The spread of many-core processors opens the possibility of new economic models for providing processor capability and tailoring it to customers' needs. Today customers do not have that option. They must purchase hardware that is capable of meeting either their maximum processing requirements or their typical requirements. In the former case they are paying for processing power they rarely use; in the latter, their systems perform inadequately when peak demand is placed on them. Joseph Sloan and author Rakesh Kumar have proposed models for on-demand pricing of multicore processing power. These models range from the capability of buying on-the-spot hardware upgrades only, to flexibly upgrading or downgrading within a range as requirements change, to 'Cores on Rent' and 'pay per use' models that separate ownership from use, the latter of these allowing billing of users only for the processing power they actually use during a payment period. There is nothing new about this general concept. Large servers and mainframes already use a Capacity-On-Demand (COD) model to assign costs to customers. However, these subscribers tend to be few in number and well established, so that ensuring compliance is relatively simple. Applying the principle to a vast number of largely anonymous customers poses new security and privacy challenges that will have to be worked out. The change to flexibility in processing capability will also require innovative changes in software licensing models.