Vietnam’s artificial intelligence industry is entering a new phase. After years of focusing on AI applications such as chatbots, speech recognition and computer vision, the country’s largest technology companies are now investing in something far less visible but arguably more important: the computing infrastructure that powers AI.
Across the country, companies are building AI factories, GPU cloud platforms, high performance data centers and AI supercomputers using NVIDIA’s latest hardware. The goal is not only to serve domestic demand but also to reduce dependence on overseas cloud providers and strengthen Vietnam’s ambitions in sovereign AI.
Although Vietnam remains far behind AI powerhouses such as the United States and China, investments made over the past two years suggest the country is laying the foundation for a domestic AI ecosystem capable of supporting enterprises, government agencies and researchers.
FPT: Building Vietnam’s first AI Factory
Among Vietnamese technology companies, FPT has made the most aggressive public investment in AI infrastructure.
In April 2024, the company announced a US$200 million investment to build an AI Factory based on NVIDIA technology, making it one of Southeast Asia’s largest AI infrastructure projects.
Unlike a traditional data center, an AI Factory is designed specifically for AI development. It combines GPU clusters, high speed networking, storage systems and software platforms that allow customers to train, fine tune and deploy large AI models.
FPT initially built its AI Factory using NVIDIA DGX H100 systems equipped with H100 Tensor Core GPUs, before expanding the platform with NVIDIA H200 GPUs as demand for larger language models increased.
In 2026, FPT became one of the first companies in Asia to announce deployment of NVIDIA HGX B300, built on NVIDIA’s Blackwell Ultra architecture.
Each HGX B300 platform integrates eight Blackwell Ultra GPUs connected through fifth generation NVLink and provides approximately 2.1 terabytes of HBM3e GPU memory, enabling significantly larger AI workloads than previous generations.
FPT says its AI Factory platform now offers access to more than 1,000 GPUs through GPU Cloud services, allowing businesses to rent computing capacity on demand instead of investing millions of dollars in dedicated AI servers.
The company is also expanding the platform beyond Vietnam, targeting customers in Japan and across Southeast Asia.
Viettel: Building sovereign AI infrastructure
While FPT is targeting commercial AI services, Viettel has positioned AI infrastructure as part of Vietnam’s national digital strategy.
In early 2026, the telecommunications group announced that it had begun operating Vietnam’s first NVIDIA DGX B200 AI supercomputer.
The DGX B200 is NVIDIA’s flagship enterprise AI platform, integrating eight Blackwell GPUs connected through fifth generation NVLink. According to Viettel, the system delivers approximately 1.5 exaFLOPS of FP8 AI performance, making it one of the country’s most powerful publicly announced AI computing platforms.
Viettel has not disclosed the number of DGX systems or GPUs deployed, making it impossible to estimate the total size of its AI cluster.
Instead of competing directly with public cloud providers, Viettel is focusing on applications requiring high levels of security and data sovereignty, including government services, telecommunications, cybersecurity and regulated industries.
Its nationwide fiber network, cloud infrastructure and data center footprint provide a strong foundation for expanding AI services across the country.
VNG: Bringing enterprise AI to the cloud
Another major player is VNG, which has chosen a cloud first strategy.
In 2025, the company merged VNG Cloud and GreenNode, its AI infrastructure business, into a unified AI platform designed to deliver infrastructure, AI platforms and enterprise AI services under one brand.
Rather than building dedicated AI supercomputers, VNG focuses on providing GPU computing as a cloud service.
According to the company, its AI Platform supports GPU instances powered by NVIDIA HGX H100, NVIDIA L40S, RTX 4090 and NVIDIA A40 accelerators connected through high speed InfiniBand networking.
Customers can use these resources for AI model training, fine tuning and inference while also accessing more than 100 open source foundation models, including Llama, Qwen and DeepSeek.
VNG has also expanded its cloud infrastructure in Ho Chi Minh City by adding two new Availability Zones to improve redundancy and reliability for enterprise workloads.
Its strategy reflects growing demand among Vietnamese businesses that want AI computing capacity without investing in expensive GPU clusters themselves.
VNPT: AI infrastructure for digital government
State owned telecommunications operator VNPT is integrating AI into its existing cloud and digital government infrastructure.
Instead of announcing large GPU clusters, the company has focused on expanding VNPT Cloud, which supports applications in e-government, smart cities, healthcare, education and enterprise digital transformation.
VNPT executives have repeatedly highlighted the importance of sovereign AI, arguing that organizations should maintain control over their data, AI models and computing infrastructure instead of relying entirely on overseas AI providers.
While VNPT has confirmed continued investment in AI cloud services and data centers, it has not publicly disclosed the GPU hardware powering these platforms.
CMC: Expanding AI ready cloud infrastructure
CMC is pursuing a strategy centered on enterprise cloud services.
The company continues expanding its data center capacity while integrating GPU computing into CMC Cloud, enabling enterprises to access AI infrastructure without purchasing dedicated hardware.
Although CMC has confirmed investments in AI cloud services, it has not publicly disclosed detailed specifications such as GPU models or cluster size.
Its focus is on providing AI infrastructure as a service for businesses adopting generative AI, data analytics and machine learning applications.
Why AI infrastructure matters
Modern AI systems require enormous computing power.
Training a large language model may require thousands of GPUs working simultaneously, while inference services supporting millions of users also demand substantial computing resources.
As a result, AI infrastructure has become a strategic asset alongside electricity, telecommunications and cloud computing.
Countries able to build domestic AI infrastructure gain greater control over sensitive data, reduce dependence on overseas providers and improve their ability to develop national AI capabilities.
Vietnam’s recent investments reflect this broader global trend.
Still behind regional leaders
Despite rapid progress, Vietnam remains in the early stages of AI infrastructure development.
Singapore hosts hyperscale AI data centers operated by Amazon Web Services, Google Cloud, Microsoft Azure and Oracle.
Malaysia has attracted billions of dollars in AI infrastructure investments from NVIDIA partners and hyperscale cloud providers.
China continues deploying AI supercomputers containing tens of thousands of GPUs, while the United States remains home to the world’s largest AI clusters operated by companies such as OpenAI, xAI, Meta and Microsoft.
By comparison, Vietnam’s largest publicly announced AI projects currently involve GPU clusters measured in thousands, rather than tens of thousands, of GPUs.
However, analysts say Vietnam is pursuing a different strategy.
Rather than competing directly with global hyperscalers, the country is focusing on building sovereign AI capabilities that allow government agencies and domestic enterprises to develop and deploy AI while keeping sensitive data inside Vietnam.
The road ahead
Vietnam’s AI race is no longer defined by who develops the best chatbot.
Instead, competition is increasingly centered on who controls the country’s computing infrastructure.
FPT is building AI factories powered by NVIDIA’s latest Blackwell systems. Viettel is investing in AI supercomputing. VNG is expanding enterprise GPU cloud services, while VNPT and CMC continue strengthening cloud infrastructure for government and businesses.
Together, these projects represent the first generation of Vietnam’s AI infrastructure.
Whether the country can become a regional AI leader will depend not only on software innovation, but also on how quickly it can continue expanding GPU capacity, AI optimized data centers and high performance cloud platforms in the years ahead.
