[mikeyouse]

There is a ton of value waiting to be created in Michigan, especially centered around the auto industry. The margins aren't anywhere near as high as they are in software companies, but would you believe that "List 1" and "List 2" below have about the same combined annual revenues (~$375B).

1. GM, Ford, Chrysler

2. Apple, Google, Microsoft, Oracle, Salesforce, Yahoo!, Facebook, Adobe, Netflix, Tesla, Twitter, Pandora

There is a lot of value in the supply chain that could be captured by clever software or manufacturing techniques.

[lifeisstillgood]

Really?

My (UK based, vaguely remembered) understanding is at least two of those companies in list 1 went bust in the last five years, and that over the past twenty or more years an awful lot of effort has been spent squeezing inefficiencies out of the supply chain (the auto industry is the poster boy for JIT manufacturing, for pushing debt fueled capital expenditure into your own supply chain etc)

Manufacturing scares me as the incumbent software providers (SAP etc) are huge beasts, well embedded and actually good at writing software.

My preference would be to find industries where the main software provider was actually Excel, or just pen and paper.

[BjoernKW]

If you're talking about actual supply chain management and manufacturing you're probably right. There are entrenched providers in those areas you can hardly take on as a startup.

However, speaking of SAP and the likes: In any large company there are so many support functions that literally run on Excel or some cobbled together SAP atrocity the opportunities are almost limitless.

[mikeyouse]

The good thing about trying to size a market with only a few major players is that all of the expenses are obvious. Auto makers typically have COGS nearing 80% of their revenue, so for that $375B in annual sales, you'd see $300B in annual COGS. Some of that is in-house but much of it is spent on suppliers and the rest of the value chain.

It's obviously easier to make a MVP of a chat app than to design a ML painting algorithm to save 1% in paint every year (or whatever the specifics may be), but there are definitely opportunities there if you're looking for a 'big market'.

[lifeisstillgood]

I think that's confirming my intuition. Getting that innovative paint algorithm to the point it can simply plug and play with current robo-paint-arm in 8/10 factories will involve:

- developing the algo (ie knowing the domain intimately) - verifying the algo against the four major robo-paint-arms in production (probably more) - meeting the safety testing criteria (expensive) - demonstrating the savings in paint vs cost of other expenses (ie we save 1% paint but the stop start nature clogs up 10% more often which is a manual fix) - sales and marketing

wow, I just see problems don't I ?

But compare that to "If you tie this box to every cow, you will be able to tell which ones are active, which lethargic, which have eaten grass from the top of the field and can then sell their milk for more"