Network Capacity, which is limited by interference between simultaneous transmissions of neighboring links, is a fundamental performance metric for wireless ad hoc networks. Ref. [1] showed that the capacity of wireless ad hoc networks with omnidirectional antennae does not scale well as the node density increases. Other related work ([2], [3], [4], etc.) demonstrated that the network capacity scales better with the use of directional antennae. Most previous investigations, however, were based on simplified but unrealizable models of antenna patterns. So far, there has not been a general framework for the analysis of network capacity when directional antennae of realizable generic patterns are used. Also, few papers have derived explicitly the impact of nulls and null-width in antenna pattern on network scalability. Specifically, the scale law of network capacity with general directional antennae has not yet been investigated. This paper addresses these issues. We study arbitrary and random networks (corresponding to the best case and random case respectively), where nodes with smart antennae are assumed to be located in a planar disk without mobility. We also point out how to generalize the result to that of generic directional antenna pattern. To conserve space here, we defer detailed analysis of the impact of null-width to the full paper.