In my limited perspective on the history of mathematics, I can name at least two big revolutions in Geometry: the introduction of Schemes in Algebraic Geometry, and the introduction of $\infty$-categories in Homotopy Theory. By "revolution" I mean a new formal framework to approach problems in a given branch of mathematics such that:
 1. Previous results can be reformulated in the new framework, at least partially;
 2. Its introduction brings substantial insight and advancement to that area.
 3. It usually requires much effort to be studied, but once obtained it provides an efficient and systematic toolkit to approach problems in the same branch.

Some of these revolutions divide the experts, whether temporarily or permanently; it could be the case that an alternative perspective makes some aspects of the theory clearer so that two frameworks can (usually with struggle) cohabit. I still count that as a 'revolution'. Let me also specify that (sub-)sub-branches are allowed and encouraged. For example, a similar shift can be seen in Knot Theory when Vassiliev began to study the cohomology of the (infinite-dimensional) space of embeddings $S^1 \to S^3$ instead of single, hand-crafted invariants on such space. It is not as radical as schemes though, since some aspects are not (yet) satisfactorily reformulated in this context.

Since revolutions are to some extent unexpected contributions, it is hard to predict them. However, I think they are not (only) the invention of a bright mathematician, but existing conditions in the mathematical 'ground' help them to stem. 

To limit the subjectivity of the answers, which will be inevitable to a certain degree, I have chosen two indicators that the ground is fertile for such a revolution to happen:
1. A sub-branch that has several interesting examples, tricks, and results, which lacks at the moment a clear pattern of strategies and a coherent formalism to explain them; 
2. A sub-branch in which formal aspects became a subject of research on its own, but no satisfactory results in terms of cohesion and efficacy have been found;

Regarding (1), I guess algebraic geometry felt a bit like that before Grothendieck. On the other hand, I see (2) happening often e.g. in category theory, and sometimes there is a struggle to find a nice reformulation. However, some axiomatizations become accepted in time, and several models are discovered to fit the same axiomatization. I find e.g. the development of cohomology theories very rewarding from this point of view, with the study of representing spectra becoming a new focus and boosting (or initiating?) the study of the stable homotopy category.

**Remark.** I did my best to make this question objective, and I would like to avoid advertisements of some areas of mathematics over some others. I understand if the question will be voted to close since big-list are quite controversial, but I think I would learn a lot from answers, so I'd give anyway a try. I also have a practical motivation: new theories usually require a lot of hours of studying to be mastered, and I'd be happier to avoid that... despite being curious about their content if they are satisfactory!

Regarding branches, I am mostly interested in algebraic topology and anything in this neighborhood, but I would be happy to listen from other geometers (e.g. differential). I think opening to other areas would make the question too broad.