The deck said seed round, €700,000, eighteen months of runway. The plan underneath it said something narrower: get to working hardware, then use the working hardware to raise the next round. That reads fine on a slide. What it hides is that the working hardware arrives in month nine, the raise takes five months on a good day, and somewhere in the middle the factory that was building your boards stops answering within a day and starts answering within a week.
Programmes die in that gap more often than they die of engineering. The team is good, the product works, the money simply ran out at a point where stopping was expensive and nobody had priced the stop.
So the rule we give clients before anything gets kicked off: secure the money for a complete development phase before you begin it. Not enough to reach the gate. Enough to reach the gate, sit through the results, decide what they mean, and carry on. If you cannot fund that, do a smaller phase rather than a partial one.
The whiteboard in the factory office
Founders imagine that a factory relationship is a contract, and that a contract holds still while you go and find more money. It is closer to a queue you have to keep standing in.
Case Engelen of Titoma, who has run this loop from Shenzhen and Taipei for two decades, describes it plainly: “Every factory they have their little whiteboard with their top priorities to work on this week. That little design project for an unproven customer is very often on the bottom of that ranking.” The mechanism is arithmetic. Larger manufacturers staff new projects from shared resources, so the engineer assigned to your NPI build is also assigned to three other programmes. Engelen again: “Cisco has a big problem, then for the next two, three weeks you’re not going to get your emails answered anytime.”
From inside that building, a funding pause is indistinguishable from a customer losing interest. The customer who was sending DFM questions twice a week has gone quiet. No new drawings, no sample orders, no PO. The project manager has three other accounts making noise. Your build slot in next month’s NPI schedule is worth money to somebody else and gets given to them. When you come back in March with fresh funding and the same enthusiasm, you are not resuming. You are re-entering the queue, behind everyone who kept moving.
Titoma’s own guidance on what makes a factory deprioritise a project is worth reading as a checklist of things a funding pause does to you automatically. Projects with no clear path to repeat business get lower priority. So do projects where the PCB, enclosure or firmware keeps changing every week, because instability reads as immaturity. A team that stops for four months and comes back with a revised design because they used the downtime to “improve things” has managed to trigger both.
Then there is the paperwork, which expires on its own schedule. Quote validity has been shortening across categories: raw materials now commonly 7 to 14 days, electronics and technology hardware 14 to 30 days, tooling and construction materials 7 to 21 days, with capital equipment the outlier at 30 to 60. Once a quote lapses the supplier owes you nothing, and in commodity-exposed lines a two week delay has been enough to move a price by 5 to 15 per cent. A three month pause does not preserve your BOM cost. It resets your entire quotation package, and every requote is a fresh negotiation with a supplier who now knows you stopped.
None of this is recoverable with charm. The way back in is the same way you got in: a funded customer who can commit to a build date.
The bill for standing still
Do the arithmetic on your own numbers, because it is more uncomfortable than the story you tell yourself about a “quiet period”.
Take a small connected-product team: four engineers and a part-time programme manager, loaded cost somewhere around €45,000 a month in the Netherlands once you include employer charges and the design partner’s retainer. A three month funding gap is €135,000 spent on a programme that produced nothing, and that is the cheap part. You will also spend the first four to six weeks after restart re-establishing what everyone knew in October: rebuilding factory contact, re-issuing RFQs, re-running samples against a supplier who has swapped a component you qualified. InnoComm, an ODM that publishes on why devices stall between prototype and mass production, puts a design or tooling change at three to six months of schedule against two to four weeks for a software-only fix. A restart is not a design change, but it lands in the same part of the calendar.
The version of this that hurts most is the one where the team does not sit idle. They keep working, because engineers do, and without factory feedback or sample builds they optimise. The design moves. Now you have burned three months of salary and arrived at a design that no longer matches the quotes you were holding, and the factory has to start its DFM review again.
