There is growing recognition that the nation requires a Learning Health System (LHS) to provide higher quality, safer, and more affordable health care--and to empower biomedical research and public health. The LHS is an ultra large-scale information infrastructure, effectively a smart grid for health. A fully-functional LHS can routinely aggregate data from disparate sources, convert the data to knowledge, and disseminate that knowledge, in actionable forms, to everyone who can benefit from it. In this way, the LHS enables a multitude of virtuous cycles of study, change, and improvement over time. The LHS is a single infrastructure that can support health care quality improvement, disease surveillance and other activities of public health, and all forms of biomedical research. Achieving a high-functioning Learning Health System at national scale has been likened to the "moon shot" of the 1960s, but this greatly underestimates the magnitude of what must be conceived as a ultra large scale socio-technical challenge. Indeed, the challenge of the LHS may be akin to that of building the Panama Canal, more than a century ago, which required, among several feats of engineering, the conquest of an infectious disease and creation of a new nation. Solving the problems prerequisite to achieving an LHS may require a significantly new science of learning systems. This presentation describes the imperative of the Learning Health System and how it is currently envisioned, the features it requires to be effective and sustainable, and the key questions a science of learning systems must address.