In his final weeks as chief executive of DBS, Piyush Gupta had delivered the best year in the bank’s history. Record profit of S$11.4 billion, a share price at all-time highs, a S$1,000 bonus for nearly every employee.
Then, in February 2025, on a stage in Mumbai, the man who had run one of Southeast Asia’s largest banks for 15 years explained that he would let 4,000 contract roles lapse. He also made an admission out loud: “For the first time, I’m struggling to create jobs.”
A month later, the DBS baton passed to Tan Su Shan, the bank’s incoming chief executive. The plan held: the workforce reduction was a structural decision that survived the change of chief executive, because the work the departing roles did had already gone to software the bank was still building.
That admission is the story of AI in Asia Pacific banking, and 18 months on it looks more like a template. In May 2026, Standard Chartered told investors it would cut more than 15% of its back-office roles, around 7,800 jobs, by 2030 as it accelerates its use of AI, with the reductions falling on corporate functions across its major operations hubs in India, China, Malaysia and Poland.
Two months later, JPMorgan Chase chief executive Jamie Dimon told analysts that AI had already eliminated 30% to 40% of headcount in some of the bank’s units. He also warned that the technology wouldn’t translate into fatter margins. “You don’t uniquely benefit from AI,” he said.
How Asia-Pacific’s banks are rewiring headcount
DBS aimed at its most expendable workforce first. The 4,000 reduction falls on contract and temporary staff across its 19 markets, delivered through natural attrition as projects roll off, with permanent employees untouched and roughly 1,000 new AI roles created.
ANZ took a more direct route. The bank expects around 3,500 employees and 1,000 contractors cut by September 2026, funded by an A$560 million restructuring charge and pointed at an A$800 million cost-savings target.
Customer-facing roles are largely spared. More than 60% of affected staff had already exited by the end of December 2025. The execution has been rough: some staff learned of their redundancies by automated email before anyone told them in person. Nuno Matos, ANZ’s chief executive, called the episode indefensible.
Meanwhile, Bloomberg reported in March 2026 that HSBC is weighing cuts of up to 20,000 roles over three to five years, concentrated in non-client-facing positions in its global service centres, a workforce heavily located in Asia.
No final decision has been made, and the bank declined to comment. But the direction fits everything chief executive Georges Elhedery has already disclosed: $1.8 billion committed to delivering the restructuring, 3,000 of the bank’s 9,000-plus applications marked for retirement by 2028, and a $1.5 billion cost-savings target reached in the first half of 2026, six months early.
Banking the savings before the AI works
Bloomberg Intelligence’s January 2025 survey of 93 bank technology chiefs projected up to 200,000 job cuts across global banking within three to five years and estimated AI could lift pretax profits 12% to 17% by 2027, worth up to $180 billion.
Citi’s research put 54% of banking roles at high automation potential, the largest share of any industry it examined. Every one of those numbers is a forecast. The cuts, by contrast, are actual: booked, charged and executed.
DBS itself supplies the clearest evidence of the gap. The bank has said it remains cautious about customer-facing AI because of hallucination risk. The jobs are being paid out on a promise the software hasn’t yet kept.
Why would rational management sequence it that way? Because the savings arrive either way. Cutting contractors and service-centre roles flatters the cost-to-income ratio whether or not the AI performs, and ANZ has already pointed to a lower ratio as proof its overhaul is working.
If the productivity forecasts land, the banks look prescient. If they don’t, the banks have still banked the cost reduction. Heads the strategy wins, tails the ratio does.
The uneven geography of banking’s AI pivot
IDC projects AI and generative AI investment across Asia Pacific will grow from $73 billion in 2024 to $370 billion by 2029, with banking among the largest contributing sectors.
The budget that funded a reconciliation team in Manila or Bangalore is being redirected to the vendors, cloud platforms and model providers replacing the reconciliation. For fintech and infrastructure firms selling into the region’s banks, every restructuring headline is a demand signal.
The pain moves too, and unevenly. The service-centre roles most exposed sit in Asian hubs, so the employment costs of boardroom decisions fall on the region’s operations workforce.
Only one jurisdiction is treating that as a policy problem. In October 2025, the Monetary Authority of Singapore, the Institute of Banking and Finance and Workforce Singapore published a Generative AI Jobs Transformation Map for the financial sector, with 11 institutions, including DBS, HSBC, OCBC, Standard Chartered and UOB, piloting workforce transition programmes under it. A regulator doesn’t build reskilling infrastructure for a change it expects to be marginal.
Tomasz Noetzel, the Bloomberg Intelligence analyst behind the 200,000 estimate, has argued AI will transform roles rather than remove them outright: “Any jobs involving routine, repetitive tasks are at risk.”
The distinction between transformation and removal is doing heavy work in every bank’s messaging. Based on the numbers disclosed so far, roughly 27,500 roles across DBS, ANZ and the reported HSBC assessment sit on the removal side of that line, and every one of them was cut before the technology replacing them went through its due processes.
Regardless of what AI has automated in Asia Pacific banking, it hasn’t yet touched the oldest rule in the industry: that the people who make the bet and the people who are the bet get paid out on very different sche
