AI Could Add $120 Billion to $130 Billion to Vietnam’s Economy by 2040

A $130 Billion Forecast With Productivity at Its Core

Artificial intelligence could contribute $120 billion to $130 billion to Vietnam’s economy by 2040, according to the Vietnam AI Economy Report 2025, developed by the National Innovation Centre (NIC) and the Japan International Cooperation Agency (JICA), with support from Boston Consulting Group (BCG). The forecast combines new demand for products and services using AI with gains from businesses and public institutions working more efficiently.

The Vietnam Journal of Science and Technology’s account of the report puts revenue from consumer demand at $45 billion to $55 billion and productivity related cost savings at $60 billion to $75 billion. Those figures make efficiency gains a larger component of the opportunity than new sales.

Reports also describe the headline projection as equivalent to 10% to 12% of Vietnam’s economy by 2040. That is a forecast, not a measured contribution today, and the different descriptions of revenue, savings and economic value need care when interpreting the total.

The report was launched in Hanoi on June 12, 2025, according to the journal and most accounts of the event. One account gives June 13 as the release date. NIC Director Vu Quoc Huy described it as the first comprehensive overview of Vietnam’s AI economy, covering opportunities across industries, the workforce, infrastructure and policy.

The practical challenge is larger than creating successful AI companies. Reaching the projected scale would require the technology to spread through manufacturing, finance, healthcare, education and government services, while producing useful results that people can verify and institutions can afford to deploy.

What the Economic Figures Actually Measure

The forecast has two main channels. The first is revenue generated by demand for products and services that incorporate AI. The second is the value of increased productivity, including savings from automation, predictive analysis and more efficient operations. Predictive analysis uses patterns in data to estimate what may happen next, helping organisations plan rather than simply react.

There is a difference between accounts of the second channel. The science and technology journal gives a range of $60 billion to $75 billion in cost savings. Another account uses the same range but describes it as increased profit from productivity. Other reports give $60 billion to $70 billion in savings.

The distinction matters because revenue, cost savings and profit are not interchangeable measures. Adding the lowest consumer revenue estimate, $45 billion, to the lowest savings estimate, $60 billion, gives $105 billion. Adding the upper estimates in the journal’s account, $55 billion and $75 billion, gives $130 billion. The published accounts do not explain how these component ranges reconcile with the headline total of $120 billion to $130 billion.

The figures therefore describe the potential scale and sources of economic value, rather than a fully explained calculation of additional gross domestic product (GDP). They should not be read as a guaranteed increase in national output or as revenue that would flow entirely to AI developers.

Public Services Are a Central Test

Public administration, education and healthcare are identified as three sectors with strong potential for AI breakthroughs. The report links healthcare applications to reducing disparities between regions, education applications to making better use of educational data, and government applications to improving public management.

The journal’s account sets out a broader group of immediate priorities: government services, environmental and disaster management, healthcare, financial services and transportation. Manufacturing and finance also feature in descriptions of the report’s industry analysis.

These priorities connect the economic forecast to services people use directly. Better public management or more effective healthcare can create value without generating a new consumer purchase. That helps explain why the productivity component is so prominent in the estimates.

The recommended approach is gradual: identify applications capable of delivering results quickly, demonstrate that they work, then expand their use. The journal describes cooperation between government and business, including a proposed national champion enterprise programme, as a way to move proven applications into wider use. It does not give a deployment timetable or measured results for those proposals.

Three Priorities and the Foundations Beneath Them

The report’s recommendations centre on three direct growth priorities: wider adoption in the public and private sectors, support for AI startups, and a deeper pool of specialist talent. Each addresses a different part of the economic forecast. Adoption creates demand and productivity gains; startups develop applications; skilled workers build, adapt and supervise the technology.

Supporting recommendations cover AI literacy, access to reliable data and computing infrastructure, and clear rules. AI literacy means understanding what these systems can do, where they can fail and when their output needs checking. The recommendations include education about risks and ethics, rather than treating familiarity with an AI tool as sufficient preparation.

Accounts describe the report’s organisation differently. One presents five broad pillars: technology, the startup ecosystem, labour, resources and international collaboration. More detailed descriptions set out the three direct growth priorities and three supporting foundations. These are different presentations of the strategy, not necessarily conflicting policy programmes.

Data access is a concrete concern. The recommendations call for better access to quality data for small businesses, universities and research institutions, alongside computing resources. They also refer to progress associated with Vietnam’s first National Data Centre. Without usable data and adequate computing capacity, an expanding workforce alone cannot deliver the projected gains.

A Large Graduate Pipeline Is Not Yet an AI Workforce

Vietnam has more than 150 universities offering courses related to information technology, with more than 60,000 graduates entering the workforce annually, according to the science and technology journal. Hanoi University of Science and Technology has introduced bachelor’s and master’s programmes in AI.

Those figures indicate a substantial education base, but they do not mean Vietnam produces 60,000 AI specialists each year. General IT training and the skills needed for advanced AI research, development and oversight are different measures.

The report recommends AI curricula meeting international standards at bachelor’s, master’s and doctoral levels, together with training delivered at scale. The journal names Japan’s GCI programme as an example and proposes structured research partnerships with international laboratories and corporations, beginning with leading institutions such as Hanoi University of Science and Technology.

JICA senior representative Kubo Yoshitomo said the agency had begun concrete initiatives with NIC and leading Vietnamese science and technology universities. Separately, accounts of the launch place this work alongside JICA’s DX Lab, established in 2022 to accelerate digital innovation in developing countries. The accounts do not specify enrolment targets or funding commitments for the proposed expansion of AI training.

