I worked at a semiconductor mixed-signal startup - rife with control-loop theory, analog design, digital design/firmware, and lots of application support/softwa

I worked at a semiconductor mixed-signal startup - rife with control-loop theory, analog design, digital design/firmware, and lots of application support/software. They do exist. It's just very hard for them to last without eventually getting acquired by an established bigCo. But the same kind of rings true for a decent chunk of startups in the software space. That said, most semiconductor startups do partnerships with bigger companies for second-sourcing or foundry help. The transaction cost problem is true whether you're a startup or a big company. If anything it's better to be a highly-partnered startup than a startup-division at a big company as you may be able to avoid lots of the bureaucratic headaches the large company is accustomed to, helping to keep employees more motivated.