We have sat across the table from more than sixty pre-seed hardware teams in the last eighteen months. About a third of them got funded inside six months. The remainder did not — and in almost every case the reasons were predictable and avoidable.
The cap table mistake that kills 40% of teams
The single most common failure pattern we see is a four-way even co-founder split made in the first two weeks, before anyone knows who is actually going to build the company. In software this is recoverable. In hardware — where the CTO will spend two years in a workshop and the CEO will spend two years on a plane — it is almost always fatal by month fourteen.
A 25/25/25/25 cap table is a promise that nobody is going to work harder than anyone else. Investors read it as a lack of conviction about who the founder actually is.
What works instead
Use a 60/25/15 split with two-year vesting cliffs that differ per co-founder. It is uncomfortable to negotiate in the first month, and vastly less uncomfortable than the equivalent conversation in month fourteen when one person has been shipping at three times the velocity of the others.
First five hires
Most pre-seed hardware teams hire a second mechanical engineer before they hire their first controls engineer. That is backwards. Mechanical iteration gets cheaper every month thanks to better CAM and cheaper fabrication. Controls debt compounds silently and eats your runway six months before your next round.
- Hire 1 — Controls / low-level systems engineer (before the second ME)
- Hire 2 — A second ME, ideally one who has shipped a consumer hardware product
- Hire 3 — Firmware / embedded engineer with RTOS experience
- Hire 4 — Perception engineer, even if vision isn't on your v1 roadmap
- Hire 5 — Operations lead who has run a contract-manufacturing relationship before
Prototype velocity is the only real metric
Pre-seed investors are not underwriting your TAM slide. They are underwriting how fast you learn. The clearest signal we have found is the number of physical iterations completed per month during the first six months. Teams in the top decile hit four to six revisions of a meaningful subsystem every month. Teams in the bottom decile hit one.
That velocity is not about how much you spend. Teams at Prelude Robotics and Kaitenzushi Labs — both of whom we worked with through their first round — ran at a burn of under $90K/month and still out-iterated competitors burning $400K.
The 'demo vs. data' inflection
At about month nine, every pre-seed hardware team faces a choice: polish the demo for the next round, or stop iterating visually and start collecting interaction data. The teams who picked data every time went on to close Series A rounds at 3.1x the median valuation of teams who kept polishing.
What your first check should actually look like
A defensible pre-seed round for a hardware team in 2026 is $2.5M to $4M on a cap between $12M and $18M. Anything smaller and you will be back in market in nine months before your risk profile has changed. Anything larger and you will have priced out your Series A lead.
Reserve 35–40% of that round for hardware spend, not 60%. The teams that overspend on hardware at pre-seed are the same ones who run out of money before they have the data to justify an up round.