The value, 200, is a guideline, not a necessity. The link furnished by Safa Aouinti , is a good starting point; the answer depends on how many parameters there are in your model to be estimated, as well as how many latent variables. While more cases (bigger N) is always better as regards precision of estimates, it is quite possible to be wasteful of resources by collecting too many cases for a given analysis.
Here is a simple sample size calculator for a family of SEM models that you may find helpful: https://www.danielsoper.com/statcalc/calculator.aspx?id=89