If I was to summarize (simplify?!) the trend that is happening in the SaaS space, I would say that the business model is moving from a high margin - low volume model to a high volume - low margin model, as described in the picture below.
This is of course not new and is coherent with the model we described in our first paper: Architecture Strategies for Catching the Long Tail. But the more we work on SaaS architectures and the more we realize the extend to which this shift in business model impacts architecture. Multi-tenancy can be explained by this shift, the emergence of service delivery platforms (or SaaS hosting platforms depending who you ask) can be explained by this shift, aggregators and marketplaces can be explained by this shift. In other words, this shift is at the very core of many (if not all) aspects of software as a service.
OK, no rocket science here, just another acknowledgment of the importance of operational efficiency as well as the need of achieving high density of customers (multi-tenancy, virtualization, SDPs...) in order to be successful in the high volume low margin era. I suppose, my recommendation to you would be to ask yourself how "high volume - low margin" (HVLM) - friendly you are with your solution and architecture. Of course, the more HVLM you are, the better prepared you will be.
In a slightly tangent topic, but still related, I was talking with Matt Baldwin today (he works with Affinity Hostway) about shared hosting vs collocation and he helped me realize that in mega data centers, power (electricity) and space ("U") are often the limiting factors. This basically means that the number of customers per watt and/or the number of customers per U becomes an key metric. Imagine you can host X number of customers in a shared environment using for example 1kilowatt and 6Us; the opportunity cost of selling the same 1KW, 6Us in collocation could be very high, as now you are only serving 1 customer which is unlikely to pay X (the number of shared customers you could have had) times the price of shared environment.
This reminded me of real estate where the value is often not the house itself but the land it is built on. If a single family house is build where a skyscraper could be built instead, the opportunity cost of selling a house is huge (or the house will have to be very expensive)