Battery intelligence: the missing layer in India's used EV market

India registered 2.45 million electric vehicles in the year to March 2026, a quarter more than the year before, and the majority were commercial: two-wheelers in delivery fleets, three-wheelers carrying passengers, and buses on state routes. A commercial vehicle is bought and sold on its economics, and the one input nobody can yet fill in is the residual value of the battery.

Source: Vahan via Autocar Professional and Autoguide India; the middle bar is derived as total registrations less two-wheelers and cars.

The battery is 30 to 40 percent of the vehicle's cost and the only component whose condition cannot be seen. A used petrol scooter is priced on mileage and condition. A used electric scooter cannot be, because the state of health of its pack, how much capacity is left and how fast it is fading, is known only to the software inside it, and that software belongs to the maker.

Four markets that depend on one number

• Resale. A commercial EV changes hands after two or three years. Without a trusted health score the buyer discounts the battery to zero and the seller's economics break.
• Financing. Lenders price the risk of the asset they cannot value. That keeps interest rates for commercial EVs high and is one reason financing, not vehicle supply, is the constraint on fleet electrification. A lender that can read the battery, which is the logic behind Exponent One and Astranova, has a structural advantage.
• Insurance and warranty. Pack replacement is the largest claim in an EV's life. Manufacturers price warranties without reliable ageing data, and insurers mostly decline to price them at all.
• Second life. The first wave of commercial EV batteries retires from FY27. A pack with 70 percent capacity left is a grid-storage asset, and Sheru has been building the energy bank that can use it. That only works if the pack can be graded quickly: the requirement is a diagnostic that takes 30 minutes rather than a full charge-discharge cycle that takes a day.

Why this is a seed-stage problem

The incumbents cannot solve it because their data is siloed by design: every OEM holds its own telemetry, every swap network its own, and none of them wants to expose the ageing curve of its packs to a competitor. The market needs a neutral grader, chemistry-agnostic, whose score is accepted by a lender, an insurer and a buyer who have no relationship with the vehicle maker. That is a software and data company with a physical test at its edge, which is a business a first institutional cheque can fund.

The technical debate in the field is whether the controller-area-network data a vehicle already produces is enough to estimate state of health, or whether a short physical test is needed for a defensible number. Our current view is that the CAN-only approach is where the market wants to end up and the physical test is where it has to start, because the first customers, lenders and resellers, need a number they can put in a contract.

What we would back

A team with battery science and fleet operations in the founding group, a grading protocol validated against warranty outcomes across at least three fleets of different makes, and a first customer who is a lender or an insurer rather than an OEM. We have the demand-side data from our own portfolio, charging sessions, swap telemetry, bus duty cycles, to test the claims before we invest. If you are building this, we would like to see the correlation between your score and real warranty claims. That correlation would be the core of our diligence.

Sources: Vahan registrations compiled by Autocar Professional and JMK Research (FY26); IESA India EV report (2026); AdvantEdge portfolio data.