I was part of a hardware startup for nearly a decade. Most real hardware companies that I recall failed for reasons other than hardware. Which fits with the a

I was part of a hardware startup for nearly a decade. Most real hardware companies that I recall failed for reasons other than hardware. Which fits with the author's overarching argument. When people imagine a hardware company failing, I think most imagine something going terribly wrong with prototypes, some failure in the field, manufacturing failures, overbuying, etc, etc. But I think those events are rarely fatal for real hardware companies. I say "real hardware companies" because those kinds of failures are endemic with crowd funded hardware startups. And I think that's why most people assume that all hardware companies fail for those reasons, because crowd funded hardware startups get so much public attention. But outside of the crowd funding bubble, my experience has been that those kinds of events are just normal bumps in the road. We were on razor thin angel funding, and we hurdled all those kinds of problems. What really kills hardware startups is what kills _any_ startup. Bad product-market fit. Over the decade of my involvement with my previous hardware startup, we watched _tons_ of fellow hardware startups die. Not because they hit some manufacturing snag, or shipped a million units of explosive doorstops. Every single one died because no one wanted their product. The classic startup tale. Perhaps I have a unique perspective on the whole software startup versus hardware startup thing because I'm a jack of all trades. I'm just as comfortable working software as I am hardware (and boy does that duality come in handy). So I've seen both sides from the trenches. Hardware people will tell you it's harder, because they don't know how frustrating real software development is and assume it's some kind of greener pasture. And software people will tell you it's harder, because hardware is a mystery to them. I've wasted days dealing with Google Cloud's SDK issues. And I've wasted just as many days chasing a defect in a board remotely with a customer half way around the world. In other words: The grass is not greener on the software side.