For some startups partnerships are essential to execution, for instance by giving access to an audience, a technology, or a process.  In those situations I have a rule that is based on a fair bit of experience: try to avoid partnering with other startups (David) or the biggest players in the respective market (Goliath).

Partnering with other startups is problematic because they may change direction and wind up competing with you, or dropping whatever it is that you are relying on, or possibly go out of business altogether.  Startups partnering with each other for something essential is too often a case of the blind leading the blind.

Partnering with the biggest of the big on the other hand tends to be problematic because they are unlikely to be sufficiently responsive.  I am specifically talking about a partnership here, which requires the Goliath to do something out of the ordinary course of business.  In other words, using Amazon S3 is not an example of partnering.  Working with Amazon on building something that requires them to modify how S3 works would be an example of partnering.  Another problem in partnering with a Goliath is that most likely the person responsible for the partnership is several layers down from top management and your whole partnership may go out the window when that person changes jobs.

So whenever possible, as a startup you should rely on mid size well established companies when it comes to strategic partnerships.  Once that have enough of a business that they won’t go away or totally change strategy, but for whom the partnership is meaningful enough and visible at the executive level to actually make it work.