The ESA mission Gaia is getting epoch astrometry of asteroids at better than 1 mas (milli-arcsec) accuracy. Initially, the measurements for 14.099 asteroids over 22 months have been made available, in the Data Release 2 (April 25, 2018). About 10 times more small bodies will be present in DR3, doe to appear in about one year. We will review how these data can be used to infer important properties on asteroid masses, on the presence of previously undetected satellites, and to determine Yarkovsky acceleration for some of them. While all these aspect are important, measuring Yarkovsky stands out for its implications in the evolution of the asteroid belt. However, we can state today that Gaia is not doing its job alone: collecting observations from the stellar occultations of asteroids is also a way to provide very accurate astrometry. We will discuss the sources of error that intervene in the process, the expected performances, and what can be inferred from the existing data set of occultation astrometry, mainly concerning Main Belt objects. We will then discuss the new challenges and perspectives for occultations by Near Earth Asteroids, and the related difficulties in prediction, observation and exploitation. The recent success on Phaethon and Apophis are probably at the edge of the performance that can be expected for predictions, in future years.