Since late 2021, a panzootic of highly pathogenic H5N1 avian influenza virus has driven significant morbidity and mortality in wild birds, domestic poultry, and mammals. In North America, infections in novel avian and mammalian species suggest the potential for changing ecology and establishment of new animal reservoirs. Outbreaks among domestic birds have persisted despite aggressive culling, necessitating a re-examination of how these outbreaks were sparked and maintained. To recover how these viruses were introduced and disseminated in North America, we analyzed 1,818 Hemagglutinin (HA) gene sequences sampled from North American wild birds, domestic birds and mammals from November 2021-September 2023 using Bayesian phylodynamic approaches. Using HA, we infer that the North American panzootic was driven by ~8 independent introductions into North America via the Atlantic and Pacific Flyways, followed by rapid dissemination westward via wild, migratory birds. Transmission was primarily driven by Anseriformes, shorebirds, and Galliformes, while species such as songbirds, raptors, and owls mostly acted as dead-end hosts. Unlike the epizootic of 2015, outbreaks in domestic birds were driven by ~46-113 independent introductions from wild birds, with some onward transmission. Backyard birds were infected ~10 days earlier on average than birds in commercial poultry production settings, suggesting that they could act as early warning signals for transmission upticks in a given area. Our findings support wild birds as an emerging reservoir for HPAI transmission in North America and suggest continuous surveillance of wild Anseriformes and shorebirds as crucial for outbreak inference. Future prevention of agricultural outbreaks may require investment in strategies that reduce transmission at the wild bird/agriculture interface, and investigation of backyard birds as putative early warning signs.