Hanoi (VNA)𒈔 – Artificial intelligence (AI) is opening up unprecedented opportunities for Vietnamese businesses, enabling them to improve operational efficiency, cut costs, increase revenues, and enhance their competitiveness on the global stage.

Vietnam has identified AI as a key strategic technology to elevate its global standing. Under the Poliburo’s Resolution 57-NQ/TW on breakthroughs in science, technology, innovation, and national digital transformation, Vietnam aims to be among the top three ASEAN countries and the top 50 globally in AI development by 2030. The government has also listed AI as a top priority in its national strategy on emerging technologies. Recently, Vietnam introduced its first legal framework for AI with the adoption of the Law on Science, Technology and Innovation, and the Law on the Digital Technology Industry. Phan Thi Thanh Ngoc, Director of AI Consulting at VNPT AI - a subsidiary of the VNPT Group, Vietnamese firms are not just adopting AI quickly, but also building their own ecosystems and platforms to serve domestic needs and compete with global tech giants. VNPT began its AI research and development over five years ago and now offers a diverse range of AI products for multiple customer segments.

In the public sector, VNPT uses AI to enhance digital governance. For individuals, the company is developing virtual AI assistants. In the enterprise sector, its AI-powered solutions help businesses increase revenue, improve profits, and enhance customer experiences. One standout product is VNPT’s AI-based eKYC system, which ranks among the world’s top 10 for accuracy. VNPT has also developed a Vietnamese-optimised large language model (LLM) that currently leads domestic rankings. All these technologies are fully developed in-house, demonstrating Vietnam’s capacity to compete internationally in AI. Tech giant VNG is also making comprehensive investments in AI across infrastructure, platforms, and applications. Its cloud subsidiary GreenNode launched Southeast Asia’s first large-scale AI Cloud in June 2024. Within six months, it had generated significant revenue due to high market demand. Vietnam is now one of the first three Southeast Asian countries to develop its own LLM, with Zalo’s KiLM model achieving performance on par with GPT-4 in just six months. Notably, around 20% of Zalo’s 75 million users—roughly 15 million people—now use AI-powered features like translation and AI avatars daily. AI is also transforming banking operations. Previously, loan processing required large teams to manually enter and cross-check data. With AI document extraction from GreenNode, banks now process millions of documents annually, saving billions of Vietnamese dong in operational costs. Processing time has improved by at least 30%, allowing banks to serve more customers efficiently. GreenNode Product Director Vu Thanh Tung observed that both government and enterprise sectors are accelerating practical AI applications. In public administration, AI is automating resume processing. In insurance, it helps digitise and streamline claims handling and other paperwork-heavy processes. Despite these advancements, broad AI adoption in Vietnamese businesses still faces several barriers including high infrastructure costs, deployment challenges, and limited awareness. “If business leaders lack a broad understanding and long-term vision to assess the value of AI, they may hesitate to invest in it. Some also worry AI could replace jobs,” said Ngoc. She emphasised the need for clear investment strategies and a focus on safety, security, and privacy. “Vietnamese companies should prioritise ‘Made in Vietnam’ AI solutions tailored to local needs,” she advised. From a technical standpoint, Tung pointed out two key concerns of data and security. “With biometric authentication, deepfake attacks are emerging as a new threat. Meanwhile, effective AI deployment depends on having clean, well-organised, and centralised data,” he said. To address this, GreenNode is developing Data Bloom, a platform that consolidates fragmented data sources. This allows even non-technical business teams to access and analyse data for AI applications. “These AI models don’t have to be ChatGPT. There are open-source, low-cost alternatives like DeepSeek, Qwen, or LlaMA,” Tung added./.