Frobenius, Geometrization, and the Archimedean Obstruction: The Langlands Correspondence Across Weil's Three Columns

This is an expository synthesis, not a report of new theorems. We survey the Langlandscorrespondence as it manifests in the three columns of AndréWeil’s celebrated analogy—numberfields, function fields of curves over finite fields, and function fields of complex curves (Riemannsurfaces)—and we advance a single organizing thesis: the availability of a global Frobenius endomorphismis the structural variable that determines which column admits unconditional proofs. Where a curve over afinite field supplies a Frobenius, the machinery of shtukas, Drinfeld’s lemma, and Vincent Lafforgue’sexcursion operators delivers the automorphic-to-Galois parametrization for every reductive group.Where the base is a complex curve, Frobenius is replaced by flat connections and the moduli stack ofbundles, and the categorical correspondence is now a theorem of Gaitsgory, Raskin and collaborators.Where the base is SpecZ, no global Frobenius exists and the correspondence remains conjectural.We develop three frontier topics through this lens: (i) the geometrization of local Langlands byFargues and Scholze, in which perfectoid tilting manufactures a Frobenius one place at a time onthe Fargues–Fontaine curve; (ii) the precise sense in which Lafforgue’s excursion operators are thetrace-of-Frobenius decategorification of the de Rham geometric correspondence, via Grothendieck’ssheaf–function dictionary; and (iii) the archimedean obstruction—the structural reason no shtukastyleargument reaches the place at infinity, and hence the number-field column. Throughout wekeep the geometric meaning of the word “field” (sections of a structure sheaf) in view, so that thealgebraic function field appears as the generic stalk of OX and the geometric field as a section ofan OX-module. A running GL1 thread exhibits each column’s abelian case as an unconditionaltheorem—class field theory, Fourier–Mukai, Lubin–Tate—isolating nonabelian reciprocity as the truelocus of difficulty, and an appendix proves the abelian case of geometric Satake in full.