According to a widely shared view, manufacturing is currently un- dergoing its fourth industrial revolution, termed “Industrie 4.0” in the high-tech strategy of the German government. Smart Factories with vertically and hori- zontally integrated production systems are enabled through the realization of machines, storage systems and utilities as Cyber-Physical Systems (CPS), which are able to share information, act, and control each other autonomously. The development of CPS requires the collaboration of different disciplines, like mechanical engineering, electrical engineering and computer science. This cre- ates new challenges for Requirements Engineering (RE), which needs to estab- lish a common perception of the targeted CPS for the involved stakeholders. This paper will elaborate the specific challenges in RE for CPS based on a lit- erature review. Natural Language Processing (NLP) is used as an approach to automatically translate shared informal requirements specifications to formal domain specific models for the involved disciplines, to develop a comprehen- sive RE methodology for CPS.