We consider the problem of constructing a communication infrastructure from scratch, for a collection of identical wireless nodes. Combinatorially, this means a) finding a set of links that form a strongly connected spanning graph on a set of $n$ points in the plane, and b) scheduling it efficiently in the SINR model of interference. The nodes must converge on a solution in a distributed manner, having no means of communication beyond the sole wireless channel. We give distributed connectivity algorithms that run in time $O(poly(\log Δ, \log n))$, where $Δ$ is the ratio between the longest and shortest distances among nodes. Given that algorithm without prior knowledge of the instance are essentially limited to using uniform power, this is close to best possible. Our primary aim, however, is to find efficient structures, measured in the number of slots used in the final schedule of the links. Our main result is algorithms that match the efficiency of centralized solutions. Specifically, the networks can be scheduled in $O(\log n)$ slots using (arbitrary) power control, and in $O(\log n (\log\log Δ+ \log n))$ slots using a simple oblivious power scheme. Additionally, the networks have the desirable properties that the latency of a converge-cast and of any node-to-node communication is optimal $O(\log n)$ time.