If you use a branch or feeder from one building to supply power to another building, you generally can run only one branch or feeder for that purpose [225.30]. There are five conditions under which you can exceed this limit [225.30(A) - (E)]. The third one is "capacity requirements."

This is pretty straightforward. If the building meets these requirements, you can exceed the single feeder/branch circuit limit for supplying it:

  1. Capacity requirements exceed 2000A.
  2. Supply voltage is 600V or less.

What isn't so straightforward is answering the question of why you're using this method of supply, rather than a service, in the first place.

There may be a good reason, such as very limited space. Suppose that’s your reason. If it's early enough in the project to change the site layout or structure design to allow for enough space to provide a service to that structure, doing so is nearly always advisable. Discuss this with the architect.

The fact the NEC allows this method does not make this method an "equal alternative" to a service. While it might possibly save money in construction, it comes at a cost in several other ways over the life of the installation. Wherever practical, the engineering decision will nearly always result in a service instead.