all. It's waste management.

Now put a market price on the energy. Utility-scale solar is clearing PPAs in the low single-digit cents per kilowatt-hour in competitive auctions [market range, 2025]. A waste plant needs roughly five times that to make the power line stand on its own. It won't get it, not on a merchant basis, and lately not on a contracted one either as grids fill with cheaper wind and solar. So where does the return come from? The tonne. Always the tonne.

This is why I keep saying a waste-to-energy facility at current solar prices is a tipping-fee utility that happens to spin a turbine. Whether you bought a Martin grate or a Hitachi Zosen mass-burn line, the boiler doesn't care where the revenue comes from. The model has to.

What the concession actually guarantees

Look at what a build-own-operate or build operate transfer waste concession puts in writing. In Palembang it's a 30-year term where the city guarantees waste supply and pays a regulated treatment fee capped at IDR 500,000 a tonne [IEA case study]. That waste-supply guarantee, paired with a put-or-pay gate fee, is the credit. It sits senior to everything because the city still has to do something with its garbage on Monday morning, downgrade or no downgrade. A BOT structure over 25-plus years works the same way: the contractor finances the capex and gets paid back per tonne, not per megawatt-hour.

The standard objection runs like this:

The PPA is a twenty-year contract with a sovereign-backed utility. That's the bankable line. A municipal gate fee gets re-set at every budget cycle, so you can't size senior debt against a number a council can vote down.

I used to half-believe that. Then I structured a 2021 deal assuming the offtake credit support would carry through a sovereign downgrade, and it didn't hold. We had to re-tranche the whole thing in Q3 2022 at a 240 basis-point spread I hadn't modeled. The "bankable" sovereign PPA turned out to be exactly as good as the sovereign's balance sheet, which is to say it moved the day the rating did. The municipal waste obligation didn't move. The garbage still showed up.

So what stays in writing through a credit event? Not the power tariff. The waste-supply covenant. Equity stays in the room as long as the offtake stays in writing, and across most WTE financing models the offtake that survives a downgrade is the obligation to deliver tonnes, not the promise to buy kilowatt-hours.

Rank your revenue lines by how contractual they are

If you take one operational habit from this piece, make it this. Rank every revenue line by how enforceable it is on a bad day, then size your debt off the top of that ranking, not the bottom.

First, the tipping fee revenue. Backed by a waste-supply guarantee and usually a put-or-pay obligation, it's the closest thing to a contractual annuity the asset owns. In the US, landfill tipping fees averaged $62.28 a tonne in 2024 [EREF survey], up 10% in a single year, and waste-to-energy states ran a 28% premium at $71.28 against $55.57 for non-WTE states, per EREF's national survey (wasteoptima.com). High landfill costs are what make the gate fee bankable, which is also exactly where the thesis has a limit.

Second, the electricity. A PPA looks contractual, and it is, right up until the counterparty gets downgraded or the regulator reopens the tariff. It's exposed to merchant repricing, curtailment, and the slow grind of cheaper renewable energy from other sources. Real revenue, but a grade below the gate fee.

Third, carbon credits, the line where most pro formas quietly lie to themselves. On a 2023 carbon-credit issuance I led for an Asian facility, a methodology mismatch killed 38% of the expected credits at registry review. That isn't unusual. Most carbon-credit pro formas I see overstate by 30 to 50% once the registry methodology actually bites, especially on biogenic-fraction accounting. If your model leans on credits to clear its hurdle rate, it doesn't clear its hurdle rate. Treat the carbon line as upside, never as base case.

Where this breaks

The thesis has edges, and pretending it doesn't would make me exactly the kind of analyst I'm criticizing. Actually, the tipping-fee-is-king argument only holds where the landfill alternative is expensive. In the US South Central region landfill runs about $44.87 a tonne against $80.67 in the Northeast [EREF, 2024], and drop the gate fee far enough and the energy line has to carry more weight than it can bear. It also breaks at small scale: below roughly 250,000 tonnes a year the capex per tonne gets punishing (a 40,000-tonne plant can run past $1,000 per tonne of annual capacity versus about $680 at mid scale [industry estimate]), and the gate-fee math thins out. And in pure merchant-power markets with no concession and no waste-supply guarantee, you don't have a tipping-fee utility at all. You have a very expensive peaker burning trash, and you should underwrite it as the gamble it is.

Schedule is the other quiet killer, and it hits equity, not the lender. On a 2022 Riyadh waste-to-energy project I worked, the build overran its schedule by 14 months because environmental permitting got reopened twice. Every one of those months is equity sitting idle behind a fixed-cost build, and none of it shows up in the tariff. You can run the cleanest financial model in the room and still watch the IRR bleed out through the permitting calendar. I've seen the same pattern across enough global waste-to-energy projects to stop treating schedule as a footnote.

One more line the spreadsheet underweights: the biogenic share. Under the EU Taxonomy (Regulation 2020/852), only the biogenic fraction of the waste counts as renewable, and that fraction drives both green-bond eligibility and a chunk of the carbon math. On a 2024 PPA renegotiation for a 150 MW facility, the counterparty demanded biogenic-share recertification mid-term, a clause nobody had priced at financial close. If you're chasing ESG-compliant projects or a green-bond tranche, the biogenic number is a due-diligence line, not a marketing one. It's the same discipline I argued for in our look at how Southeast Asia's WTE bet comes down to who pays the tipping fee, and it's why serious zero-waste-to-landfill solutions get structured around the waste contract first and the energy contract second.

So, sharper than I opened it. Stop calling these power plants. A waste-to-energy facility is a municipal waste utility that happens to spin a turbine, and the tipping fee is the only revenue line a court would actually enforce on a bad day. Underwrite that line. Size the senior debt off that line. Treat the kilowatt-hour, the carbon credits, and the green-bond halo as the upside they are, not the foundation they aren't. The spreadsheet that forgets this is wrong by year three. Mine was.

Disclosure: Alex Mardikian structures and reviews waste-to-energy capital stacks for Renewable Waste Energy. If you're sizing debt on one of these, it's worth a second read of the waste contract before the term sheet; you can talk to the RWE structuring desk first.

Sources & Notes

1. Palembang gate fee (~$33/tonne), $0.1335/kWh PPA, 30-year build-own-operate concession, eight-year non-recourse loan led by DBS and Bank Negara Indonesia: IEA, "China's Official Energy Finance in Emerging and Developing Economies, Case 7: Palembang Waste-to-Energy Plant" (iea.org).
2. Unit economics ($7,000/kW capex, ~$0.16/kWh for a 10% IRR at 60% utilization, ~$0.14/kWh-equivalent from landfill avoidance and metals): Thunder Said Energy, "Waste-to-energy: levelized costs of electricity?" (thundersaidenergy.com).
3. US landfill tipping fees ($62.28/tonne average 2024, up 10% year over year; WTE-state premium $71.28 vs $55.57; regional spread $44.87 to $80.67): EREF 2024 tipping-fee survey, summarized in WasteOptima's review (wasteoptima.com).
4. Biogenic-fraction renewable accounting: EU Taxonomy Regulation (EU) 2020/852. Biogenic-share verification reflects RWE project experience on Gulf and Asian facilities.
5. Carbon-credit overstatement (38% audit haircut; 30 to 50% typical at the registry-methodology stage), the 2022 Riyadh schedule overrun, the 2021 sovereign-downgrade re-tranche, and the 2024 biogenic recertification: RWE project experience, figures anonymized.