I bet one of the leading factors is the number of sites blocking the AWS servers. As a scraping service, the fact that Amazon makes all their public IPs known (https://forums.aws.amazon.com/ann.jspa?annID=1701) is really inconvenient for anyone "crawling" the web. Rackspace also makes their list available, albeit incomplete (http://www.rackspace.com/knowledge_center/article/cloud-site...).

As more websites setup default blocks for all accesses from AWS services, the Moz exit was inevitable.