The path-dependency (or difficulty of pivot) is an interesting angle. Certainly worth considering. Most product needs iteration if not outright pivoting. Perhaps the mitigating factor is that if the product is tangible, the need is also. SW is so powerful, as a concept, that i wonder if SW some startups may suffer from "too many degrees of freedom". Thus the ~inevitvble pivot -- combined with the ability to re-purpose a core of valuable skill-- is almost a more inherent characteristic of SW startup vs HW one. When considered in context of problem solving and product/market fit, &tc.
The issue, as I understand it, is not so much path-dependency per se, but the long lead times involved in the manufacture of new hardware at scale. Software has path-dependency too, but changing source code to modify a new software product is faster than changing physical machinery, processes, materials, suppliers, and logistics to modify a new hardware product.
This is a good comment, and sheds more light on the subject. The long lead time is a question of agility. This is subtly but importantly different that path-dependency (considered more of a lock-in to an existing technology). The link is that the ways of making the company more agile (investing in specialized manufacturing/prototyping and/or inventory) tend to create a more path-dependent trajectory (or opportunity set). So, perhaps this is better said that HW has 'interial' dependency on initial trajectory if not true path dependency per-se. Even provided that the startup is low-capital/outsourcing model. I hope this does some justice to incorporating your point.
Nice comment. I think there is indeed an inherent "inertial" in hardware business.
A software/website service can have different usage. Users may prefer to use a smaller part of the service, or using the main service for purposes not originally conceived by the creators. In this way there is a chance of pivoting.
However a piece of hardware has a very focused usage. All parts are integrated to provide a single function for the user.
Also, unlike software, there is no straightforward way to collect real-time data on how a user uses their hardware.
pg: if possible, I'd love to read your thoughts on how the economics of these businesses might evolve as they attempt to grow, especially as compared to software-only businesses.