Climate exposure starts upstream, where you can't see it

The problem

Most companies model climate risk where they have coordinates: owned facilities, distribution, logistics. Raw material sourcing sits three tiers up, unmapped, and that is where ecosystem depletion and climate sensitivity actually bite. A business can hold a well-modelled logistics risk profile and still not know that a critical input comes from a single basin under severe water stress. The blind spot is structural: assessment follows the org chart, not the material flow.

Why your existing tools don't answer this

Hazard platforms need coordinates. You give them a list of assets, they return flood, heat and drought scores against it. That works for what you own. Nobody sells the hard part, producing the location list for the farms, mines, mills and processors you don't own and were never asked to track. We build that layer. It sits underneath the hazard analytics you already pay for and makes them answer a question they currently can't.

Hazard is only half the exposure

Flood and heat layers will not tell you that your leather depends on pasture that is degrading, that a fibre depends on an aquifer drawn down faster than it recharges, or that a crop depends on pollinator populations in decline. Those are dependencies on natural systems, not exposure to weather, and they sit outside every hazard model on the market. Pricing them is natural capital work, the methodology behind the first Environmental Profit & Loss rollout across Kering.

Why it matters

• Cost: supply interruption and input price volatility hit margin directly, this year.
• Timeline: a current operating risk, not a 2050 scenario.
• EUDR: from 30 December 2026, large and medium operators must hold plot-level geolocation for cattle, cocoa, coffee, palm oil, rubber, soy and wood. You are already funding the map. This turns it into a risk answer.

The approach

Resolve the supplier estate first: one supplier appears as a dozen different name strings across ERP, procurement and customs records, and material origin sits in bills of lading, mill lists, certifications and audit PDFs in several languages. Frontier models read and reconcile that at a scale that was not economic a year ago, then geolocate to site level and join it against hazard, water, land use and ecosystem data. The point is breadth: every supplier gets screened, not the twenty someone pre-labelled critical, because the scoping is the blind spot. The machine does the volume. The judgement about which dependencies actually threaten the business is ours.

What you receive

A ranked exposure register: input by origin by hazard and dependency, each line confidence-graded against the strength of the underlying data, with mitigation options attached, dual sourcing, requalifying an alternate origin, forward cover, inventory buffering, supplier investment.