| Episodes | 62 |
| Mentions | 89 |
| Cited here | 27 |
| First — last | #80 — #715 |
| Top guests | Luke Iseman, Omer Kilic, Eric Klein |
| Related | kickstarter · venture capital · open source hardware |
A hardware startup is a new company whose product is a physical device, a structure that differs from a software company less in what the business must achieve than in how it must execute: build something people want, find a way to distribute it, find a way to market it and sell it are the same objectives in either domain.[163] The distinguishing constraints are two specific absences: there is no built-in distribution channel comparable to an application store reaching millions of users instantly, and there is no recompiling, because manufacturing has to physically happen.[268] The full inventory of work explains the difficulty, since a hardware company must deliver a product and a platform, software and electrical and mechanical engineering, new product introduction and contract manufacturers, then distribution, supply chain, logistics and multiple compliance certifications.[327]
Structural differences from software
The advantage hardware holds over software is defensibility through time: a software product can be displaced within a month by someone smarter writing a better one, whereas a product that took six to twelve months to develop is not easily knocked off its position.[87] A second advantage is that people expect to pay for hardware, so building a business on something customers already accept they must buy removes a problem software companies spend years solving.[268] Against that, at the earliest stages, ignoring per-unit cost as one should at that point, nothing in hardware need take more than about a week longer than the software equivalent, whether that is populated boards or a printed enclosure.[268]
The organisational difference is concrete. Within the first fifteen employees a hardware company often has to cover eight, nine or ten distinct disciplines that must work together effectively, which is a coordination problem a software company does not have.[495] The difference also shows up in the accounts, because a hardware company has a cost of goods sold, so the profit and loss statement contains materials as well as people, which changes every financial conversation about the business.[495]
Costs follow the same pattern. A software company’s costs are chairs, computers, snacks and people, while a hardware company adds tooling that can run to six figures, prototype boards that cost thousands to obtain quickly, and regulatory approvals.[295] At product level the figures can be much larger: one company’s pre-sales target determined whether it could continue at all, against non-recurring engineering costs running into millions for a single custom display.[475]
Hardware companies are also tied to physical locations, so an engineer’s employment options are limited to companies within reach of where they live, while a software engineer can in principle work from anywhere.[597] The hardware community stays small through selection rather than through lack of awareness: the work is hard and the comparative payout in software is better, so most people who could do it choose not to.[466]
Capital
The historical reason hardware was underfunded is margin. Selling a widget at twenty or thirty points of gross margin is a genuinely hard business, so both investors and experienced founders were cautious about it, and the sector underperformed as a result.[402] What changed was the definition rather than the economics: hardware stopped being a widget that sits in a house collecting dust and became the mechanism for putting software into someone’s life away from a screen, which is what made it investable again.[402] The falling-component-cost effect that produced the smartphone era subsequently ran in robotics, where motor controllers, battery technology and laser rangefinders all dropped far enough to open markets that were previously priced out entirely.[495]
Investor appetite remains narrower than the opportunity, with money going to consumer gadgets rather than heavy machinery even where the unglamorous market is demonstrably large, because one is fundable and the other is not.[151] The semiconductor version of the same constraint is that everyone starts without their own production, because even venture funding only covers the designers, the tools and the licences; owning production comes decades later out of revenue, if at all.[151]
Bootstrapping
The case for bootstrapping rests on the actual numbers: unless the work is unusually expensive, hardware development runs to thousands rather than hundreds of thousands, which is within reach of a few years of saving.[94] The limit of bootstrapping is scale rather than survival, since an order-of-magnitude jump in production capacity, such as moving from a small facility to one ten times the size, comes from investment because the margin on the existing product cannot fund it.[151]
There are three viable scales and the middle one is the trap. Small one-off projects work, and a very large raise allows hiring a team, but in between there is neither the money to hire people nor the option to ignore the work, so the founder does everything personally and continuously.[104]
Cash timing
