Abstract: As the source of the digital economy, digital innovation has become an important strategic tool to generate new drivers and promote high-quality economic development. How to optimize the enterprise digital innovation mechanism and improve the level of enterprise digital innovation is an important problem. However, existing researches mostly focus on the influence of a single factor on digital innovation, which is lacking of an integrated framework to systematically analyze the mechanism of digital innovation, and also neglect to discuss the synergistic effect between multiple factors. To further explore the influence mechanism of digital innovation and promote the practice of digital innovation, the "2020 top 100 innovative enterprises of digital economy" published by Internet Weekly as the initial basis for case sample screening, this study finally obtained 20 internet enterprises listed in Shanghai and Shenzhen A-shares. With the help of TOE theoretical framework, the study used fuzzy set qualitative comparative analysis and necessary condition analysis method to explore the joint effect of five antecedent conditions at three levels of technology-organization-environment on the digital innovation of internet enterprise. The conclusions are as follows: (1) There are no necessary conditions for high-level digital innovation. That is, a single condition does not constitute the bottleneck of high- or low-level digital innovation. (2) The driving mechanism of high-level digital innovation is divided into four recipes: namely, technology and environment synergy with big data capability empowered (H1), technology and environment synergy with innovative capability supported (H2a), technology and environment synergy with big data capability and innovative capability joint supported (H2b), absorptive capacity and digital context dual-wheel driven (H3). Each configuration is composed of multiple antecedent conditions, i.e., "multiple concurrency". Different combinations of antecedent conditions lead to different configurations, which can achieve high-level digital innovation, i.e., "all roads lead to the same destination". (3) There are four configurations that result in not-high level digital innovation. Among them, configuration NH1a has high innovative capability, but lacks digital infrastructure, big data capability and absorptive capacity. NH1b has high innovative capability and digital context, but lack of digital infrastructure and absorptive capacity. NH2a has a high absorptive capacity, but the other four conditions are deficient. NH2b has high digital infrastructure, big data capability and absorptive capacity, while the other two conditions are lacking. There is causal asymmetry between the driving mechanism of the four not-high configuration and high-level digital innovation. The following management implications can be drawn. Firstly, to achieve a high-level digital innovation, internet enterprises should pay attention to the coordination and improvement of various capabilities, and the thinking should shift from "local optimization" to "configuration coordination". For example, when internet enterprises have high level digital infrastructure, they should combine the digital context and enhance the innovative capability in a targeted way. Secondly, managers need to choose the digital innovation path according to their own conditions, which is suitable for the enterprise's resource endowment and external environment. Thirdly, internet enterprises should closely combine with the actual situation of China's digital economy and make good use of the basic resources of external environment.

Key words: digital innovation, internet enterprise, TOE framework, fuzzy set qualitative comparative analysis, necessary condition analysis