Pragmatic, Not Profligate: How DBS, Maybank and CIMB Are Using AI to Redefine ASEAN Banking

Southeast Asia’s biggest banks are quietly rewriting the AI rulebook. While US and European giants make headlines for multi‑billion‑dollar technology budgets, ASEAN leaders such as DBS, Maybank and CIMB are showing that targeted, pragmatically governed AI deployments can deliver outsized value without matching Western spending sprees. The contrast reveals an emerging “ASEAN model” of banking innovation: less focused on flashy experimentation, more on disciplined industrialization of AI where it moves the needle on risk, revenue and cost.

A new analysis from Tech Wire Asia highlights the scale of the divergence. JPMorgan Chase now spends about 18 billion dollars a year on technology, rolling out AI tools to roughly 250,000 employees and experimenting across more than 450 use cases. By comparison, Singapore‑based DBS, Malaysia’s Maybank and Malaysia/Singapore‑anchored CIMB have far smaller IT budgets, but their AI programs are already delivering concrete financial and operational results that rival Western peers on a relative basis.​

DBS stands out as the region’s industrialist of AI. The bank has deployed more than 1,500 AI models across roughly 370 use cases as of mid‑2025, spanning credit, fraud, collections, marketing, operations and HR. Those models generated an estimated 750 million Singapore dollars in economic value in 2024, and DBS is targeting around 1 billion Singapore dollars in 2025 after three years of sequential doubling. Crucially, the bank treats AI not as a set of pilots but as a core part of its operating system: models are integrated into existing processes, governed centrally, and measured against hard P&L metrics.​

The use‑case mix is telling. In retail banking, DBS uses AI to score loans, detect fraud patterns and personalize offers, improving both risk management and cross‑sell conversion. In corporate banking, models help with early‑warning signals on client health and with working‑capital optimization tools that strengthen relationships. On the operations side, AI supports workflow routing, call‑center assistance and anomaly detection, cutting handling times and error rates. The bank’s leadership argues this systematic “AI everywhere it makes sense” approach is more powerful than chasing headline‑grabbing experiments.​

Maybank is pursuing a different, partnership‑centric path, leveraging hyperscale cloud and productivity tools to accelerate adoption. In August 2025, it signed a five‑year, 1‑billion‑ringgit (about 240‑million‑dollar) strategic agreement with Microsoft, designating Azure as its primary cloud platform and rolling out Microsoft 365 Copilot to about 44,000 employees. Rather than building its own large language models from scratch, Maybank is focusing on securely embedding Copilot and other AI assistants into day‑to‑day tasks—drafting documents, summarizing customer interactions, generating code and surfacing insights from internal data.​

The bank frames this as an “enterprise‑wide productivity and risk‑control play.” Front‑line staff use AI to prepare proposals and answer client queries faster; risk and compliance teams use it to scan policy documents and regulatory updates; and IT teams leverage generative AI to speed up development and testing. Maybank’s leadership sees this as a cost‑effective way to level up the skills and effectiveness of tens of thousands of employees without the complexity of building heavy in‑house AI stacks.​

CIMB, meanwhile, has leaned hardest into customer‑facing generative AI, especially in small‑business lending and customer service. Its flagship GenAI‑powered chatbots now handle a significant share of customer interactions across multiple markets, achieving about 94 percent accuracy on intent recognition and resolution, according to the Tech Wire Asia report. That performance allows the bank to automate routine queries while escalating complex cases to human agents with full context, improving both efficiency and satisfaction.​

The payoff for SMEs has been particularly notable. CIMB reports a 288 percent increase in SME financing over a recent period, supported by AI‑enhanced risk models and digital journeys that shorten approval times. By combining data from transaction histories, alternative sources and sector benchmarks, its systems can evaluate small‑business borrowers more quickly and with greater nuance than traditional scorecards alone. For entrepreneurs across ASEAN’s “missing middle,” this can mean access to working capital that was previously either too slow or unavailable.​

What unites these three banks is not just the use of AI, but their disciplined approach to governance and focus. Analysts note that, unlike some Western peers, ASEAN banks have typically prioritized a smaller number of high‑impact domains—credit, fraud, productivity—before experimenting widely. Data quality, model risk management and regulatory alignment are treated as core program pillars, not afterthoughts. This is partly necessity: with tighter budgets and fewer in‑house data scientists than global giants, they must extract more value per model deployed.​

The regional regulatory mood has helped. Supervisors in Singapore, Malaysia and other ASEAN markets have encouraged experimentation under structured guidelines, issuing AI‑risk management and model‑governance frameworks that stress explainability, fairness and accountability. Banks that demonstrate robust controls can innovate faster, knowing what compliance guardrails they must stay within. Some regulators are also exploring “suptech” tools—AI used by supervisors themselves—to analyze regulatory returns and transaction data, creating a feedback loop where both banks and watchdogs become more data‑driven.​

There are challenges. Talent remains a bottleneck, especially for advanced machine‑learning engineering and AI product management. Competition from big tech and startups makes retention difficult, and banks must invest continuously in training existing staff to work effectively with AI systems. Cybersecurity and model‑risk incidents are another concern: as more processes rely on AI, the potential impact of a flawed model or compromised data set increases. Institutions are responding with red‑team exercises, ethical‑AI committees and layered defenses, but the risk surface is expanding.​

The comparison with JPMorgan and other Western majors raises strategic questions. If DBS can generate roughly 1 billion Singapore dollars of annual economic value from AI on a smaller base, and Maybank and CIMB can show material productivity and lending gains with partnership‑driven strategies, the traditional assumption that “more tech spend automatically means better outcomes” looks increasingly shaky. Instead, the emerging lesson is that capital efficiency, governance and domain focus may matter more than raw budget size.​

For ASEAN’s broader financial sector—including mid‑tier banks and fintechs—the examples set by DBS, Maybank and CIMB are likely to become reference playbooks. Smaller institutions cannot match JPMorgan’s 18‑billion‑dollar tech spend, but they can emulate the region’s champions: pick a few high‑impact domains, industrialize data pipelines and model governance, and integrate AI deeply into business processes rather than treating it as a lab project. As the World Economic Forum and others have noted, such pragmatic digital strategies are essential if ASEAN is to close financing gaps for SMEs and households while maintaining financial stability.​

If these trajectories hold, Southeast Asia’s banks may prove that in the AI era, being smarter about where and how to deploy capital beats simply being bigger. For global observers, that makes the region not just an interesting market, but a laboratory for the future of AI‑enabled banking.