Artificial intelligence has transformed the financial sector over the past three years, and Vietnam is expected to be one of the region’s fastest-growing AI-fintech markets. According to IMARC Group, the market is projected to reach US$607.24 million by 2033, growing at an annual rate of more than 21%.
Yet despite the optimism, widespread AI adoption remains elusive.
According to the Vietnam AI Adoption Status & IT Hiring Report 2025, while around 73% of organisations have adopted AI in some form, nearly six out of seven companies remain stuck between pilot projects and full operational deployment.
The gap between experimentation and real business impact reveals a deeper challenge facing Vietnam’s fintech sector — one that cannot be solved simply by investing in more AI.
Highlights
AI Doesn’t Fail First — Data Does
The biggest obstacle is not AI itself.
It is data.
Unlike traditional software, AI learns from patterns within data rather than following predefined instructions. As a result, the quality of that data — its accuracy, consistency, completeness, and relevance — directly determines the quality of AI-generated decisions.
In financial services, that relationship becomes even more critical.
“This is particularly important,” said Hieu Nguyen, CEO of FiinGroup, “because every credit decision, risk assessment, compliance review, fraud alert, and investment recommendation ultimately depends on data quality, consistency, and traceability.”
Poor, outdated, incomplete, or biased data inevitably produces unreliable results.
Unfortunately, this remains the reality for many SMEs across Southeast Asia, including Vietnam.
“Over the years, many firms have accumulated a lot of data,” said Chia Hock Lai, Chairman of the Responsible Fintech Institute, “but it is messy, inconsistent, and scattered across legacy systems, so AI cannot learn or operate reliably on top of it.”
FiinGroup’s 2025 Report found that fragmented data across banking, accounting, tax, and e-commerce platforms continues to limit AI implementation in Vietnam’s SME lending sector.
Without standardised datasets, many lenders still rely heavily on manual credit assessment, reducing both the effectiveness and scalability of AI-powered lending.
“In financial services, trust and compliance are non-negotiable,” Hieu added.
When AI systems cannot produce trustworthy or explainable decisions because of poor data, businesses become understandably reluctant to deploy them at scale.
AI Is Not a Transformation Strategy
Even with clean data, successful AI adoption requires more than technical implementation.
It demands organisational change.
Speaking to ThanhNienNews, Tran Kim Sa, Director of Solutions Consulting, observed that many local businesses tend to “start with big problems and build AI models with high expectations.”
Hieu sees a similar pattern.
Many fintech companies, he said, “adopt AI as an application rather than embedding it into business processes.”
The misconception is that introducing AI automatically leads to transformation.
It does not.
AI only creates value when it becomes part of day-to-day operations — from customer onboarding and fraud detection to credit assessment, compliance monitoring, collections, underwriting, and portfolio management.
“Deploying a chatbot, purchasing AI licences, or running a successful pilot does not necessarily create business value,” Hieu said.
As a result, companies often invest more money into improving AI models when the real bottlenecks lie elsewhere — in data quality, workflow design, governance, or management discipline.
Without addressing those underlying issues, many AI initiatives struggle to move beyond the pilot stage.
The SMEs That Win Start With the Problem, Not the AI
Limited funding, shortages of skilled talent, and a lack of AI expertise often make SMEs cautious about expanding AI initiatives beyond experimentation.
However, these characteristics can also become competitive advantages.
Smaller organisations are often more agile, enabling them to adapt quickly, test new approaches, and implement focused AI solutions faster than larger enterprises.
Across ASEAN, several fintech success stories demonstrate this approach.
Surprisingly, they did not begin with AI.
“They started with a business problem,” Chia Hock said.
Digital lending platforms such as Kredivo and Akulaku focused on improving credit decisions for underserved borrowers rather than launching company-wide AI transformation programmes.
Instead of relying solely on traditional credit histories, they used alternative data — including transaction records, device information, behavioural patterns, and digital footprints — to evaluate borrowers with limited financial histories.
The strategy not only helped maintain portfolio quality but also improved credit-risk assessment for thin-file customers, expanding access to financing for millions of Indonesians and supporting broader financial inclusion.
For Chia Hock, AI success is measured by business outcomes rather than technological sophistication.
Those outcomes include:
- Faster customer onboarding
- Higher approval rates
- Broader financial inclusion
- Better risk management
What Actually Separates AI Leaders From Everyone Else
Many of today’s leading fintech success stories were built using traditional AI and machine learning rather than the latest generative AI models.
The underlying principles, however, remain unchanged.
According to Chia Hock, organisations should:
- Strengthen their data foundation before launching AI pilots.
- Begin with modular, department-level projects before attempting enterprise-wide transformation.
- Invest in developing existing employees.
- Build leadership support and an organisational culture that embraces AI.
Hieu believes another factor distinguishes AI leaders from the rest.
“Leading fintech companies treat AI as an organisational capability, not just an IT project,” he said.
As AI models become increasingly accessible, a sustainable competitive advantage will come from three assets:
- Proprietary data
- Deep domain expertise
- Strong operational integration
“The companies that combine these three elements effectively are the ones that move beyond pilot mode and create sustainable advantages,” he said.
Vietnam’s AI Future Won’t Be Decided by Better Models
As competition intensifies and customer expectations continue to evolve, the cost of remaining in perpetual pilot mode may soon exceed the risks of full deployment.
The next phase of AI adoption is unlikely to be defined by larger language models or bigger technology budgets.
Instead, it will be determined by how effectively businesses use AI to make faster, smarter, and more reliable operational decisions every day.
Whether Vietnam ultimately captures the projected growth of its AI-fintech market will depend less on access to AI itself and more on how quickly companies shift their mindset — from viewing AI as another technology initiative to treating it as a core capability that shapes how they compete, serve customers, and create value.
Until that shift happens, many firms may continue investing in AI without ever truly becoming AI-driven.
Editor’s Note:
This article was contributed by NewInAsia Senior Editorial Fellow Le Ha Phuong (Nora), a freelance podcaster and former radio producer who creates current affairs content by blending expert insights with engaging storytelling.
In this piece, she explores why many of Vietnam’s fintech firms remain stuck in AI pilot mode despite growing investment and adoption, examining how data quality, organisational readiness, and operational strategy — rather than the technology itself — ultimately determine whether AI delivers real business value.
To pitch your story or share your perspectives on AI, fintech, digital transformation, or innovation across Asia, contact the NewInAsia editorial team.
Highlights
Read the Chinese article here.
Related posts:
Why Having 30,000 Teachers Proves You Can Scale Supply — But Not Learning
The AI Era Will Reward a Different Kind of Manager: One Who Redesigns Work
AI for Startups: Predicting the Future Isn’t Magic—It’s Data
The Most Radical Innovation in a Hyper-Connected Market Might Simply Be Leaving Your Product Alone
