NVIDIA’s Four Strategic AI Proposals for Vietnam

The emphasis on Vietnamese ownership makes this more than a request for access to a foreign chatbot. It points toward a domestic model that can be adapted to local language and use cases while remaining part of a broader technology ecosystem.

2. Expand national AI infrastructure and GPU capacity

NVIDIA recommended that Vietnam create a national plan for AI computing capacity serving government agencies, companies, universities and researchers. The company estimated Vietnam’s current capacity at approximately 10,000–15,000 GPUs. South Korea was cited as a comparison, with roughly 50,000–60,000 GPUs and a stated plan to reach 200,000–250,000 within five years.

Those South Korean figures are a benchmark for the scale of infrastructure under discussion—not a confirmed target or purchase commitment for Vietnam. The practical challenge is to coordinate access to computing across public research, education and commercial development rather than leaving capacity concentrated in a small number of organizations.

3. Train more people and support AI startups

The third proposal focuses on the workforce and innovation pipeline. NVIDIA wants to expand its Deep Learning Institute programs across Vietnam and integrate them more deeply into education and training. It also proposed supporting Vietnamese AI startups, researchers, graduate students and undergraduates through its ecosystem.

This pillar addresses the human-capital bottleneck behind AI development. More GPUs alone will not produce useful national models or competitive applications without people who can train, evaluate, deploy and commercialize them.

4. Connect Vietnam’s AI ecosystem to regional and global markets

NVIDIA also proposed helping bring Vietnamese AI products, companies and capabilities first into ASEAN markets and then to the wider global market.

That gives the plan an economic objective beyond domestic adoption. Vietnam would be positioned not only as a user of AI infrastructure and models, but also as a source of products and services that can scale outside the country.

How the proposals build on existing partnerships

The four pillars extend work NVIDIA is already doing with Vietnamese technology companies, including FPT, Viettel and GreenNode, on AI infrastructure and capability development.

NVIDIA’s sovereign-model proposal also follows an earlier government collaboration. In December 2024, NVIDIA announced work with Vietnam on an AI research and development center intended to focus on software development and collaboration with enterprises, startups, government agencies and universities.

Viettel’s sovereign-AI initiative provides another domestic foundation for the proposed program. Its work with NVIDIA focuses on high-performance computing infrastructure and Vietnamese AI foundation models, including a cluster of 22 NVIDIA DGX B200 systems reported as capable of up to 1.5 exaFLOPs.

Together, these partnerships suggest that the August proposals are intended to coordinate and scale existing activity rather than begin from zero. FPT, Viettel and GreenNode represent infrastructure, model-development and deployment capabilities that could support the national framework, although the meeting itself did not specify how responsibilities would be divided among them.

What happens next

Vu Hai Quan said the proposals align with Vietnam’s strategy for science, technology, innovation and digital transformation. He asked ministry focal units and NVIDIA to translate the ideas into a joint action plan covering tasks, resources and an implementation roadmap. NVIDIA said it was ready to help develop the plan and pursue early deliverables.

The key test will be execution. Vietnam will need to turn four broad cooperation areas into measurable decisions about model ownership, data governance, shared computing access, training programs, startup support and export pathways. Until that roadmap is published, the August 20 meeting should be read as a strategic framework—not as a completed national AI deployment deal.