From "Applying AI" to "Running on AI": Việt Nam's New National AI Strategy, Quyết định 1671/QĐ-TTg

Việt Nam has replaced its national AI strategy. Quyết định 1671/QĐ-TTg, signed on 28 August 2026 by Deputy Prime Minister Hồ Quốc Dũng on behalf of the Prime Minister, approves the “Chiến lược quốc gia về trí tuệ nhân tạo đến năm 2030, tầm nhìn đến năm 2045” (National Strategy on Artificial Intelligence to 2030, with a vision to 2045). It took effect on the day it was signed (Điều 3 k1). It was drafted by Bộ Khoa học và Công nghệ (the Ministry of Science and Technology, MOST) under Tờ trình 158/TTr-BKHCN of 28 July 2026.

The decision is five weeks old. It is news this week because MOST has now started work on it. At the ministry’s press conference on 30 September 2026, Deputy Minister Trần Trung Tính listed carrying out the strategy among MOST’s priority tasks, along with expanding AI support for civil servants. Several of the strategy’s own deliverables also fall due in 2026, which leaves under three months.

What it replaces

Điều 3 k2 states that the decision replaces Quyết định 127/QĐ-TTg of 26 January 2021, the “Chiến lược quốc gia về nghiên cứu, phát triển và ứng dụng Trí tuệ nhân tạo đến năm 2030” (National Strategy on Research, Development and Application of AI to 2030). Programmes and projects already running under the 2021 strategy continue until they finish or are folded into the new one (Điều 3 k3).

A new strategy was expected. The implementation plan for Luật Trí tuệ nhân tạo 134/2025/QH15 (the AI Law, in force since 1 March 2026), issued as Quyết định 367/QĐ-TTg on 28 February 2026, required the strategy to be issued in 2026 and reviewed at least once every three years.

The title shows the change in approach. The 2021 strategy was about “research, development and application” of AI. The first principle of the 2026 strategy is a move to “chuyển đổi AI quốc gia” (national AI transformation). AI should stop being an add-on to existing products and processes and become “a core capability from the moment a problem is defined, designed, organised and operated”. Institutions and AI governance are described as the “breakthrough”, with people, infrastructure and data as the foundation.

The 2021 targets and the 2026 targets

	Quyết định 127/QĐ-TTg (2021)	Quyết định 1671/QĐ-TTg (2026)
Regional ranking by 2030	Top 4 in ASEAN, top 50 in the world	Top 3 in Southeast Asia in AI R&D
Long-term vision	—	Top 10 in Asia by 2045
Brands	10 reputable AI brands in the region	At least 10 AI brands competitive regionally and internationally
Compute	3 national big-data and HPC centres	About 560 EFLOPS BF16 (about 1,100 EFLOPS FP8), roughly 500 MW of power
Centres	3 national AI innovation centres	3 AI centres of excellence in Hà Nội, Đà Nẵng and TP Hồ Chí Minh

Phụ lục I lists 32 indicators, with an owner for each. The headline 2030 targets in Điều 1 are:

An AI economy worth about 6% of GDP, productivity gains of at least 10–15% in sectors that use AI, and about US$5 billion of exports of AI-integrated products and services.
Eight Vietnamese-language AI models, two in each size band: under 7 billion parameters, 7 to under 70 billion, 70 to under 300 billion, and 300 billion or more. A national platform for evaluating Vietnamese-language models will also be set up.
AI in the state: 100% of online public services integrated with and supported by AI, 100% of administrative files supported by AI processing, 100% of civil servants using an official AI platform every week, and 60% of state steering decisions supported by data and AI.
People: 10,000 advanced AI specialists, 50,000 graduates able to apply AI, 500,000 workers retrained, and 10 million trained workers with basic AI skills.
Business: at least 100 AI start-ups valued above US$10 million, 500,000 SMEs, household businesses and cooperatives supported with AI tools, and 60% of AI-adopting firms reporting measurable gains.
Money: total AI investment of 1–1.5% of GDP a year, with about 0.6% of annual state budget spending going to AI.

The 97 tasks, and the ones due in 2026

Phụ lục II lists 97 tasks in 14 groups, each with a lead agency and a deadline. Most run from 2026 to 2030, but several are due this year or next:

Bộ Công an (Ministry of Public Security) is to issue rules or guidance on controlling cross-border data flows in AI activity and export controls on sensitive AI technology in 2026 (task 73), and mechanisms for opening, sharing and securely accessing data in 2026 (task 12). It is also to form a data market and data-sharing platforms for AI in 2026 (task 11), and rules on property rights in data, IP and lawful use of content for AI in 2027 (task 13). Under Điều 2 k3c, it also builds and runs the Cơ sở dữ liệu quốc gia về trí tuệ nhân tạo (National AI Database).
MOST is to release a first version of the national AI platform for public servants by the end of 2026 (task 34). It also owns a 2026 set of indicators for measuring how far each ministry and province has transformed with AI, with annual public rankings (tasks 94–95). The national AI ethics framework and its monitoring mechanism are listed for 2026 (task 64), and a benchmark suite for Vietnamese-language models for 2027 (task 15).
Bộ Nội vụ (Ministry of Home Affairs) is to issue a national scheme for retraining workers at risk of being replaced by AI in 2027 (task 3).

Who it affects

A strategy is a policy decision, not a decree, so it sets no fines or licensing duties for private companies. In practice it still matters to businesses in three ways.

First, the state as a customer. Ministries and provinces must publish their priority AI problems, the data available, their procurement needs, selection criteria and expected budgets so that businesses can bid (Điều 2 k8g). The strategy also tells them to favour “Make in Việt Nam” AI products in IT, digital-government and public-service projects, with preferential treatment in contractor selection. Local AI vendors should watch for these lists. Foreign vendors should expect a preference for domestic suppliers.

Second, sovereignty language. The strategy prefers open-source and open-weight models that can be inspected and deployed independently. It warns against dependence on any single market, platform, model or supply chain. It asks for FDI in AI research centres to come with technology transfer, localisation and, “in cases provided by law”, data stored in Việt Nam. The cross-border data-flow rules due from Bộ Công an in 2026 are the item foreign AI providers should follow most closely.

Third, expectations for companies. Điều 2 k9 and k10 ask businesses, especially major digital-technology firms and key state-owned enterprises, to adopt their own AI plans, retrain staff, fund retraining for workers displaced by automation, and comply with the data, AI and cybersecurity laws. These are policy expectations, not legal duties, but state-owned enterprises tend to treat them as instructions.

Open questions

Deadlines for ministry plans. Every ministry and province must issue an implementation plan (indicator 13), but Điều 2 sets no deadline for it. Whether one follows from the 30 September briefing remains to be seen.
The ethics framework. LuatVietnam reports that a national AI ethics framework has existed since 10 March 2026, yet task 64 still lists “issuing” the framework and its monitoring mechanism as a 2026 deliverable. It is unclear whether this means a revision or a separate monitoring instrument.
Two AI databases. The National AI Database under Bộ Công an and the compliance infrastructure under the AI Law and its implementing decree will need clearly divided roles.
Tax incentives. Tax breaks for AI work and for AI data centres are listed only as items to “study” (tasks 76 and IV.1c), with no date attached.

Sources

This post is general information, not legal advice.