Brain development starts with the formation of the neural tube followed by the emergence of radial glial (RG) cells that act as neural stem cells (NSCs) and sequentially produce neurons and then glial cells at later stages. Astrocytes are considered key players in normal brain function and during pathogenesis of various neurological and psychiatric diseases. However, the molecular mechanisms that control astrogliogenesis during development remain poorly understood. Hence, derivation of astrocytes from human pluripotent stem cells (hPSCs) has been inefficient, variable, and long-lasting. Current methods require extended periods of cell culture (6 months and longer), use of xenogenic material, and forced expression of transcription factors. Here, we developed a highly efficient chemically defined astrocyte differentiation strategy that requires neither animal serum nor genetic manipulation. Surprisingly, our in vitro approach bypasses neurogenesis, which is known to always precede astrogliogenesis similar to in vivo brain development. Specifically, by modulating gliogenic cell signaling pathways, we first differentiated hPSCs into FABP7+ RG-like cells in 7 days. These cells were then stepwise differentiated into astrocytes with over 95% efficiency in less than 30 days. Extensive molecular and functional characterizations (e.g. single cell analysis, calcium imaging, co-culture experiments), cell grafting into mouse brains, and disease modeling (e.g. Alexander disease) confirmed the specific identity and translational utility of hPSC-derived astrocytes. Lastly, our method was automated using a robotic cell culture system, which enabled the standardized production of billions of human astrocytes. Collectively, our results shed new light on cell fate specification by identifying remarkable plasticity in neural lineage progression that can be exploited to engineer hPSC-derived astrocytes for high-throughput drug screening and regenerative medicine.