Abstract:    Integrated circuit (IC) is the most important component of electronic devices, which plays a decisive role in national security of information. Unfortunately, the majority of Chinese IC enterprises have to be embedded in a low position of the global value chain (GVC) due to the low development level of China IC technology. It is urgent for China to develop core technology system of the IC industry. Since international new ventures (INVs) target global market and utilize global resources to create value at their inception, exploring the embedding path of GVC of INVs in China′s IC industry will not only enhance the core competitiveness of China IC industry and China enterprises, but also develop organizational learning theory and localize international entrepreneurship theory.
    There are six key elements in INVs′ learning. The first one is feedforward learning which includes the design of strategic vision, the identification of opportunity in the initial stage, and the leadership in the middle stage of entrepreneurship. The second element is accidents-driven learning. When succeed accidentally, INVs will summarize the trend of market in order to transform the occasional success into new profitable way. The third is information search and collection. Trying to collect more external information will benefit INVs to accumulate and absorb knowledge promptly so that they can identify new market opportunity. The fourth one is experiment which includes strategy adjustment, new product and services development and so on. The fifth element is knowledge storage and retrieval. INVs can store and retrieve the learned knowledge to enhance their cognitive capability. The last is continuous improvement which includes technology improvement after experiment failure, knowledge update after application, and upgrade of organizational process. 
    Based on organizational learning theory, this paper argues that the learning model of INVs can be classified into information exploration learning, technology exploration learning and exploitation learning. Information exploration learning refers to the process of collecting external information, which includes feedforward learning, accidents-driven learning, and information search and collection. INVs using information exploration learning can achieve competitive advantage resulting from information asymmetry. Technology exploration learning includes accidents-driven learning and experiment. It refers to the process of finding new technology, new process and new product by experiment so that INVs can accomplish learning-by-doing. Although facing high risk and uncertainty, INVs using technology exploration learning can enhance innovation performance dramatically. Exploitation learning includes knowledge storage and retrieval and continuous improvement aiming to exploit current organizational knowledge fully and upgrade learning curve. Although facing lower risk and prompter performance feedback, exploitation learning isn′t useful for sustainable development.
    Organizational learning is indispensable throughout INVs′ whole process of GVC embedding. It benefits INVs to embed GVC by means of increasing productivity, enhancing market competitiveness and avoiding entrepreneurial risk. According to extant literature, GVC embedding is useful to increase productivity. Meanwhile, GVC embedding helps INVs to enter into international market by accessing external knowledge. Organizational learning fosters INVs to consider their management rules and norms seriously, which is useful to enhance market competitiveness. INVs with strong learning capability can improve the storage of knowledge and retrieve and create knowledge efficiently. Consequently, INVs reduce market entry cost and avoid entrepreneurial risk.
     Using cross-case analysis, this paper explores GVC embedding paths of three companies including Semiconductor Manufacturing International Corporation (SMIC), Spreadtrum Communications (SPRD) and TongFu Microelectronics (TFME). 
    As the biggest IC manufacturing enterprise in China, SMIC focuses on producing IC solely or jointly with IC designing companies. From its inception, SMIC has been adopting information exploration learning by cooperation with foreign IC corporations. The learning elements of SMIC include feedforward learning, knowledge storage and retrieval, and continuous improvement. By continuous cooperation with international famous IC corporations, SMIC has achieved great performance and upgraded its position of GVC, which can be called horizontal embedding path of GVC. 
    In terms of SPRD, it has been dedicating to innovate independently since its establishment. The learning elements of SPRD include feedforward learning, accidents-driven learning, knowledge storage and retrieval, and experiment, which can be regarded as technology exploration learning. During the process of GVC embedding, SPRD has been concentrating on research and development (R&D). By establishing research affiliates in developed countries like U.S., SPRD learns and imitates technology from host countries′ enterprises closely, which is useful to broaden information channel. Eventually, GVC embedding strengthens SPRD′s R&D capability, which can be called upstream embedding path of GVC.
    As a joint venture, TFME mainly engaged in IC packaging test. In the learning process, it used many learning elements such as information search and collection, knowledge storage and retrieval, and continuous improvement, which can be regarded as exploitation learning. During the process of GVC embedding, TFME has been exporting intermediate goods to foreign top corporations by seeking and widening product line of packaging test, which forces TFME to learn advanced technology and management experience and upgrade its learning curve. As a result, GVC embedding consolidates TFME′s market share, which can be called vertical embedding path of GVC.
    In sum, horizontal embedding path of GVC is used for INVs with information exploration learning by cooperation with international corporations. Meanwhile, upstream embedding path of GVC is used for INVs with technology exploration learning by establishing R&D center in both domestic and foreign countries. Moreover, vertical embedding path of GVC is used for INVs with exploitation learning by strengthening its original position of GVC. INVs should choose appropriate embedding path of GVC to internationalize based on the features of their learning elements. Specifically, those using horizontal embedding path of GVC should focus on strategic planning, take advantage of accidents, and collect information. Furthermore, those using upstream embedding path of GVC should emphasize R&D by continuous experiment. Meanwhile, those using vertical embedding path of GVC should concentrate on extant knowledge application and improvement. Once embedded, INVs should improve and upgrade GVC, which requires INVs to construct comprehensive knowledge system and adopt organizational learning in the whole process.

Key words: international new venture, China′s chip industry, organizational learning, embedding path of GVC