The Iranic language family spans 101 ISO 639-3 coded varieties and over 200 million speakers across a vast arc from the Caucasus through the Middle East and Central Asia to South Asia and China. Yet this richness is barely reflected in modern NLP: resources, models, and benchmarks are overwhelmingly concentrated on Iranian Persian, leaving Kurdish, Pashto, Balochi, Gilaki, Zazaki, and the Pamir languages computationally invisible. This survey provides the first comprehensive landscape analysis of Iranic NLP, covering 495 papers published between 2020 and early 2026. We examine language coverage, NLP tasks, datasets, modeling approaches, and evaluation practices, and identify persistent structural gaps: script heterogeneity, dialectal fragmentation, data scarcity, and an evaluation monoculture that has saturated Persian while leaving near-zero baselines for every other variety. Together with this survey, we release a companion website providing interactive language maps, a curated resource browser, and a live publication index at https://sigiranic.github.io.