"How Much Is Normal" Is the Wrong Question

Founders ask what a typical hardware seed round looks like, hear a number, and reverse-engineer a plan to fit it. That gets the causality backward, and experienced investors spot it in the first five minutes. The amount you raise is an output of a plan, not an input to one.

The question that has an answer is narrower: what specific, verifiable thing will this company be able to prove when the money runs out, and what does getting there cost? Answer that and the raise size falls out. Fail to, and you either underraise and stall six weeks short of a working tool, or overraise at a valuation you cannot grow into.

Raise to the Next Milestone, Not to Profitability

Nobody funds a physical product from concept to profit in one round. Money arrives in tranches, and each buys the removal of one category of risk. The valuation step between rounds is what the market pays for that. For hardware the ladder is consistent:

  • Technical feasibility. A works-like prototype proving the core mechanism does what you claim, plus evidence people want it. Retires "can it be built."
  • Design maturity. A design-validated unit from production-intent parts, passing its safety and emissions tests, with a BOM costed by a real factory. Retires "can it be built repeatedly at a price."
  • Manufacturing readiness. Tooling cut, a pilot run through the line, yield and defect data in hand. Retires "can it be made at volume."
  • Commercial traction. Units sold to strangers at a margin. Retires "will anyone pay," and it is what a Series A is buying, in the sense of what counts as traction.

Pick the milestone you can credibly reach, then price the path to it. Funding two rungs at once usually means raising too much at a stage where your valuation cannot support the dilution, a tradeoff worth understanding through equity dilution.

Build the Number Bottom-Up

A use of funds that reads "engineering, marketing, operations" is a red flag. Build it line by line, in the order the money gets spent.

Development. Industrial design, mechanical engineering, electronics, firmware, and any companion app, each quoted rather than guessed. Cross-check against what it costs to develop a new product; a number far below the range means something is missing from the scope.

Prototypes and iterations. Not one build. Plan on three to five hardware iterations, each with printed or machined parts, boards, and assembly labor. Teams budget the first and get surprised by the rest.

Testing and certification. Emissions and safety testing, battery shipping qualification, and category-specific approvals. A connected consumer product commonly runs $15,000 to $60,000 across certifications, and a retest is not free.

Tooling. The largest discrete line item and the one that most often blows the round. Injection molds run $8,000 to $80,000 each depending on size, cavitation, and steel, so a product with six molded parts can absorb $150,000 before a sellable unit exists.

Pilot production and first inventory. A pilot run of a few hundred units, then a first purchase order the factory wants a deposit on. Inventory is working capital, not expense, and founders routinely leave it out entirely.

Payroll. A four to six person hardware team in a US market runs $80,000 to $130,000 a month fully loaded. Multiply by the runway, and pay the founders something. Unpaid founders are a burn rate that shows up later as a resignation.

Everything else. Patent filings, entity and contract legal work, product liability insurance, freight and duties, accounting, software. Individually small, collectively 10 to 15 percent of the round; the full catalog is in the hidden costs of hardware development.

The Reserve: 25 to 35 Percent, Not 10

Add up the bottom-up plan, then add a quarter to a third on top. Say so openly in the pitch; the alternative is worse. Hardware schedules slip for structural reasons, not sloppy ones. First mold trials rarely produce acceptable parts, and each correction adds one to three weeks. A certification failure means a board respin and a retest queue. A component goes end-of-life and forces a redesign. A factory closes for a month over a national holiday. Any one costs six weeks of burn, and a typical program hits two or three.

The runway is longer than the work, too: plan for the work, plus three to six months of operating room after the milestone, plus the four to six months the next raise takes. A round that funds exactly the engineering plan puts you on the road with no money and no milestone.

What the Ranges Actually Look Like

With the caveat that these are outputs, not targets, US hardware rounds cluster like this. Pre-seed and friends-and-family: $150,000 to $600,000, buying a works-like prototype and demand evidence. Seed: $1 million to $3 million, buying design validation, certification, tooling, and a pilot run. Series A: $4 million to $15 million, buying production launch, inventory, and a sales motion. The boundaries are laid out in pre-seed versus seed. Regulated medical devices and anything with a long clinical path sit well above these bands; simple mechanical consumer products sit below.

Not All of It Has to Be Equity

Funding everything from one priced round is expensive. Different costs suit different instruments.

Non-dilutive federal grants fund technical risk well: SBIR Phase I awards commonly run $150,000 to $310,000 and Phase II into the millions, six to twelve months from submission to money, which is why the SBIR application should start long before you need cash. Tooling and inventory suit debt better than equity, being assets against a known order, and once purchase orders exist purchase order financing funds production without touching the cap table. A sensible plan is often a modest equity round for the team and design, a grant for the hard technical piece, and debt or deposits for the first production run.

Five Ways Founders Get the Number Wrong

Minimizing dilution by underraising. A round 30 percent too small ends in a bridge at a flat or lower valuation, costing more dilution than raising properly would have. Raising for the perfect scenario. A model with no failed mold trial and no certification retest is a marketing document. Forgetting working capital. Profitable on paper and out of cash is the standard hardware death. Ignoring the raise itself. Four to six months of founder time and burn, during which engineering slows. Setting the amount from a valuation target. Deciding to give up 15 percent and back-solving produces a number unconnected to the plan, and investors ask about the plan.

Price the Plan Before You Price the Round

Projects House builds the technical side of the use of funds with founders preparing to raise: a scoped development plan, quoted tooling and certification, realistic iteration counts, and a schedule investors can test. Send your concept and milestone through our contact form.