Green Link posts eightfold income growth on disciplined SME focus and AI investment

Green Link says digital banks fail by mimicking traditional models as agentic AI offers a way to serve SMEs at lower cost.

Asia’s digital banks are struggling to reach profitability as high operating costs, customer acquisition spending and weak trust make scale expensive without disciplined lending.

Patrick Sze, Chief Technology Officer at Green Link Digital Bank, said digital banks still carry the full cost base of regulated banks, including core systems, cybersecurity, compliance, risk management, regulatory reporting, data governance, fraud monitoring and customer operations.

“Digital banks do not fail because they are digital,” Sze said. “They fail when they behave like smaller traditional banks without the trust and scale of traditional banks.”

Green Link Digital Bank reported eightfold income growth and became Singapore’s first profitable digital bank, a result Sze linked to its focus on business banking and micro, small and medium-sized enterprises. He said digital banks need to serve specific underserved segments rather than compete with traditional banks on the same products and customers.

Profitability also depends on avoiding growth that adds cost without value. Sze said banks risk confusing customer expansion with value creation, especially when investing in AI, stablecoins and tokenisation without clear business cases.

“If growth is not supported by profitability and clear ROI, then we are just accumulating more cost and more complexity,” he said.

Green Link has joined the MAS Pathfinder AI programme as a core member and has built on-premise AI capabilities, including local GPU power and locally running large language models. Sze said customer and transaction data must remain protected, with AI systems secured, auditable and compliant.

He said agentic AI could have the largest impact because it can plan, reason, execute workflows and escalate exceptions to humans. For MSME loans, AI agents could collect documents, review transaction history, identify missing information and prepare credit recommendations.

For digital banks, the test is whether they can use AI to serve low-margin, high-volume SME customers at lower cost without scaling faster than their controls and business cases allow.