Data centre downturn would hit bank fees before loans
Japanese and Singaporean banks’ exposure to data centres and AI remain low, said CreditSights.
Banks in Asia—particularly Japanese and Singaporean banks—have little exposure to data centres and artificial intelligence (AI), and any risks are more likely to hit revenue than credit losses, according to CreditSights, a Fitch Solutions Company.
MUFG and Sumitomo Mitsui Financial Group’s (SMFG) data centre exposure remain under 2% of credit exposure, and SMFG is now tightening its credit exposure to the segment, according to a report released on 16 November 2026.
Mizuho’s and Nomura’s exposure to data centres appear smaller, the report said.
Singaporean banks’ exposure is higher, with data centre financing between 3% to 5% of total loans.
CreditSights noted that financial regulators and central banks are beginning to tighten scrutiny to the connections between the financial system and AI, and the emerging risks between the two. It also observed a rise in client inquiries regarding banks’ exposure to AI and data centres.
At present, AI and data centres are mainly a revenue exposure, largely within capital markets, CreditSights said, with banks’ roles largely as underwriters and as intermediaries in capital activity.
“Banks with strong origination and distribution platforms are best positioned to capture fees over several years, whilst several banks this quarter acknowledged that the pace of investment may be choppy over time,” CreditSights said, citing sentiment, political factors, interest rates and cost of capital, and energy supply and cost concerns for the latter.
CreditSights warned that underwriting can turn into balance sheet risk if deals are not distributed as quickly as expected.
A slowdown would most likely show first on slower fee growth rather than credit losses, it added.
A second and third order impact revolved around recession risks, slower loan growth, and slower consumer credit, CreditSights said.
“AI investment has been a major driver of GDP growth, so a severe slowdown creates recession risk, along with slower loan growth and weaker asset quality anywhere in the AI supply chain,” CreditSights wrote.
A more sustained downturn in the AI space could lead to a deleveraging in markets, it added.


