Development banks are pricing transition risk

The World Bank and BNDES put $2.86 billion behind Brazil's hard industries this week, and the round number is already doing what round numbers do, which is stand in for the deal itself. The figure that will decide whether the package was a transaction or an announcement is the $1.06 billion of public money inside it, a slice that exists to do something the commercial market has so far avoided: put a price on industrial decarbonization assets that have no single owner, and then find lenders willing to accept it.

The public share is not a subsidy attached to a deal; it is the deal's first instrument, sized and structured so the rest of the capital stack has something to trade against. That distinction gets lost whenever a development bank announces money, because the announcements are written to sell the commitment rather than explain the mechanism. A first-loss slice is a price before it is a gift, and Brazil is the cleanest test: the asset class is hard, the public tranche is large, and the syndication either happens or it does not.

The mechanics are routine inside development finance and rare outside it: a multilateral takes a first-loss position, or writes a guarantee, and the risk that no commercial lender will underwrite moves onto a balance sheet that can hold it. What the public money buys is a reference point, the terms on which an asset can be traded, rather than the asset itself. With the anchor in place, a syndicate can be assembled on conditions the market can read, and an industrial project that had no single owner acquires a group of them. Whether that group materializes has a public answer: names on a loan.

Hard industries are the awkward case because the assets resist the categories lenders use: a hydrogen or industrial-decarbonization project in Brazil carries a commodity-linked revenue stream, a construction phase that outlasts most bank mandates, and no single utility or sovereign standing behind it. Those are exactly the features a public anchor is meant to neutralize, and exactly the features that leave the private market waiting for someone else to move first.

Guarantees and first-loss slices do different work, and the difference decides which one gets repeated. A guarantee is contingent capital, called only if the borrower fails; a first-loss slice is money committed up front and subordinate to everyone else. The guarantee crosses borders more easily because the public lender never owns the downside, while the slice sends the stronger signal, since it sits in the same collateral pool the commercial banks are pricing. The $1.06 billion, rather than the $2.86 billion, is the number to track.

The EIB's first nuclear loan is worth $46.18 million to a Finnish reactor startup, and as a loan it barely registers; as a precedent it is the larger transaction. The EU budget does not fund nuclear, and the loan is built to work around that constraint, with a guarantee doing the load-bearing work — a credit enhancement that moves risk onto the institution without the institution taking ownership of the project. The $46.18 million is what shows up in the table; the guarantee is what the next deal copies.

Small loans carrying large guarantees are how a development bank moves a new asset class through its own mandate, and for nuclear the EIB has now produced a structure other lenders can cite. The economics favor repetition: a guarantee consumes less of a public balance sheet than an equity stake and is easier to standardize than a one-off concessional package. If the template holds, the next nuclear financing should arrive with a commercial tranche already attached, because the guarantee has told lenders where the floor sits.

There is already a finished version of the structure to point at. The Gennaker financing, a $3.5 billion close, required two things before commercial lenders would take construction risk — the EIB inside the syndicate and a municipal utility holding 25% of the equity — and only then did 16 commercial lenders come in. That sequence is the template: a public lender in the deal, a local anchor on the equity, and private capital arriving once both were in place. The commercial banks came in behind two institutions that had agreed to hold the risk first.

Both cases point at the same gap: nuclear and heavy industry are capital-intensive, politically exposed, and hard to hedge, which is why commercial lenders have stayed out, and the public anchor is the substitute for the hedge that does not exist on their side of the table. The EIB bought a template; the World Bank and BNDES are trying to buy a market. The second is the harder purchase, and it is the one with a scoreboard.