If you must pause, pause deliberately: freeze the design at a named revision, tell the factory a date and honour it, pay for something small so you stay on the books, and accept that the schedule restarts rather than resumes. That is damage control. It is not a plan.
Size the phase before you sign anything
Founders underestimate development budgets because they price the engineering and forget everything the engineering triggers. Two practitioner sources are useful for sizing, and they are useful because they disagree about scale while agreeing about shape.
Titoma puts total development for an electronic product, taken through design, engineering, prototyping, injection moulding, testing and CE/FCC certification, at “around US$50,000 to millions”. Their own typical fees give a floor: roughly $5,000 for electronics architecture, $5,000 for industrial design, $10,000 to $15,000 for mechanical engineering, around $15,000 for firmware, with a minimum certification budget of $20,000 and individual certificates running $1,000 to $20,000 each. Prototypes ladder up from $1,000 for a proof of concept to $8,000 to $18,000 for pilot units off soft tooling.
Glencoyne’s NPI budget framework prices the same journey higher and by gate, which is the more useful shape when you are deciding how much to raise. Their worked example puts the EVT stage at about $35,000 ($25,000 of prototype units, $10,000 of lab equipment), DVT at about $55,000 ($50,000 of pre-tooling units, $5,000 of certification pre-scans), and PVT at $135,000 to $143,000, dominated by $80,000 of injection mould tooling, $40,000 of full FCC and CE certification, and $15,000 of assembly jigs and fixtures. They recommend a 15 to 20 per cent contingency on NRE on top. Note that this excludes engineering salaries entirely, so it is the bill that arrives alongside your burn rather than instead of it. Both frameworks also price certification light: our certification post puts EU plus US market access for a connected device with a pre-certified radio at €60,000 to €120,000 all-in, and that is the line to carry into your model.
Neither set of numbers is your number. What both make clear is that the money is back-loaded. Roughly two thirds of the non-recurring spend in that framework sits in the last gate, which is exactly the point at which a team financed to reach EVT has nothing left. If you have read our piece on who owns the calendar, you already know what those gates test. The financing question is different: the cheapest gate is the first one, and the phase most founders fund is the one that costs least.
So build the budget backwards from a decision point rather than forwards from a milestone. Ask what you need to hold in the bank on the day the DVT results land, including the cost of the answer you do not want.
Money that arrives in slices
Tranched investment is common in European deep tech and it is not automatically a bad deal. FounderCatalyst’s guide describes the usual shape: an initial tranche, say £250,000, with a larger amount released on demonstrated progress. Public co-investment schemes often work this way by design.
The trap is what the second tranche is conditioned on. If the milestone is “complete EVT”, you have signed up to a structure that funds the cheapest part of your programme and puts a renegotiation directly on top of your most expensive one. Push milestones to sit after a decision rather than after a build. “EVT complete plus reliability results reviewed plus DVT scope agreed with the CM” is a milestone that releases money when you actually need it. “EVT complete” releases it two months early and then leaves you exposed.
The market data explains why this matters more in 2026 than it did five years ago. The gap between primary rounds has stretched: Carta measured a median of 696 days, roughly 23 months, between rounds in Q2 2025, against around 600 days two years earlier. Bridge rounds took 16.6 per cent of all cash raised on Carta in Q2 2025, up from 11.8 per cent a year before. Blended estimates from Carta, PitchBook and AngelList reporting suggest around 38 per cent of seed-funded startups now raise an extension before they ever see a priced Series A, and that only 15 to 20 per cent reach a Series A within two years, down from over 30 per cent in the 2018 to 2020 vintages. Those are directional figures from a mostly software dataset, and hardware timelines are longer, not shorter.
The good news buried in the same data is that repricing has calmed down: Carta put down rounds at 11.4 per cent of new rounds in Q1 2026, against a 22 per cent peak in 2023. Raising again is survivable. Raising again on a schedule set by a factory that has stopped waiting for you is the problem.