Startup Strength Needs Capital and Customers

Vietnam ranked second in Southeast Asia in 2024 for both the number of AI startups and investment focused on AI, according to the journal’s account. The ranking points to activity in the sector, although the account does not provide company counts, investment totals or the underlying ranking methodology.

The recommendations seek to turn that activity into businesses capable of competing domestically and internationally. An AI startup acceleration programme would provide business and technical mentoring, investment opportunities and partnerships. National funding programmes are proposed to improve access to capital on favourable terms.

These measures address more than the cost of developing a model. A startup also needs customers, technical support and opportunities to demonstrate that its application works in an organisation’s daily operations. The report connects startup development to broader adoption rather than treating company formation as an end in itself.

Arnaud Ginolin, a BCG managing director and partner leading its public sector and industrial goods work in Vietnam, cited digital economy growth of up to 20% annually. Another account describes growth above 20%, while others give 20%. All indicate rapid expansion, but digital economy growth is not the same measure as AI’s contribution to GDP.

The Policy Framework and Key Dates

The strategy sits alongside several named policy instruments. Resolution No. 57-NQ/TW concerns science, technology, innovation and national digital transformation. Resolution No. 68-NQ/TW supports development of the private economic sector. National Assembly Resolution No. 193/2025/QH15 provides for piloting special mechanisms to advance science and technology.

The journal also identifies Decision No. 127/QD-TTg as the national AI strategy and describes AI as a strategic national priority for 2030. These measures establish policy direction, but the economic forecast still depends on implementation, investment and adoption.

• 2022: JICA established its DX Lab to support digital innovation in developing countries.
• 2024: Vietnam ranked second in Southeast Asia for AI startup numbers and investment focused on AI.
• June 12, 2025: The Vietnam AI Economy Report 2025 was launched in Hanoi. One account gives June 13 instead.
• First quarter of 2026: A later account cites an AI adoption rate of 26.5%, compared with 23.5% in 2025.
• 2030: Vietnam’s strategy identifies AI as a national priority.
• 2040: The report projects an AI economy worth $120 billion to $130 billion.

The accounts do not set out annual milestones for reaching the 2040 total, a complete programme budget or a schedule for measuring progress against the forecast.

Later Adoption Figures Show Progress, Not Proof

A later account citing Microsoft’s Global AI Diffusion Report puts Vietnam’s AI adoption rate at 26.5% in the first quarter of 2026, up from 23.5% in 2025. That is an increase of 3 percentage points, or about 12.8% relative to the earlier rate.

The same account gives Singapore an adoption rate of 63.4%, Malaysia 21.8%, the Philippines 20.1% and Thailand 12.4%. Vietnam’s figure is above the latter three but 36.9 percentage points below Singapore. These comparisons concern adoption, not the financial return from using AI.

At the CRIF Forum Vietnam 2026, NIC Deputy Director Do Tien Thinh described extensive use within government and the startup community. Speaking about government officials and civil servants, he said:

Among government officials and civil servants, more than 90 per cent are currently using AI on a daily basis,

That statement concerns a specific workforce group and is not directly comparable with the broader adoption rate. The later account also places AI’s current economic contribution at around $40 billion to $50 billion, but does not explain its calculation. It cannot reliably be used as a baseline for measuring progress towards the 2040 forecast.

Similarly, studies by NIC with Google and BCG link nearly 30% of Vietnam’s GDP growth to digital technology. That category includes AI alongside other technologies. It is neither a claim that AI produces 30% of GDP nor a separate confirmation of the $130 billion projection.

Employment and Trust Will Shape the Returns

The productivity forecast also has a labour dimension. Thinh said corporations working with NIC had hired relatively few additional IT employees over the previous one or two years as they adopted AI. He described companies with thousands of skilled IT workers needing fewer people to check and validate results produced by the technology.

That account suggests slower recruitment in some businesses, not a measured national total of jobs lost. It nevertheless shows why graduate numbers alone are an incomplete measure of readiness. Training must address how work changes, including the judgement required to verify automated results.

The later account describes Vietnam’s Law on AI as classifying systems by high, medium and low risk, while providing mechanisms for infrastructure, data, workforce development and a national AI development fund. It also describes controlled testing programmes, known as sandboxes, lasting one to three years with a possible further three year extension. It does not give the law’s number or commencement date.

Data reliability presents another constraint. Sachin B Singh, head of enterprise solutions for Asia Pacific at Dow Jones, cited a study suggesting more than 40% of information or answers received from AI came from sources such as Reddit and Wikipedia. Without the study’s methodology, that figure should not be treated as a universal measure of AI output.

His practical concern was traceability: organisations need to establish where an answer came from and whether its underlying information is reliable. That becomes more consequential with agentic AI, systems that can take actions rather than simply generate text. The economic opportunity depends on useful automation supported by verification and human accountability, not adoption alone.

Key Points

• The Vietnam AI Economy Report 2025 projects $120 billion to $130 billion in economic value by 2040.
• The science and technology journal gives $45 billion to $55 billion in consumer revenue and $60 billion to $75 billion in productivity savings.
• Other accounts give a lower savings ceiling of $70 billion; the component ranges are not fully reconciled with the headline total.
• Priority areas include government services, healthcare, education, finance, transportation and environmental management.
• Vietnam’s education base includes more than 150 universities offering IT courses and more than 60,000 graduates annually, not all of them AI specialists.
• A later account puts AI adoption at 26.5% in early 2026, compared with 23.5% in 2025.
• Funding, specialist skills, reliable data and verified results remain central to turning the forecast into economic gains.