Data rates of mobile communications have increased dramatically during the last decade. The industry predicts an exponential increase of data traffic that would correspond to a 1000-fold increase in traffic between 2010 and 2020. These figures are very similar to ones reported during the last Internet boom. In this article we assess the realism of these assumptions. We conjecture that wireless and mobile Internet access will emerge as a dominant technology. A necessary prerequisite for this development is that wireless access is abundant and becomes (almost) free. A consequence is that the projected capacity increase must be provided at the same cost and energy consumption as today. We explore technical and architectural solutions that have realistic possibility to achieve these targets. We ask if Moore's law, which has successfully predicted the tremendous advances in computing and signal processing, will also save the day for high-speed wireless access. We argue that further improvements of the PHY layer are possible, but it is unlikely that this alone provides a viable path. The exponential traffic increase has to be matched mainly by increasing the density of the access networks as well as providing a modest amount of extra spectrum. Thus, the future research challenges are in designing energy- and cost-efficient short-range architectures and systems that support super-dense deployments. A non-technical complication is that such infrastructures are likely to lead to highly fragmented markets with a large number of operators and infrastructure owners.