The Drop 2026 on funding the build-out

At the cement Ripple, a late-stage infrastructure investor told founders to stop celebrating their first-of-a-kind plant and come back when they had ten. Lenders will not take technology risk and ask for 10 or 20 plants first, Kyle McEneaney of The Schmidt Family Foundation said at another Ripple. Corporates all want to be the third, fourth or fifth pilot and never the first, said a founder making low-cost ammonia at a third. The capital that builds infrastructure is waiting for plant ten. Plant one is where start-ups get stuck.

The Drop framed this theme as a question of speed, whether data centres, energy systems and factories will arrive fast enough to justify what is being spent on them now. The 19 Ripples, live podcasts and Expert Sessions gathered here mostly asked something earlier, who pays for the first one. They did not agree on whether the obstacle is money at all.

Two times or twenty

Shakil Karim of Actia Capital Partners gave the arithmetic at an afternoon Ripple on infratech. Infrastructure investors count on an almost guaranteed two to two and a half times their money, and a seed investor needs a credible path to 20. A first-of-a-kind project should target three to five times while it piles up risks, from the technology to a single offtaker. When Shakil asked for an example of that middle done really well, nobody offered one. On New Wave, Jerome Ternynck described the same trap from the company's side. A start-up that has de-risked its technology meets venture investors who cannot make 50 times their money on a 50 million valuation, and private equity and debt that find it too early.

Whether the gap is a shortage of money divided the sessions. The capital for the next stage is not really missing, only scarce, Julian Ryba-White said on Climate Drift. Jerome called capital commoditised at every layer of the stack. A think-tank researcher at the FOAK cliff Ripple said the money for the middle stage exists but goes to the projects with the lowest credit risk. Kyle disagreed that the best companies find capital, pointing to strong climate companies raising oversubscribed rounds that nobody will lead or price. Asked for the biggest risk, Marie Ekeland said money is not pouring into these sectors.

The theme's own build-out explains some of where it does pour. Outside AI and data centres, one investor at the pre-seed Ripple saw little chance that anyone would fund the next round. Thomas Kristensen of LGT Capital Partners described firms raising funds so large that they need several exits of $50bn to $100bn, pushing prices up. Now and then the build-out is the buyer. During it, Abhiram Kannan said, no cost is too high for chipmakers recovering xenon and krypton from air.

A word that sounds like risk

The investor at the cement Ripple had a diagnosis. "This is a language problem, not a money problem," they said, adding that 90 of every 100 companies do not know how to talk to infrastructure capital. Julian traced part of it to one word. FOAK works with venture investors, but to the capital that pays for plants it sounds very risky, while a first commercial deployment isolates the risk. Sam Cash said capital wants to underwrite a theme it already sees, and suggested a climate company might do better calling itself a resilience company. Concrete made Biozeroc's first pitch easy, Liv Andersson said, because everyone has touched it. Now the start-up has to explain a catalysis step inside someone else's plant.

The same investor allowed for something harder. "It is possible that we have the thesis wrong," they said. If a material has to pass from one buyer to the next with a premium attached, no amount of work will solve it. A commercialisation-stage investor at Kyle's Ripple judged the appetite for a green premium in most of the world to be zero.

Other people's balance sheets

Much of the practical advice was about splitting risk and passing pieces of it on. A waste-sorting deal that IFM Investors liked stalled at its investment committee because waste the company could not control and a pyrolysis supplier it had never used were stacked on top of the technology. In hindsight, Adelaide Morphett said, a delayed draw that financed only the sorting might have sequenced the risk. Julian splits a hard-tech business into a technology company, a developer and an operating company, and treats each project as the unit to be financed. The direct-air-capture founder at the FOAK cliff Ripple gave up trying to fill every role and formed a special-purpose vehicle with an experienced developer.

Big companies proved unreliable counterparties. Bailey Morrow of HSBC Innovation Banking leaves the offtake out when sizing a loan, because an offtaker once walked away. At the moat-or-museum Ripple, industrial pilots were described as still running two or three years on, and Maria Wasastjerna tells every founder to charge for them. When one of a start-up's five corporate investors lost interest after changing its chief executive, its founder said that was fine. As the only strategic backer, it would have killed the business.

Clocks that do not match

Speed returned as a mismatch of clocks. A Planet A company has spent a year negotiating an NDA with a DAX company. Nick de la Forge said Chinese suppliers deliver in about a third of the time at a third of the cost, and Liv gave nearly the same fractions for deploying in China. CTRL-S, which buys the IP of wound-down climate start-ups, finds the technology often sound. "The IP never failed, but it was just the wrong moment in the funding cycle," Silvan Aeschlimann said. In one case, a corporate partner wanted six months to decide when the start-up had one month left. Inside a corporate, one furnace experiment waited about 20 years to become a spin-out.

Fund lives set the pace further up. Marie built 2050 as an evergreen fund because a ten-year closed-end fund forces an exit in seven or eight years, while Paebbl, one of its companies, expects its real performance five to ten years from now. A US climate GP at the pre-seed Ripple runs 15-year funds. In Europe, another participant said, LPs walk out when a fund mentions more than ten. Evan Finkel said a GP's record becomes statistically stable about when they near retirement, so Integra Global Advisors tracks several KPIs over time. Thomas uses five to seven years of good picks only as a filter.

At the afternoon infratech Ripple, Toba Spiegel asked what the ideal investor for the middle would look like, and how it would win over LPs without the usual venture pitch. No one had an answer. Partial ones came up during the day. Borja Gumuzio Morenés's fund buys hardware into separate vehicles, Jerome built Rypples as a holding company, 2050 is evergreen and steward-owned, and CTRL-S keeps failed start-ups' knowledge intact until the timing comes right. 2050 has 60 million, which Marie called still too low. Whether patient money of that kind can pay for the first plants of a build-out this size is the question these sessions leave open.