Full EPC refresh/replacement. Perhaps the least likely option in the near term because a) it's difficult with today's state-of-the-art technology; and b) many operators have relatively new EPCs working just fine. Nevertheless, NTT Docomo has a target of 2016 to introduce vEPC into its main 4G backbone network (Docomo has particular reasons for this aggressive schedule) and, working over a three-year view, preparation for an EPC refresh is already important for large, progressive operators.

Capacity expansion. As operators upgrade capacity, it may make sense to do this with additional virtual instances deployed "in pool" with the physical EPC functions, rather than adding new chassis. One implication is that, by deploying in pool, the operator has to go with the same vendor for classic and virtual EPC.

Overlays for the Internet of Things. IoT services will generate a different traffic profile to the consumer smartphone users (e.g., more transaction-oriented) and operators can create vEPC instances optimized for this type of service. SK Telecom in South Korea is pursuing this model, and AT&T expects parts of its connected car service to run on this type of platform.

Mobile virtual network operators. Sophisticated MVNOs tend to want their own core networks, so that they can control their services independently of the host operator. This can be time consuming to set up and inefficient. Using NFV operators can offer MVNO wholesale customers with their own virtual "core-as-a-service." Elephant Talk, which provides mobile virtual network enabler services for Vodafone, is an example of this model.

Greenfield operations. The first vEPC to be deployed in commercial service was announced back in 2013, on behalf of a new operator in Myanmar, so clearly this model is workable. There aren't that many "classic" greenfield opportunities any more, but there are some, as the current speculation around Dish Networks shows, and you could also put MVNOs and IoT providers building their own cores into this category.

Virtual private enterprise networks. This is an intriguing opportunity that isn't yet well-developed, but has the potential to be important in the high-end enterprise market, and therefore to operator revenue and profitability. The idea is to offer an enterprise a hosted virtual private network configured to its specific needs (and configurable by the customers). Segments with proprietary processes and unique research, such as aerospace, finance, pharmaceutical, etc., are the best opportunities for this type of service. You could also think of it as an extension of the wireline Cloud VPN services now in the market.