Disclosed are various embodiments for prefetching of objects referenced on a network page. An encoded network page referring to at least one item is retrieved. The same item is included on a second network page. In response to an indication of user interest in the item on the first network page, at least an initial portion of a video resource associated with the indicated item and included on the second network page is retrieved. In response to a user selection of the same item, the retrieved initial portion of the video resource is rendered on the second network page.