In the longer term, differentiated QoS can have much larger implications by affecting the balance of power between ISPs, their users, and content providers (including also service providers such as Google or application providers such as Skype). ISPs are under pressure to deliver ever faster connections to users and content providers, yet Internet access is becoming a commodity, with the price of subscriptions falling steadily. The trend can be reversed by turning the ISP network into a “ platform,” that is, offering specific QoS and performance levels to users and content providers alike, thereby making the ISP attractive to deal with (“the best video delivery,” “the best gaming experience”), as opposed to just one of many alternatives on the market.

In addition, by positioning itself as a distinctive “platform,” an ISP should be able to maintain, if not expand, its revenue stream; at a time when ISPs must carry out considerable investment in upgrading their networks (fixed and mobile alike), this could be a welcome evolution. Yet this would also imply a reshuffling of innovation pat-

terns. So far the Internet has been very successfully driven through innovation “at the edge,” outside of the networks (consider Google, Amazon, Skype, i Tunes, and all the Web 2.0 providers). In the future, innovation could equally be coming from the ISPs on their platforms. It is not clear for now whether this will substitute for or complement innovation at the edge, that is, whether innovation at the edge will be reduced (because innovative upstarts would be shut out) or further spurred.

What is more, technically no one knows yet how such differentiated QoS offerings could be implemented across the various networks that typically make up the fabled Internet cloud. This brings me to mention some significant differences between the EU and the U.S. In the U.S., the provision of broadband Internet access is concentrated in a few hands, namely those of the remaining local exchange carriers providing ADSL (AT&T, Verizon, Qwest) and the large cable TV providers. The official FCC policy is to bank on competition between the relatively few providers of these infrastructure

platforms (ADSL, cable, mobile). The Internet cloud would then give way to a limited number of single-firm platforms, each controlled by one of these providers, with two or more platforms being present at any given location in the U.S.

In the EU, the prospects for infrastructure competition are dimmer, since only a few areas (Benelux, parts of France, Germany, and the U.K.) are now served by competing broadband infrastructures (cable and ADSL). In most of the EU, it is thought that the rollout of competing broadband networks—effectively from scratch— cannot be achieved without some form of access to incumbent networks, at least in a starting phase. This means the ISP landscape in Europe looks different than in the U.S. and is likely to remain so in the foreseeable future: fewer competing infrastructures, but more market players, many of which rely on access to the incumbent’s network. Furthermore, that landscape is structured along national lines. In the end, it is difficult to conceive how differentiated QoS could be successfully