I'm going to be controversial for a minute. Will cloud actually make IT cheaper? Public cloud service providers often state that their costs are lower, using criteria such as Power Usage Effectiveness (PUE) to show how they can run services more cost-effectively than the in-house equivalents.
Indeed, such providers have the benefit of scale - and can therefore negotiate cheaper deals with equipment suppliers; its also in their commercial interests to be architecting and running IT as efficiently as possible. Without a doubt, it is possible to take specific examples of workloads and demonstrate how running them in the cloud costs less than running them in-house.
Cost is a movable feast, however. If equipment already exists, under-utilised, in the data centre, it might be cheaper to use it rather than rent processor cycles from a third party. There's also the operational overhead - but in a bureaucratic organisation, this might be just as high wherever resources are. Equally, getting data in and out of the cloud from other systems might involve considerable manual intervention, which can be hard to quantify in financial terms.
The potential for people to take matters into their own hands and use cloud resources anyway suggests another cost - associated with the risk of things going wrong. Cloud provider due diligence is a cost in itself, which may be seen as necessary by the organisation but not by a department eating to get something done. However, if there is a data leak or governance breach, the result could be expensive.
To top it all, we need to take Jevons' Paradox into account. Having researched coal use in the 19th Century, William Jevons proposed that technologies that increased the efficiency of using a resource, only caused the resource to be used more than it was previously. In other words, there will always be more we can do with IT, and if we make it cheaper by using a cloud provider, we'll simply increase the number and scale of workloads.
Perhaps the lesson is to avoid simplistic statements like "cloud is cheaper" and to keep in mind the broader costs involved in IT. Coming back to the see-saw analogy I talk about in this video , it's important to recognise that efficiency gains on one side need to be balanced with operational overheads and risks on the other. Only by doing so can we get a true picture of the cost of cloud.