Crowdfunding will fund your campaign before it funds your product
Crowdfunding can genuinely finance a development phase. We covered elsewhere why the platforms have consolidated around established brands, so treat this as the mechanics underneath that finding: what the money does on its way to you.
Start with the fixed leak. Kickstarter takes 5 per cent of total funds raised, and payment processing takes 3 per cent plus $0.30 per pledge, with small pledges charged at 5 per cent plus $0.08. Nothing is charged if you miss the goal.
Then the variable leak, which is much larger. LaunchBoom, one of the bigger campaign agencies, tells creators to expect to spend 15 to 25 per cent of whatever they raise on ads. Jellop, a Kickstarter Premier Partner that has worked on over 7,500 campaigns, reports that most creators spend around 10 per cent, usually 7.5 to 12.5 per cent. Jellop’s own commission runs about 15 per cent of pledges it can attribute to its work, which they say typically lands near 5 per cent of a project’s total raise. Full-service agency engagement at LaunchBoom starts at $25,000 plus a 5 to 15 per cent commission. A professionally produced campaign video runs $5,000 to $15,000 before anyone has bought a single impression.
Stack it and the honest working assumption is that somewhere between a quarter and a third of your headline number never reaches the product. Jellop’s own guidance on minimum viable ROAS assumes roughly 20 to 33 per cent cost of goods, about 15 per cent ad management, and about 10 per cent platform and payment fees, which is why they put the break-even at around 2 to 2.5 times return on ad spend.
The part that decides the outcome happens before the campaign is live. Jellop’s analysis of 5,207 projects found paid promotion accounted for an average of 41 per cent of total pledges, with 80 per cent of projects falling between 20 and 60 per cent. The remainder comes from the platform’s own traffic and, decisively, from the list you built yourself. LaunchBoom’s published campaign data shows what building that list costs and returns: reservations taken with a $1 deposit came in at $8 each on one campaign and $35 on another, with higher deposits costing proportionally more ($42 per reservation at $50, around $120 at $100). Those reservation lists converted to backers at 22.34 and 33.24 per cent. Their two worked examples turned lists of roughly 23,000 emails into $358,602 and $813,019 of direct sales at 9.1 and 8.7 times return on ad spend.
Read those numbers as a job description. You are running a paid acquisition business for three to six months, with a funnel, a cost per lead, a deposit product and a conversion rate, and the hardware is the thing you sell at the end of it. Founders who like that job do well at it. Founders who expected a funding event get a marketing bill and a partially funded programme.
If you go this route, budget it as a phase with its own cost line, and set your funding goal against what lands in the bank rather than what appears on the page.
The number to write down
Before kickoff, work out one figure: the cash required to carry the team, the factory engagement and the NRE bill from today through to the far side of the next gate decision, plus the contingency, plus the months your next raise will take. If that number is bigger than what you can secure, cut the scope of the phase until it fits. Ship an EVT-only programme with an honest EVT-only budget rather than a PVT ambition financed to month seven.
The factory will forgive a smaller project. It will not wait through a quiet one.
Sources
- Titoma: NRE costs for electronic product development
- Titoma: Electronics manufacturing in China, challenges and advantages
- Titoma: How to approach a manufacturer for mass production
- Glencoyne: Hardware NPI budget framework, concept to mass production
- InnoComm: Why smart device projects stall between prototype and mass production
- AuraVMS: Supplier quote validity periods and price lock agreements
- Predictable Designs: What it really costs to launch an electronic product
- FounderCatalyst: Guide to raising tranche-based investments
- Causo Hub: Down rounds and bridges in 2026, source by source
- Value Add VC: Seed extension rounds in 2026
- Kickstarter: Fees
- LaunchBoom: How much does a Kickstarter campaign cost in 2026
- LaunchBoom: How to build a pre-launch email list
- Jellop: FAQs on fees and ad budgets
- Jellop on Kickstarter: Beyond ads, what creators ask us most