Telegram was not disrupted during the AWS crash, so they probably were not using it (or had a decent fail-over mechanism to a backup system). Telegram's user-base is two orders of magnitude larger than Signal, so 'we use AWS because we have to' argument clearly is bogus and nonsense.