Hardware Startups: Why Physical Products Are Different and How to Survive It

Four constraints that define hardware

• Unit cost is the business model. A software feature costs nothing to serve to the next customer. A physical product costs the bill of materials, assembly, freight, duty and returns every single time. If landed cost is more than roughly a quarter to a third of retail, a retail channel will not work.
• Iteration is measured in weeks. A design change means new parts, new prints, sometimes new tooling. Teams that batch several questions into each prototype round move far faster than teams that build one change at a time.
• Tooling is a capital wall. Injection mould tooling for a modest consumer product commonly runs tens of thousands of dollars and takes six to twelve weeks. That money buys repeatability, not learning — so do the learning first.
• Inventory is cash. A production order converts funding into boxes. Forecast badly and the company is solvent on paper and broke in practice.

Where the money goes, stage by stage

How hardware startups actually fail

• Cutting tooling before the design is settled. Tool modifications are possible but slow and sometimes impossible; a steel change can cost more than the original cavity.
• Designing to a prototype, not to a process. A 3D-printed part that snaps together beautifully may be unmouldable, unassemblable at volume, or too slow to produce profitably.
• Discovering certification late. Radio, safety, energy efficiency and packaging rules shape enclosures and boards. Finding them after tooling is the classic expensive surprise.
• Crowdfunding the wrong number. Campaign pricing set from prototype cost, before landed cost and fulfilment are known, produces a funded company that loses money on every unit.
• Single-sourcing a critical component. One allocated microcontroller can stall an entire product line for months. Choose parts with second sources wherever the design allows.

The sequence that keeps a hardware startup fundable

Where the money actually goes

The cash cycle that kills companies

• Model growth in cash, not revenue. Build a 24-month cash schedule with deposits, freight and payment terms, not a monthly profit-and-loss.
• Delay tooling until the design is genuinely frozen. A steel change order costs less than a new tool but still costs weeks.
• Order the first run small even when the unit price stings. The cost of unsold inventory exceeds the savings on volume pricing.
• Qualify a second source for every long-lead component before it becomes the reason the line stops.
• Treat certification as a design input, not a checkpoint - the standards dictate creepage, materials and enclosure design.

Signals a hardware startup is on track

• The bill of materials cost is known within 10% before tooling is cut.
• Design for manufacturing review has been run with the factory that will actually build it.
• A functional unit built on production tooling has passed a real user test, not a demo.
• Compliance testing is scheduled with a lab and the pre-scan results are in hand.
• Return rate assumptions are based on a pilot batch in customer hands, not a guess.
• The team can state the landed cost, the wholesale price and the retail price, and the gaps between them hold at scale.

Key takeaways

• Hardware discipline is financial as much as technical; the cash cycle, not the engineering, decides survival.
• Freeze the design before tooling and keep the first production run deliberately small.
• Certification and freight are line items to plan from day one, not surprises at month twelve.
• Physical products still win where the value is inseparable from the object - and that is a permanent category, not a nostalgic one.

What hardware investors actually diligence

Diligence areas and the evidence expected

Evidence to collect during development

• Volume-based BOM quotes, refreshed each quarter.
• Yield and defect data from every build, however small.
• A documented DFM review with actions closed.
• Signed purchase intent from real buyers, not survey interest.
• A working capital model showing the cash cycle, not just burn.

Key takeaways

• Hardware diligence centers on unit economics, manufacturability and supply.
• Collect yield and BOM evidence during development, not at fundraising.
• Model working capital; inventory timing kills more hardware firms than burn.