One Harvard question pivoted Singularity Energy from battery control to carbon data
A customer question turned Singularity Energy from battery-cost optimization into carbon intelligence; a $4.5M seed and a utility customer followed.
What was changed
When Wenbo Shi founded Singularity Energy around 2019, carbon was not on his mind: the plan was intelligent control of batteries and EV charging, with the objective of saving customers money. Working with the Harvard Innovation Lab on a battery-plus-solar installation, he was asked, 'Can you actually consider carbon as a signal?' The university wanted the battery not only to cut costs but to cut the campus's carbon footprint. Shi, a power-systems engineer, realized that if he could measure how clean or dirty the grid was hour by hour, carbon became another optimization objective.
The reframing changed the math. Analyzing the grid's carbon emissions hourly instead of using the common annual averages, Shi found that optimizing for cost alone would increase pollution. University of California, Davis researchers, working with his data, published a paper in April 2022 showing annual averages can overestimate the grid's carbon intensity by 33% in some cases and underestimate it by 22% in others, though most of the time they land within ±5% of hourly estimates — a gap big enough to matter for companies hunting emissions cuts.
Monetization was the struggle: the initial hypothesis that large companies would pay did not pan out, and utilities showed no interest until Eversource, the Fortune 500 New England utility with a net-zero-by-2030 objective, reached out to compute the carbon intensity of its line losses — about 5% of electricity is lost this way nationally. 'Singularity allows us to use the much better location-based calculation,' said Vandan Divatia, the utility's VP of transmission policy, adding that Eversource would begin using the data state by state and could eventually go customer by customer.
Why it worked
A customer's framing question redefined the company's objective function from cost to carbon.
Hourly grid data showed that cost-optimal charging decisions are not carbon-optimal, giving the pivot technical substance.
State net-zero mandates put utilities under pressure to cut emissions, turning them into paying customers for carbon intelligence.
Academic validation from UC Davis made the data credible in a market full of rough emissions approximations.
What can be applied
The customer question that breaks your framing can be the pivot: a side constraint like carbon can become the whole product when real-time data proves the common industry averages wrong.
Aftermath
Eversource planned to roll Singularity's data out on a state-by-state basis, potentially down to individual customers, while Shi argued grid decarbonization is impossible without utilities because they own the system and see the kilowatt-hours flowing through it. The next step, he said, is turning carbon visibility into action, with end-users back in the picture later.