The paradigm of enterprise technological innovation is undergoing a significant transition from a traditional "value chain" framework to a more complex, multi-stakeholder "value network". This transition is driven by the rise of global trade protectionism, accelerated by the urgency for domestic technological self-sufficiency, particularly in critical sectors like the semiconductor industry, where "chokepoint" technologies pose severe national security and economic stability risks. The semiconductor industry’s innovation ecosystem is shifting toward dense regional networks, demanding collaborative and non-linear approaches to value creation, contrasting sharply with the sequential and linear nature of the traditional value chain model. This paradigm shift exposes enterprises to new challenges, including cognitive limitations, technological uncertainty, high coordination costs, and capability obsolescence, which exacerbate the risk of falling into the "valleys of death" across the innovation lifecycle. These valleys manifest as critical breaks in the network nodes from basic research through application development to industrialization and scaling.

This study systematically investigates how enterprises can effectively navigate these challenges, focusing on the six distinct types of innovation valleys of death that emerge within this value network context. Adopting the "Awareness-Motivation-Capability" (AMC) perspective, which is highly suited for analyzing enterprise responses to external changes and strategic challenges, we conduct a multi-case study on eight semiconductor subsidiaries of the Changguang Group, an enterprise group integral to China's high-tech development. We track their archival records, patent portfolios, and financial data between 2014 and 2024 to provide robust longitudinal evidence.

Our empirical findings reveal two distinct strategic paths for overcoming innovation valleys, contingent on the type of core technology being pursued.

First, for "chokepoint" technologies (those significantly lagging behind international leaders and facing external restrictions), the most effective strategy follows the path of "social-expectation-based awareness—developing intrinsic motivation—cultivating core capabilities". Social expectation awareness, driven by comparison to industry benchmarks, compels the enterprise to adopt a differentiation strategy to break free from "chain"-based path dependency. Intrinsic motivation, rooted in a sense of mission and long-term commitment, facilitates continuous learning and the tacit knowledge accumulation necessary to overcome network monopolies and master crucial know-how. Finally, cultivating core capabilities allows the firm to achieve rapid product transformation, scale production efficiently, and realize sufficient network benefits, thereby securing the autonomous control and industrialization of critical core technologies.

Second, for "game-changer" technologies (asymmetric technologies originating from internal, primary innovation that are difficult to imitate), the successful approach is "self-expectation-based awareness—developing extrinsic motivation—cultivating dynamic capabilities". Self-expectation awareness, based on the enterprise’s past performance, provides a stable internal anchor for sustained investment in basic research, mitigating the risk of divergent goals among multiple stakeholders (such as government, research institutes, and market-driven firms). Extrinsic motivation, often triggered by external pressures (e.g., policy support, competitive attacks), facilitates the coordination of diverse actors and the infusion of external resources, thereby resolving the challenges of adoption network delays in the transition from application to product. Crucially, dynamic capabilities (sensing, seizing, and reconfiguring) enable the enterprise to quickly adapt to market changes, integrate resources, and reconstruct business models, effectively addressing diffusion network delays and ensuring the rapid dissemination of asymmetric technologies and successful business model innovation.

This study contributes to the literature on innovation management by extending the AMC framework into the context of innovation valleys of death in value networks and providing a differentiated strategic response mechanism for "chokepoint" and "game-changer" technologies. Practically, the findings offer vital guidance for enterprises in critical industries to proactively respond to paradigm shifts, successfully commercialize R&D outcomes, and enhance national competitiveness.