NVIDIA outlines four strategic AI priorities for Vietnam

2. Expand AI infrastructure and GPU capacity

NVIDIA proposed a national computing-capacity strategy serving government agencies, companies, universities and research institutions. At the meeting, the company said Vietnam currently has an estimated 10,000–15,000 GPUs, while South Korea has around 50,000–60,000 and is aiming for 200,000–250,000 GPUs within five years.

The comparison highlights a challenge that goes beyond buying more chips. Vietnam would also need coordinated planning for computing resources, data centers, access for researchers and businesses, and mechanisms for turning infrastructure into useful AI products and services.

The national-level proposal would build on projects involving domestic companies including Viettel, FPT and GreenNode. Viettel’s existing NVIDIA collaboration combines an NVIDIA DGX-based high-performance computing cluster with work on foundational AI models for Vietnam.

3. Train talent and support research and startups

The third pillar focuses on people. NVIDIA proposed expanding its Deep Learning Institute nationwide, integrating more advanced AI training into university curricula, and supporting Vietnamese AI startups, researchers, doctoral candidates and students.

This is a critical complement to infrastructure investment. GPUs and data centers create value only when teams have the expertise to train, evaluate, deploy and operate AI systems. Stronger links between universities, research programs and startups could also help move ideas from laboratories into commercial products.

The proposal is consistent with the planned role of NVIDIA’s research presence in Vietnam. Under an agreement announced in December 2024, NVIDIA and the Vietnamese government agreed to establish an AI research and development center and an AI data center in the country. The facilities were intended to support software development and connect companies, startups, government agencies, universities and students.

4. Take Vietnam’s AI ecosystem into ASEAN and beyond

NVIDIA also proposed helping Vietnamese AI products and companies enter ASEAN markets before expanding globally through the company’s networks.

This places market access alongside technical capability. A national AI ecosystem needs more than models and domestic computing capacity; businesses also need customers, partners, distribution channels and opportunities to deploy their products in markets with related needs.

ASEAN could provide an initial regional step for products designed around local languages and operating contexts. However, the proposal remains a cooperation direction rather than a commitment to specific product volumes, revenue targets or commercialization deadlines.

How the proposals build on existing cooperation

From separate projects to a national framework

Before the August meeting, NVIDIA had already established relationships with Vietnamese companies working on AI infrastructure, models and applications. FPT, Viettel and GreenNode were among the domestic partners referenced in the broader cooperation landscape. The four pillars place those activities in a more connected framework: local models need infrastructure, infrastructure needs skilled people, and products need markets.

Scaling Viettel’s sovereign AI initiative

Viettel’s March 2026 agreement with NVIDIA focused on two foundations of sovereign AI: high-performance computing and foundational models. Viettel said it was deploying NVIDIA DGX B200 systems and developing an AI ecosystem for real-world systems in Vietnam.

The August proposals extend that logic from an enterprise partnership toward capabilities that could serve the wider national ecosystem, including government, businesses, universities and research organizations.

Giving the 2024 government agreement a clearer path to implementation

In December 2024, the Vietnamese government and NVIDIA agreed to establish the Vietnam Research and Development Center and an AI data center. The latest proposals suggest more specific uses for that relationship, including local model and software development, workforce training, startup support and ecosystem coordination.

The next step: a joint action plan

Following the meeting, Minister Vu Hai Quan directed the Ministry of Science and Technology’s focal units to work with NVIDIA on a joint action plan. The plan is expected to define responsibilities, resources, tasks and implementation timelines. NVIDIA agreed to help develop the plan and pursue early, measurable results.

The significance of the meeting therefore lies less in announcing another standalone commitment than in trying to turn existing work on models, GPUs, training and commercialization into a coordinated program with clearer deliverables.

Whether the roadmap produces a stronger Vietnamese AI ecosystem will depend on how it is translated into concrete projects, how resources are allocated and whether the existing initiatives can be connected into a single, functioning system.