The mammalian neocortex is characterised by precise patterns of synaptic connections, which dictate how information flows through cortical circuits. The degree of convergence and divergence of excitatory information results from highly non-random, short-range and long-range glutamatergic synaptic connections (1-11). Previous seminal work has shown that connections across cortical layers can form preferentially between sister neurons born from the same individual progenitor cell (12). However, cortical neurons also form synaptic connections with many neurons within the same layer (1-6) and this intralaminar connectivity has been shown to reflect specific patterns of longer-range connectivity (1-3, 9-10). Given that neurons within the same cortical layer can be derived from a heterogeneous population of embryonic progenitor cells (13-18), we examined how defined progenitor pools influence local excitatory synaptic connectivity in mouse primary somatosensory cortex. We find that upper layer neurons derived from a subpopulation of progenitors, called short neural precursors, connect out-of-class, such that they preferentially form intralaminar connections with neurons derived from other types of progenitor. Similar local connectivity rules have been associated with differential long-range connectivity (1). Indeed, while we find no differences in basic intrinsic or morphological properties, optogenetic circuit mapping reveals that progenitor pool predicts the long-range thalamic input a neuron receives, as well as a neuron's output to deeper layers of cortex. Neurons derived from short neural precursors receive more input from a higher-order thalamic nucleus and provide stronger output to layer 5a, which is associated with cortico-cortical projections. In contrast, adjacent neurons derived from other progenitors receive more input from a first-order thalamic nucleus and provide stronger output to layer 5b, which is associated with subcortical projections. These data suggest that the emergence of progenitor pools has enabled the cortex to generate distinct local and long-range subnetworks for the differential routing of excitatory information.