Today was a day of lengthy conversations on many topics. One that I’m still wrestling with is the question: “what is the role of vertical integration in the private space industry?”

Currently, space tech companies seem to employ vertical integration significantly more than other industries. This phenomenon is especially well illustrated by SpaceX – which does everything from bend its own sheet metal for the rocket exterior to fabricate its own PCBs.

Two factors stand out as reasons why vertical integration is especially attractive in the space industry:

Strong risk aversion – the huge expense of any single rocket and payload demands strong assurance that the launch will succeed. This in turn pushes exacting standards on components. Because launches are still relatively infrequent, space companies can’t afford to use repeat interaction as a test of whether an external product is worthwhile: they can’t go online and get an aggregate of hundreds or thousands of reviews.

Lack of standardization – most space hardware is still customized for each specific payload. This case-by-case customization makes it far more difficult for an external company to provide a product that can interface with the products of multiple customer companies.

Given these considerations, Coase (and, of course, reality) would suggest that vertical integration makes sense. However, I think that the space tech industry would be far more efficient if it made sense for most space companies to specialize.

The trick then, is to create innovations that change the necessity for such strong risk aversion and create more standardization. A lot of the work with cubesats is moving things in that direction, but its interesting to think about what other steps one could take to improve efficiency without the resources of Elon Musk.

Speaking of Elon Musk and vertical things, the Grasshopper continues to make me giddy.