Financial models for hardware and manufacturing businesses

Software businesses can model working capital loosely and survive it. Hardware businesses cannot. Components are paid for months before the finished unit ships, the unit ships weeks before it is invoiced, and the invoice is paid on terms the channel dictates. That gap is the single largest financing requirement most hardware companies ever face, and a percentage-of-revenue inventory assumption erases it from the model entirely.

Cost built from the bill of materials

Unit economics start at the BOM: component by component, with the price break curve at each volume tier, assembly labour, test and yield loss, tooling amortisation, freight, duty, and warranty reserve. That produces a landed cost per unit that moves with volume and with the sourcing decisions your operations team is actually making.

A learning curve applies on top: unit cost declining as a function of cumulative volume produced, at a rate stated as an assumption rather than assumed silently. Channel margin then sits between landed cost and end price — direct, distributor and retail treated separately, because a blended gross margin across those channels is not a number anyone can plan with.

The cash conversion cycle, modelled on real timing

Inventory is modelled on production and shipping lead times, safety stock policy and the build ahead of a launch or a seasonal peak — not as a ratio. Payables follow supplier terms, including the deposits contract manufacturers require before a run. Receivables follow channel terms, which are typically worse than founders assume.

The output is a monthly cash conversion cycle and a working capital curve with a visible peak: how large, which month, and what has to be in place before it. That single chart has changed more financing decisions in hardware engagements than any revenue forecast.

Capacity, capex and step costs

Manufacturing capacity does not scale smoothly. Tooling, line additions and test equipment arrive as lumps of capex with lead times, and utilisation against installed capacity determines when the next lump is required. The model shows the step, its timing and its funding, rather than allowing volume to grow through a ceiling that physically exists.

Scenarios that reflect hardware risk

The downside cases that matter are specific: a certification delay, a component on twenty-week allocation, a yield problem in the first production run, a tariff change on the landed cost of the main assembly. Each is modelled as a switch with its cash consequence, so the conversation with a board or lender is about a quantified exposure.

The same structure supports a valuation and extends into a fundraising model when the working capital peak needs financing — which, in hardware, it almost always does.