The Third Stack: Epistemic Discretion and the Power to Decide

The essay posits that Vietnam’s strategy in global AI infrastructure is based on positioning itself as a co-author of global AI, achieving relevance by accumulating influence at the edge rather than competing centrally. This approach aims to secure "epistemic self-determination," meaning the right to decide what knowledge about AI is created and how it is governed within the national context. The strategy blends clear strategic intent with pragmatic action, involving the willing integration of global technology providers like FPT to develop a domestic "stack" encompassing infrastructure, capability, and education.
The core of this strategy involves controlling the underlying infrastructure—the "stack"—to determine power structures rather than focusing solely on technical performance. This necessitates an open yet sovereign ecosystem that navigates external powers by engaging them productively without conceding control. The essay suggests this path is not about isolation but about assembling a third stack, drawing selectively from global capabilities while cultivating national sovereignty over the shape and deployment of AI systems.

* Vietnam inserts itself as a co-author of global AI infrastructure through open-source, transnational efforts aligned with industrial priorities.
* The strategy focuses on accruing influence at the edge rather than competing at the center.
* A key component involves opening the "core tech stack" (language models, cloud infrastructure, training data) to domestic partners.
* This involves creating a national sandbox for experimentation to develop locally trained GPT-style models and support AI education in schools.
* Infrastructure choices determine power and authority, acting as political decisions rather than purely technical ones.
* The strategy emphasizes an open, comprehensive, and state-regulated electronic ecosystem.
* AI sovereignty is defined as the authorship of the perceptual field itself, deciding what knowledge is perceived and how it is structured.
* Vietnam seeks to assemble a "third stack" by drawing from different global powers without fully aligning with any single one.
* The strategy involves technology transfer, ecosystem development, and workforce capacity building through partnerships like FPT.

The narrative resists the simplistic framing of AI geopolitics as a culture war between established blocs. Instead, it shifts the fault line to infrastructural level, focusing on configurations like proprietary systems versus open-weight models and centralized command versus distributed improvisation. The pattern of inquiry reveals that true sovereignty in this context is not about self-reliance but about epistemic control—determining what counts as knowledge and how it translates into actionable reality for both machines and humans.
The concept of the "third stack" and "epistemic self-determination" functions as a powerful mechanism to challenge the "dictatorship of no alternatives." It suggests that strategic agency is achieved by managing the substrates—data regimes, licensing schemes, and compute dependencies—which are the actual conduits of power in AI governance. The pattern observed is the tension between global openness and localized control; while openness facilitates borrowing, the risk lies in a new perimeter hardening around these systems if external regulatory forces impose strict standards.
The framework implies that successful geopolitical maneuvering requires moving beyond ideological stances to focus on tangible, situated logics embedded in infrastructure. The implication for policy leadership is that national agency resides in designing the structure through which knowledge flows and is processed, rather than merely reacting to external rules. A critical question remains whether infrastructural diplomacy can effectively secure this nuanced epistemic authority against the inherent pressures of systemic closure or fragmentation.