For example, in 2021 Twitter didn't report monthly users, but "monetisable daily active users", a metric that Musk did not understand.
I think that 650 million of X comes from a report of the SEC, but it also includes just users of Grok, not just the social media section.
So it's a different type of user, over a different time period, for a different service.
Looking at this from an investment standpoint (as if, e.g., I were establishing funding requirements for a pension fund or an annuity or some other such financial vehicle), MAUs are a nice to have because they can establish a counterfactual[0] for volume, but I'd be looking at the performance of X's products relative to the financial performance of X's customers buying those products (e.g., for advertisers, things like conversion and customer acquisition cost come to mind), legal exposure (because of, e.g., CSAM distribution) and other tail risks, and also whether there's an existential risk to the liberal democratic order that enables the financial vehicle I'm working with to operate at all, among other portfolio-level risks.
--
0: Well, kinda. The counterfactual actually comes in when we ask questions like, "If the reach changes, what happens to the economic performance?" Relying on MAUs as a baseline measure of the reachable population based on observation enables us to establish these kinds of counterfactuals.