The banks and investors writing the checks for the artificial intelligence spending boom are beginning to take a harder look at the risks behind the spending.

DBS Group Holdings CEO Tan Su Shan said in a Bloomberg Television interview on October 11, 2026 that the market is repricing the risks tied to artificial intelligence capital expenditures. She framed the shift as a healthy development for the sector, even if it may sting for some borrowers.

Tan's assessment carries particular weight because DBS is not watching from the sidelines. The Singapore lender has arranged approximately S$25 billion (around US$20 billion) in AI-related deals this year, giving her a front-row seat to how lenders and investors are sizing up the sector.

Financiers Demand More

At the core of Tan's message is that financiers are no longer treating every AI project as a sure thing. Investors are now pushing for higher returns on investment before committing capital, and they are paying closer attention to how transactions are structured. In practice, financing terms are increasingly being calibrated to the strength of individual projects and borrowers rather than flowing uniformly across the AI buildout.

The aim, as Tan described it, is to make sure these arrangements hold up over the long run rather than merely looking impressive at signing. She said the tighter approach is a positive one: more discipline on returns and structure, in her view, makes for a sturdier market.

Tan did flag one potential consequence. Weaker players in the AI ecosystem could face possible credit rating downgrades as investors grow pickier about where they place their money. Credit ratings are a widely used gauge of a borrower's ability to repay, and a downgrade can make future debt fundraising more difficult and more expensive. Whether weaker players do end up facing rating actions, and how borrowers adapt to the tougher terms, will be among the threads to watch as the repricing runs its course.

Where DBS Has Placed Its Bets

DBS's AI financing work this year has centered on hardware. The bank's deals have focused primarily on semiconductors, memory chips, and data centers — the infrastructure layers that anchor the broader AI spending boom.

Notably, DBS has not loaded these loans onto its own balance sheet. The bank has syndicated most of the financing, meaning it arranged the deals and then distributed much of the lending to other investors. Syndication is a common structure in large project financing: the arranging bank earns fees for putting the deal together while spreading credit risk across a wider group of lenders. The result is minimal direct exposure for DBS itself.

A Bank That Also Builds With AI

Tan took the helm at DBS in March 2025. Since then, the bank has leaned heavily into AI on the operational side, not just as a lender to the sector.

DBS runs somewhere in the range of 1,500 to 2,000 AI models internally, spread across hundreds of use cases. The bank says these systems generate approximately S$1 billion in annual economic value.\nTan has described the goal as becoming an "AI-enabled bank with a heart," a phrase meant to pair technology adoption with a commitment to its workforce and operational integrity.