This week, researching some very cool startups, I came across a few pain points. Enjoy.
In 2024, machinery damage or failure accounted for over half of all shipping incidents globally — 1,860 out of 3,310 reported casualties. Fire and explosion came in at 250. Collisions at 251. The single most common thing that goes wrong with a ship is the machinery inside it.
So how well do underwriters actually understand what's happening inside the machines they're insuring?
Not well. Not well at all.
Hull premiums are still calculated using fleet age, classification records, flag state, and claims history. The global hull premium reached $9.67 billion in 2024, insuring a fleet valued at an estimated $1.54 trillion. That's a lot of money changing hands based on backward-looking data.
There's no live connection between how a vessel actually performs at sea and what the owner pays for cover. Two identical ships on the same route, same age, same flag — one meticulously maintained, one running on borrowed time — same premium.
Without performance data flowing back to anyone, operators are stuck in a binary. Over-maintain and you burn cash on servicing that wasn't needed. Under-maintain and you're gambling on breakdown — which, as we've established, is exactly what's driving the biggest claims.
Break it down further: main engine damage alone accounts for over 60% of machinery claims cost, followed by auxiliary engines at 20% and turbochargers at 10%.
These aren't unpredictable events. They're mechanical wear patterns that could be tracked, modelled, and anticipated. But they aren't. Not at scale.
Steel, spare parts, dry-docking, labour — all up, and still climbing. The cost of regular machinery breakdown claims is soaring, with repair costs directly eating into already thin margins.
Hull & Machinery insurance is marginally profitable at best. General rate reductions paired with rising claim costs can push the market back into loss-making territory in a very short space of time.
When your largest claims category is also the one you have the least real-time visibility into, inflation doesn't just hurt.
IMO's Carbon Intensity Indicator ratings. The EU Emissions Trading System extending to maritime from 2024. FuelEU Maritime from 2025. Shipowners now have to track, report, and reduce emissions — and the ageing global fleet is making that harder, not easier. Delayed scrapping means older tonnage stays in service longer, which raises machinery claim frequency.
Many mid-sized operators don't have the digital infrastructure to comply efficiently. But here's what's interesting: compliance and insurability are starting to converge. The vessels that can't prove their efficiency will increasingly struggle to get affordable cover.
Regulation is doing what the market hasn't — demanding data.
Aviation has real-time telemetry. Agriculture has precision monitoring. Logistics runs on sensor networks. Maritime? It's decades behind.
But things are shifting. New entrants are appearing in the MGA space — Managing General Agents, companies that underwrite insurance on behalf of a larger insurer but bring their own data and their own pricing models. Names like AI Marine and Ceto.ai are already in the market. It's the model that disrupted motor insurance. The question is whether it can do the same for marine hull.
The gap between what vessel data exists and what underwriters actually use is enormous. Engine sensor feeds, AIS signals, maintenance logs, fuel consumption patterns — the raw material is there. Nobody's wiring it into the pricing.
Premiums that flex based on engine health, route efficiency, and emissions compliance — not last year's claims record. Maintenance schedules generated by the insurer, not the operator, because preventing a claim is cheaper than paying one.
A world where the best-run ships pay the least, and the data proves it.
AI is the obvious accelerant. Machine learning models trained on engine sensor data could predict breakdown weeks before it happens — turning reactive maintenance into preventive maintenance, and turning preventive maintenance into a pricing signal. The same models that help an operator comply with CII ratings could feed directly into their insurance premium. One data pipeline solving three problems at once: maintenance, compliance, and cover.
That convergence is the real opportunity. The pain points aren't separate — they're the same problem viewed from different angles. Whoever builds the platform that merges them has something genuinely new.
But the risk landscape is about to get harder, not easier. Novel fuels like methanol, ammonia, and hydrogen introduce completely new hazard profiles. Insurers will be covering more complex machinery with less repair history. The fleet is getting older and greener at the same time — and both trends push risk in the same direction.