The hangar doors were closed. The engines were cold. And yet an Airbus A220 was burning. airBaltic's loss of a grounded aircraft to a maintenance fire is a quiet kind of disaster — no mayday, no radar track going dark, no altitude to lose. But for the people who price aircraft risk, it's anything but quiet. **This is the first hull loss in the A220's operational history.** The type entered service in 2016. Nearly a decade of flying, and the first total write-off happened not at 35,000 feet but in a maintenance bay. That distinction matters enormously to how the loss gets processed. Aviation insurers separate ground damage events from in-flight hull losses in both categorization and premium modeling. But the deeper problem here isn't the category — it's the absence of comparables. Hull valuations for in-production narrowbodies typically anchor to prior total-loss settlements for the same type. The A220 has none. Actuaries building a settlement model are working without a floor. For a mature type like the 737 or A320, loss history is thick enough to price confidently. For the A220, underwriters must lean on proxy data from comparable-vintage narrowbodies, manufacturer book values, and lessor replacement cost estimates. The tension is structural: manufacturer book values tend to run higher than lessor replacement costs in a soft delivery market, and that gap is exactly where the negotiation lives. **airBaltic makes this harder to absorb.** The Latvian carrier operates one of the most A220-concentrated fleets on the planet — a deliberate single-type bet that other operators hedge against. Lessors still placing incremental ATR72-600 orders signal what fleet diversification looks like as a strategy. airBaltic chose the opposite: one type, one network logic, one training pipeline. One destroyed airframe isn't just an insurance claim — it's a gap in a system with no slack. Somewhere, an underwriter is now building the model that will price the next A220 that does go dark at altitude — and this maintenance fire is the only data point they have.