Systems and methods are disclosed for delivering video content over a network, such as the Internet. Videos are identified and pre-processed by a web service and then separated into a plurality of segments. Based on user interest, video segments may be pre-fetched and stored by a client associated with a user. Upon receiving a selection from a user to play a video, the first video segment may begin playing instantaneously from a local cache. While the first video segment plays, subsequent video segments are transmitted from the web service to the client, so that the subsequent video segments will be ready for viewing at the client when playback of the first video segment has finished.