A serious question - please post comments if you have thoughts!

Does the ratio of service providers to OEMs in an industry correlate indirectly with the defect rates in its critical components?

In other words, as the manufacturing processes required to produce a product become more reliable, is production shifted away from OEMs?

You won't be surprised that my question relates to metal AM - and the degree to which OEMs can generally outspend (in both R&D and acquisitions) the smaller job shops. When a critical process in the industry is unreliable, OEMs can invest the capital expenses to either solve the problem (through R&D and often resulting in trade secrets) or acquire companies who have. But as the process matures, smaller service providers can be more competitive, as their overhead is (citation needed) in many cases lower.

As a result: Until the process (for example, metal powder bed fusion) is fully industrialized and reliable, it's very difficult for small shops to enter the market. But once the technology is well understood, mom & pop shops are able to flourish.

A concrete example: Today, OEMs like GE, Airbus, and Philips dominate the metal additive industry, and the proprietary R&D they do makes insourcing components more cost competitive than buying them from service providers. If you start a job shop today, it might be 12-18 months before it reliably creates revenue. But if & when additive becomes a more predictable process, the time to revenue (and profit) will be shortened, and OEMs will find it increasingly attractive to outsource their parts.

^ This is a half baked theory - I'd love to hear your perspective!

Edit:

This question prompted a good discussion on twitter!
Another note to the example above: Arcam EBM is (and I don't mean this as a criticism) less fully industrialized than laser metal powder bed fusion.
There are (by my count) 23 firms in the US who own Arcam machines; three are job shops, one is owned by Arcam, and the rest are OEMs or research institutes. On the other hand, there are many dozens (by my count at least 70) of service providers who have laser based machines. If the ratio of OEMs to job shops were consistent across the technologies, you'd expect there to be over 500 firms in the US running laser machines in house - which sounds *much* too high to me.