Some interesting news over the last week on financing against contracts. We all love a contract. It answers some of the most important questions before you finance a business: Is there a real buyer? Will they stay? And if they leave, whatâs left?
On October 27, Terradot closed a $34MM senior secured loan from HASI for enhanced rock weathering projects in the US and Brazil. The loan is underwritten against fixed-price, multi-year offtake from Frontier, the corporate carbon removal buyersâ coalition that includes Google, Salesforce, JPMorgan, Shopify and Stripe. The hard part of weathering has always been proving the tonnes are real, so the deal also carries carbon delivery insurance, arranged by Marsh with CFC, which covers the risk that the tonnes arenât delivered. HASI called it the largest disclosed debt financing for this method. Terradotâs CFO, Julia Sekula, said: âLenders only back what they can predict.â My read is that it took both a fixed-price contract and a bespoke insurance wrap to make this financeable. Marsh and CFC also previously wrapped Chestnut Carbon <> Microsoft carbon credit deal the same way. This feels a playbook already.
Now look at Heidelberg Materials. On October 22, it paused the GeZero carbon capture project at its Geseke cement plant in Germany, citing CO2 transport and storage infrastructure that isnât ready and the lack of a viable business case, even with âŹ191MM in EU Innovation Fund grants already secured. Itâs not the first pause, either. In August, Heidelberg put its C$1.36 billion Edmonton project on hold. That plant was planned as NAâs first full-scale cement capture facility, at 1 million tonnes/year. Its model assumed a C$170 per tonne carbon price by 2030. Albertaâs actual industrial price is C$95 this year, scheduled to reach only C$115 by 2030, and credits trade around C$40â45. Ottawa had offered up to C$275MM plus a 50% investment tax credit. That was real money for a well understood technology, and it still wasnât enough.
The two carbon removal technologies arenât the same, and weathering is arguably the harder one to verify. But one maybe had the better contract, while the other was relying on a policy price path.
Then thereâs enforceability. GM signed a 10 year rare-earth magnet purchase agreement with e-VAC Magnetics, which backed a roughly $500MM new plant in South Carolina with 90%+ its output sold under this contract. GM sent a termination notice on July 1, 2026, and VAC started arbitration on July 7. Energy Fuels, which is buying VAC for $1.9B, has since amended the merger agreement so the price can move with the arbitration outcome.
So what does âwill they stayâ look like? On October 22, Formic raised $50MM: a Series B led by Mercato Partners, plus debt from S2G Investments. Factory customers pay one monthly fee that covers robots, software, maintenance and financing, with nothing upfront. Formic runs hundreds of robots across more than 100 facilities and reports a 97% renewal rate. That renewal rate is probably what S2G is underwriting against: lots of small customers who choose to stay, instead of one big customer who can walk. This is an exciting deal as because it shows robot fleets becoming a financeable asset class that looks like equipment leasing, which I wrote about in the construction autonomy piece.
Over the years my IC drilled one habit into me: go one layer below the contract. Why does the customer want this? How much does it matter to them? And why do we believe theyâll keep paying once it stops being convenient?
That brings us to the last question, whatâs left if the buyer leaves. On October 27, General Compute raised a $100 million Series A, up from $60MM in May. It sells bare-metal inference capacity on SambaNova, Positron and Cerebras chips, and the round followed one unnamed customer contract worth $100MM/year. The debt came first. In July, Upper90, the lender that first took Nvidia GPUs as collateral back in 2021, put up a $400MM facility reportedly secured by SambaNova chips. Iâd imagine itâs hard to underwrite the value of those alternative chips on their own, so the lender has to take a view on whatâs left if the buyer leaves. Something like, inference demand is enormous, so if this customer walks, someone else probably wants the capacity. And when alternative chips become a commodity, an early financier taking that view and risk would expect to share meaningfully in the upside.
One might say weathering, carbon capture, magnets, factory robots, alternative chips have very little in common. But each financing came down to similar things. We spend a lot of time in this industry talking about technology. But sometimes, itâs the paper.
Building something where the contract or financing is the hard part? Iâm always up for a conversation. Reply or reach me at Meng@Fridaymorninglabs.com anytime. :)


