Energy is key to Vietnam’s AI sovereignty

Energy is key to Vietnam’s AI sovereignty

Reliable, affordable energy must be treated as a strategic foundation of Vietnam’s AI ambitions, alongside data, chips and computing infrastructure.

According to Dr James Kang, Senior Lecturer in Computer Science at RMIT University Vietnam, artificial intelligence (AI) sovereignty has at least four foundations: algorithms, data, chips and energy. 

“Vietnam is unlikely to compete with the United States and China on all four. It does not have the resources to build the world's largest general-purpose AI models, control the global supply of advanced AI chips, or match the enormous datasets available to the world's biggest technology companies,” Dr Kang said. 

But there is one area where Vietnam can make a strategic choice of its own: energy.

He emphasised: “AI does not live in the cloud. It runs on physical machines in data centres, and those machines need electricity continuously for computing and cooling. If Vietnam wants genuine AI sovereignty, energy sovereignty must be part of the conversation.” 

Reliable energy is vital to Vietnam’s AI future. (Image: Pexels)Reliable energy is vital to Vietnam’s AI future. (Image: Pexels)

AI ambitions will depend on reliable power

The scale of Vietnam's emerging AI infrastructure is already significant. 

Mr Tran Quy Hung, an AI expert and PhD candidate at RMIT, noted that Vietnam currently has 41 active data centres with a combined power capacity of about 221 megawatts. Viettel is also developing a 140-megawatt hyperscale facility in Ho Chi Minh City, while other large projects are being planned.

According to him, the issue is not simply how many megawatts Vietnam can generate, but whether enough electricity can be delivered reliably to facilities that cannot afford interruptions. 

Vietnam may have a better opportunity by developing specialised AI around Vietnamese data, language and industries where local knowledge matters. 

Dr Kang added: “Specialised AI still needs computing infrastructure, which needs data centres. And data centres need reliable electricity. Energy is therefore not separate from AI sovereignty; it is one of its foundations.” 

Cooling could become a strategic advantage

Data centres also require substantial electricity to remove the heat generated by their servers. Studies commonly estimate that cooling accounts for around 30-40 per cent of total data-centre energy consumption, although the proportion varies depending on the facility and technology. 

For Vietnam, the challenge is particularly relevant given its hot climate. The electricity needed to keep computing infrastructure at appropriate temperatures is both an engineering consideration and a significant part of the economics of AI infrastructure.

Dr Kang suggested: “Vietnam's long coastline could offer an opportunity to explore alternative approaches.” 

Rather than locating every future data centre inland and relying heavily on electricity and water for cooling, Vietnam could consider coastal locations and seawater-based cooling systems where environmental and engineering conditions allow.

Mr Hung highlighted international examples showing that alternative cooling approaches are becoming increasingly technically feasible. China has moved towards commercial deployment, with an underwater data centre operating in Hainan and a 24-megawatt underwater facility in Shanghai linked to offshore wind. 

“Vietnam should study these projects rather than simply replicate them,” Mr Hung said. Questions around maintenance, connectivity, corrosion, security and cost remain as the technology develops.

Alt Text is not present for this image, Taking dc:title 'dr-james-kang-and-phd-student-ai-expert-tran-quy-hung-rmit-vietnam.jpg'Dr James Kang (left) and PhD candidate Tran Quy Hung, RMIT Vietnam

A diversified energy strategy for AI

The source of the electricity powering AI infrastructure is another strategic consideration. 

Mr Hung said: “Vietnam should pursue a mix of technologies rather than rely on a single energy source, with renewable energy, grid expansion, energy storage and more efficient data-centre design all playing a role”.

Offshore wind is particularly relevant. The World Bank estimates Vietnam's offshore wind technical potential at hundreds of gigawatts and has identified it as an important future source of electricity. 

He also suggested that nuclear power could also contribute to the longer-term energy strategy. Vietnam has restarted its nuclear program, and in March 2026 signed an agreement with Russia for Ninh Thuan 1, involving two conventional Russian-designed reactors with a combined capacity of 2,400 megawatts.

Dr Kang said conventional nuclear power and small modular reactors (SMRs) should be considered as different layers of the energy strategy. 

“SMRs still require technological, economic and regulatory development. The large Ninh Thuan project and future SMRs should be seen as two different layers of the energy strategy: conventional nuclear can strengthen the national grid, while SMRs could eventually become an option for dedicated industrial and computing loads if the technology proves commercially and technically suitable,” he said.

AI sovereignty does not mean producing everything domestically but having sufficient control over its critical foundations. (Image: Pexels)AI sovereignty does not mean producing everything domestically but having sufficient control over its critical foundations. (Image: Pexels)

Turning infrastructure into Vietnamese capability

Dr Kang suggests three priorities for Vietnam.

First, reliable and affordable electricity should be treated as part of the national AI strategy rather than simply a utility issue.

Second, Vietnam should identify locations where energy, data-centre infrastructure, fibre connectivity and cooling can be developed together, including coastal locations where seawater cooling and offshore wind may offer potential advantages. 

Third, foreign data-centre investment should contribute to Vietnamese capabilities through skills development, research, technology partnerships and local AI applications.

“The objective should not be to become the place where the world's AI machines happen to sit. It should be to turn the infrastructure needed to run AI into a source of Vietnamese capability,” Dr Kang said.

Story: June Pham

18 August 2026

Share

Engage

Related news

aerial-view-power-plant-cooling-towers.jpg

Energy is key to Vietnam’s AI sovereignty
Media Release
Icon / Small / CalendarCreated with Sketch. 18 Aug 2026

Reliable, affordable energy must be treated as a strategic foundation of Vietnam’s AI ambitions, alongside data, chips and computing infrastructure.

Young boy holding a smartphone

Vietnam's social media proposal raises questions beyond child safety
Media Release
Icon / Small / CalendarCreated with Sketch. 29 Jul 2026

Vietnam's proposed restrictions on social media use for under-16s could improve child safety, but experts say effective implementation and unintended consequences must be considered.

thumbnail-vietnam-path-to-artificial-intelligence-sovereignty-rmit.jpg

Vietnam’s path to sovereign AI lies in specialisation, not imitation
Media Release
Icon / Small / CalendarCreated with Sketch. 27 Jul 2026

Vietnam's path to artificial intelligence (AI) sovereignty may not lie in competing head-to-head with global technology giants, but in developing specialised AI systems tailored to local needs, according to an RMIT expert.

Photo of 5 people sitting around a table in a meeting room

Vietnam-Australia team advances concrete fracture research
Media Release
Icon / Small / CalendarCreated with Sketch. 07 Jul 2026

Supported by an Australian Government-funded initiative, a collaboration between Vietnam- and Australia-based researchers has led to a pioneering publication in construction engineering.

Load more +