Smarter power: Negawatts before megawatts

When SOSV looked at ground-source heat loops for heat pumps, one of Duncan Turner's analysts worked out that the owner would earn a 7% return. The analyst got 9% from the S&P 500, and asked why anyone would bother. Duncan told the story to a small table of investors and corporates, and it caught the difficulty the session kept meeting. Nobody doubted that the existing power system could deliver far more. The doubt was over who would pay for it.

Rushad Nanavatty, who runs RMI's Third Derivative accelerator, made the case for negawatts. RMI estimates that the US needs about 270 gigawatts of new peak capacity over ten years, and that 190 gigawatts of it could come from energy efficiency, virtual power plants and grid-enhancing technologies. Nearly every home has a water heater, which is a thermal battery. Across the US they add up to about 120 gigawatt-hours, almost exactly the country's grid battery storage. Seen as a machine, the grid runs at about 40% of its capacity. "There's so much more that we could be doing to squeeze more capacity out of the existing system before we start adding more generation," Rushad said.

Duncan took the opposite view on purpose, while noting that SOSV backs efficiency companies too. Venture capital has to start any such shift, and funds are built around the outlier, a Fervo that returns the whole fund, not a device hundreds of millions of households must adopt. Putting lots of kit on the grid raises rates when everyone wants cheaper power. And the recent big swings in storage and geothermal had one customer, the data centre.

Hardware first, then the channel

A growth investor agreed about the latent capacity and had seen two or three start-ups go after water heaters. They all had to work with US utilities, which earn a return on the assets they build and so are misaligned with saving power. The investor prefers hardware, because software is rarely a moat and most virtual power plant companies are struggling for being easy to replace. The channel matters as much. A home-battery company in the investor's portfolio was first seen as an add-on to residential solar, which in the US kills interest, and took off once it was repositioned to help hyperscalers connect inference data centres faster. The investor no longer calls theirs a climate fund but an energy AI fund, doing the same things under a new name.

Asked whether an as-a-service model reassures investors, Rushad said only with distinctive technology behind it. TS Conductor sells transmission as a service, but its edge is a cable carrying three to five times the current over the same wires.

There's so much more that we could be doing to squeeze more capacity out of the existing system before we start adding more generation.

Heat pumps pay back too slowly

Duncan said about 40% of the energy used in buildings is wasted through HVAC, from filters nobody changes to heating and cooling empty rooms, and optimisation can fix much of that. Heat pumps are the opposite problem because they add electricity load. A corporate investor from the heat pump industry agreed. Converting between electrical and thermal energy loses some, which limits heat pumps as flexible assets, and end users care about their bills, not flexibility. A home heat pump costs almost ten times as much as a gas boiler, and in Belgium and the Netherlands electricity costs three or four times as much as gas. Industrial heat pumps pay back in about ten years, while buyers expect three.

SOSV hated ground-source loops, Duncan said, and backed only one, because it was so cheap. Its rig is the size of a small European fridge and can be wheeled through a house, leaving a disposable drill head 200 metres down. That brings the payback to about three and a half years.

Policy follows technology

Rushad asked whether investors and start-ups should organise to push US utilities from cost-of-service regulation, which rewards building expensive assets, to performance-based rules that reward services. Duncan was blunt. "Policy follows technology. You can't do it the other way around." After the scar tissue of the last climate cycle, Duncan now backs only bipartisan directions, such as battery manufacturing driven by defence. The growth investor said their worst investments had been in batteries, the further upstream the worse.

The water heaters came back at the end. Two to two and a half kilowatt-hours in each European tank would add up to five times the grid batteries Europe has deployed, Rushad said. Three things have changed, namely the need for speed to power, much cheaper batteries and better computing. The growth investor said companies had tried before and been bought for pennies, and urged anyone trying again to find the customer first. Duncan struggled with installation costs that far exceed the price of the equipment, and thought industrial efficiency a better place to look.

The examples bore that out. An investor described a robot that cleans fouling off container ship hulls and saves 18% to 30% of fuel. The growth investor had backed a new airframe that cuts operating costs by 30–50%, and needed to convince only a few airlines, government buyers and public backers, not 500,000 households. One of the hosts closed with the lesson. Go after one big customer, or back a company that has partnered with one. It was an awkward ending for a session about negawatts, which live in millions of homes. The water heaters are already installed. What they lack is a buyer.

This Ripple was hosted by Rushad Nanavatty (RMI-Third Derivative) and Duncan Turner (SOSV) at The Drop 2026 on 16 September.