Southeast Asian banks struggle to turn AI into revenue

In the boardrooms of Jakarta, Singapore, and Bangkok, a paradox is unfolding. Southeast Asia is arguably one of the most AI-ready regions in the world—boasting a mobile-first population, supportive regulators, and massive infrastructure investment. Yet, despite billions of dollars pouring into Artificial Intelligence, a significant gap remains between the promise of technology and the reality of the balance sheet.

A recent report by Dyna.Ai with GXS Partners and Smartkarma reveals a stark truth: while AI spending in the banking sector is accelerating, many institutions remain stuck in “pilot purgatory”—unable to translate successful experiments into sustained, revenue-generating operations.

On paper, Southeast Asia (SEA) looks like the perfect laboratory for an AI revolution.

• Hyper-Connectivity: The region is home to over 680 million people with a median age of just over 30, creating a digital-native customer base ripe for AI engagement.
• Infrastructure Boom: By mid-2024, more than US$30 billion had been committed to AI-ready data center infrastructure across Singapore, Thailand, and Malaysia.
• Regulatory Support: From Singapore’s sandboxes to Malaysia and Thailand’s frameworks, regulators are actively encouraging innovation.

However, the report notes that “higher investment alone does not guarantee business results”. While the physical foundation is being laid, the commercial transition is lagging. Global AI spending in banking is set to skyrocket to US$368 billion by 2032, but currently, only 10% of organizations using agentic AI are seeing significant, measurable ROI.

Why do pilots that dazzle in the lab fail on the front line? The research identifies three primary “commercial bottlenecks” that constrain Southeast Asian banks.

1. The Data Fragmentation Trap

AI models crave context. To offer a hyper-personalized loan or investment tip, an AI needs a 360-degree view of the customer. However, in many SEA banks, data remains trapped in silos—legacy core systems do not talk to mobile apps or credit risk engines.

Without a unified data layer, high-value models (like fraud detection or predictive credit) fail to scale because they cannot access consistent, real-time data across markets.

2. The “Trust Gap” (Adoption Lag)

This is perhaps the most human and overlooked hurdle. The report highlights a fascinating metric regarding the “adoption gap”:

“While AI models can be deployed within three months, it often takes up to nine months for frontline staff such as relationship managers to trust and actively use them in day-to-day workflows.”

If a Relationship Manager (RM) in Singapore or Kuala Lumpur does not understand why the AI is recommending a specific product, they simply will not use it. The technology is fast, but organizational trust is slow.

3. Regulatory Fragmentation

While individual domestic regulators are supportive, ASEAN is not a monolith. A credit scoring model validated in Indonesia often requires significant re-engineering to meet data sovereignty or privacy rules in Vietnam or the Philippines. This forces banks to rebuild “proven” solutions for every new market, killing the economies of scale that AI promises.

Despite these struggles, some institutions are breaking through. The report highlights DBS Singapore as a prime example, generating US$565 million in 2024 from over 350 AI use cases.

For banks looking to escape pilot purgatory, the report identifies three specific revenue engines in Southeast Asia:

• Inclusion Lending: Tapping into the estimated US$300 billion financing gap for MSMEs in the region. Using alternative data (telco, e-commerce) to lend to “thin-file” borrowers in Indonesia and the Philippines.
• RM “Co-Pilots”: Generative AI tools that reduce research time for wealth managers. One tool cited boosted advisor sales by 20% year-on-year by cutting research time by 95%.
• Personalization: Moving from segmentation to “segment-of-one” marketing. Generative AI personalization is linked to a 6% revenue uplift and a 3% improvement in ROE.

The banks that are successfully scaling AI have shifted their mindset. They are moving away from “technology delivery” and towards “outcome-based commercial models”.

To close the gap, the research suggests a playbook for SEA leaders:

1. Anchor to P&L: Stop treating AI as an experiment. Tie every pilot to a specific revenue line item (e.g., “increase MSME loan approval by 15%”).
2. Partner to Scale: 56% of bank CXOs now prefer buying off-the-shelf solutions to move fast, rather than building everything in-house. This allows them to bypass the talent shortage.
3. Solve the Trust Gap: Invest in “explainability.” If an AI tells a banker to sell a product, it must explain why in plain language. Usage only takes off when staff trust the machine.

The technology to transform Southeast Asian banking is already here, and the capital is committed. The struggle is no longer about capability—it is about connectivity. Connecting siloed data, connecting AI models to human workflows, and connecting pilots to clear financial outcomes.

As Tomas Skoumal, Chairman of Dyna.Ai, notes, the gap between pilots and revenue is “far wider than most executives expect”. The winners in this region will not be the banks with the most advanced algorithms, but those with the organizational agility to actually use them.