**Speaker**: Wei Dai (1kx) **Abstract**: In this talk, we compare the current and historical cost of adding verifiability to different types of computation. Using these data points and the recent theoretical results on linear time provers, we aim to better understand the following questions: is it possible for zk compute to be as cheap or as fast as native compute? If not, how should we expect the zk compute overhead to change over the next few years and decades? **Slides:** TBD **Video**: TBD