The software model is cash light while hardware is cash heavy, so committing money to parts months ahead of revenue runs against how the rest of the company is operating.[509] Every hardware startup therefore needs someone in the role of reality bringer, insisting that fourteen-week lead-time parts be ordered now; that person is generally treated as difficult, because startups are structurally reactive and the deadline does not feel real yet.[509]
A successful crowdfunding campaign is the beginning of the risk rather than the end of it. Founders who have raised a few hundred thousand often believe they are through the difficult part, when the difficult part has not started.[327]
Accelerators
An accelerator application is worth completing even if it is never submitted, because the questions are ones a founder should be asking anyway: how the company will make money, how equity is divided, and what the founders understand about the market that competitors do not.[268] The terms of one well-known programme are roughly twenty thousand dollars for about seven percent of the company, with founders present for a three-month batch and as much or as little help as they choose to seek; some companies extract very little from it.[268] Reapplying after rejection with visible progress is read as a positive signal rather than a negative one, because it identifies a team that is executing regardless of whether the investor participates.[268]
An accelerator is not a source of hardware expertise. Attending one expecting to be taught how to build hardware is a misunderstanding of what is on offer; what such programmes can help with is scaling a company.[531]
Operational competence
Getting something manufactured is the difficult part rather than designing it, which yields a blunt readiness test: if nobody in the company knows what design for manufacture means, a hardware startup is probably the wrong idea.[115] Purchasing is a specific and underestimated gap, since a new hardware company generally knows nothing about the mechanics of buying components, which is precisely where an inexperienced buyer gets poor terms on a reel of parts.[190] Pricing is another, because taking the component cost and adding twenty percent for margin does not work, and margin calculation is its own discipline that founders routinely underestimate.[123]
The advice that follows from the breadth of the work is to identify and accept weaknesses and buy in help for them, since specialised areas such as antenna design are where a general hardware team will otherwise lose months.[295] A category of mistakes is nonetheless unavoidable rather than careless, including choosing the wrong corporate structure, misclassifying goods for tariffs and discovering compliance obligations late; the realistic posture is to expect them, pay the penalties and keep learning rather than to plan them away.[475]
Successful hardware founders overwhelmingly have prior industry experience rather than coming directly from education, a different pattern from software that reflects how much of the difficulty is operational rather than technical.[166]
Scale and outcomes
One manufacturing consultancy’s core customers build between five thousand and a million units, but the companies that reach that scale usually start far below it, frequently after false starts and financing that fell through.[451] The companies that succeeded were not distinguishable early: a brilliant team and an interesting product did not separate them from a hundred other startups with the same characteristics, and the traction came later and was not predictable from the starting position.[451]
The visible record is heavily distorted by survivorship, with successes visible and failures almost entirely invisible, while the proportion of startups that produce a genuinely useful product and succeed is very small.[190] On one practitioner’s estimate ninety-nine percent of hardware startups go under the radar entirely, so the sample available for reasoning is the visible one percent.[267]
Funding cycles
Easy money is a warning rather than an opportunity: when capital becomes readily available everyone enters the space, which indicates a phase in the funding cycle rather than durable demand for what is being built.[207] The corresponding question during a boom is what becomes of those companies when interest cools again, since the capital that made them viable is the thing most likely to disappear.[236] That cooling did arrive; the boom period ended, fewer people start hardware companies, and of those starting companies at all more go into software or financial technology.[550]
Choice of product
One viable category is the product that can be cloned easily but survives on brand rather than on technical defensibility, which requires a different funding profile from a product protected by its difficulty.[394] A gap runs in the other direction between what people need and what gets built: affordable transportation and cheaper energy are hardware problems people genuinely have, and almost nobody starts a company to address them, because the practical work is unglamorous.[298]
Scope can also be narrowed deliberately on the strength of experience. Returning to the field, Migicovsky listed everything he had liked and not liked about running a consumer hardware company and produced a much longer list of dislikes, an exercise that makes a narrower next attempt attractive.[715]