Tan Su Shan, chief executive of DBS Group, warned at a Semafor World Economy Global Leaders dinner in Bangkok that the AI buildout, already hampered by shortages of chips, memory and power, could soon face shortages of capital as well.
The event took place on the sidelines of the World Bank and IMF annual meetings. Tan said lenders are beginning to demand real returns and that while companies anywhere on the AI supply chain may be broadly okay for the next roughly two years, it is uncertain what happens after that.
She directly referenced remarks by Broadcom chief executive Hock Tan, whose comparison last week of some AI investors to casino gamblers helped trigger a chip selloff and rattled markets. Tan said many investors appear to have "no clue what they are investing in," but argued the market reaction has a constructive element: unlike the sudden shocks of past crises, the current environment is provoking public discussion about ROI and capex, and buyers are increasingly seeking risk‑adjusted returns.
"They're bringing up this ROI question, the capex question. The market is speaking, and I don't think it's a bad thing," she said.
Role of capital markets and private credit
Tan noted the long reach of U.S. capital markets in attracting investments from sovereign wealth funds to retail traders, and warned that the concentration of dollars into a narrow set of growth opportunities creates a point at which investors will ask tougher questions about returns. "When the party will end, I don't know," she said.
She also observed that the private credit market is adjusting, which she viewed as potentially healthy and a path toward a new equilibrium.
Hyperscalers shifting from cash generators to cash consumers
According to Tan, many hyperscalers are tapping markets for financing before potential credit downgrades, but this reflects a broader shift: firms that were once massive cash generators are becoming heavy cash consumers as they invest in AI infrastructure, a pivot that has important implications for financing and valuation.
Geopolitical position and middle powers
Speaking about DBS's operations across Taiwan, mainland China and the rest of Asia, Tan said the current geopolitical landscape is an opportunity for middle powers. These countries and companies can remain open and transactional with multiple partners, which she argued promotes safety. She cautioned that unleashing AI without guardrails, shared best practices and cyber know‑how would leave everyone worse off.
AI adoption at DBS: advisers, agents and governance
Tan described how DBS was an early user of AI with its wealth advisers and suggested that in future interactions may occur between agents on both sides. She emphasized the importance of being deeply embedded in customer journeys with AI that is trusted, policy‑grounded and governed. That requires clear limits on what agents can access and do, continuous evaluation of their behavior, and the ability to terminate agents that fail to meet standards: "If they're not doing what we want them to do? Sayonara, kill the agent."
Overall, Tan Su Shan's message was that beyond physical supply constraints, the AI sector is entering a phase where questions about returns and financing will shape investment decisions, and markets and lenders are beginning to insist on more disciplined, risk‑adjusted approaches.
