Singapore’s early bet on artificial intelligence could give its economy a modest but meaningful lift as population ageing starts to bite. But the real test will not be how many companies say they use AI. It will be whether they can rebuild work around it.
The ASEAN+3 Macroeconomic Research Office (AMRO) said in its annual consultation report on Singapore that AI adoption could raise the city-state’s potential economic output by 1.06 per cent in the long run. That would offer a partial buffer against slower labour-force growth as Singapore’s population ages, although the transition may also create pressure for some workers, especially fresh graduates entering white-collar jobs.
The estimate is not a forecast. AMRO described it as a calibrated counterfactual focused on domestic supply-side effects. It does not include possible upside from external demand for Singapore’s AI-linked trade and investment, nor broader gains from innovation or the creation of entirely new types of work.
Also Read: iWOW’s US$11M placement tests investor appetite for Singapore’s ageing economy
Even so, the finding matters. Singapore’s long-term growth model has always relied on moving up the productivity curve, given its limited land, small domestic market and tight labour supply. AI is now being folded into that familiar national problem: how to make a small workforce produce more value.
Under AMRO’s baseline scenario, only about half of the potential-output gain from AI will be realised by 2035. The boost is expected to rise to 0.65 per cent by 2040 and 0.78 per cent by 2050. The drag from ageing is expected to become more pronounced from 2030, making productivity improvements more important to sustaining economic momentum.
Adoption is rising, but depth remains shallow
Singapore has moved earlier than many of its neighbours in treating AI as an economy-wide priority rather than a narrow technology-sector issue. The government has pushed national AI strategies, industry programmes and skills initiatives, while major cloud and chip players have continued to expand regional operations from the city-state.
Business adoption is now accelerating, but it remains uneven.
According to the Infocomm Media Development Authority, AI adoption among small and medium-sized enterprises(SMEs) more than tripled to 14.5 per cent in 2024 from 4.2 per cent in 2023. Among non-SMEs, adoption rose to 62.5 per cent from 44 per cent over the same period.
The gap becomes starker when company size is considered. The Ministry of Manpower’s (MOM) 2026 establishment survey found AI adoption at 23.9 per cent among firms with fewer than 25 employees, compared with 76.4 per cent among firms with more than 500 employees.
That divide is familiar across Southeast Asia. Larger companies are more likely to have the budgets, data infrastructure and technical teams needed to experiment with AI, while smaller firms often struggle with implementation costs, data readiness, governance concerns and a shortage of in-house expertise.
But AMRO’s report suggests the larger issue is not initial adoption. It is integration.
Only 3.8 per cent of firms surveyed by MOM had integrated AI into their core operations. Most were still planning, testing or using basic tools. Among firms already using AI, off-the-shelf generative AI tools accounted for 84 per cent of adoption, while fewer than half had implemented customised or proprietary AI solutions.
Also Read: SIA has scaled AI. Aviation must now govern the point of action
That distinction matters. A company using a chatbot to draft emails is not the same as one redesigning customer service, compliance, logistics, finance or product development around AI-enabled workflows. The first may save time at the margins. The second could change productivity.
White-collar work faces the biggest adjustment
AMRO also warned that Singapore’s labour market has a relatively high share of jobs exposed to AI compared with other high-income economies.
It estimated that occupations with a high degree of automation exposure account for 22.3 per cent of employment. These are roles where AI could substitute for a substantial share of existing tasks and potentially reduce labour demand. The exposure is concentrated in professional jobs such as financial analysis, business administration, sales, marketing and public relations.
That does not mean a fifth of jobs will disappear. Many roles are bundles of tasks, and AI may automate routine activities while leaving humans to handle judgement, client relationships, decision-making, compliance and complex problem-solving.
Business administration professionals, for instance, may use AI for data processing or preliminary analysis while continuing to interpret findings, manage stakeholders and ensure regulatory requirements are met. In this sense, AI may change the route into expertise rather than eliminate the need for it.
Firm-level evidence so far points more towards restructuring than mass displacement. Among firms adopting AI, 18.9 per cent reported job redesign and 13.9 per cent reported creating AI-related jobs. Only 6.2 per cent reported reducing headcount.
This pattern will be closely watched across Southeast Asia, where governments are trying to raise digital productivity without worsening job insecurity. Singapore may be better placed than most economies in the region because of its training system, fiscal capacity and concentration of higher-value services. But it is also more exposed, because many of its workers are in precisely the professional roles that generative AI can affect first.
Fresh graduates may feel the pressure first
One of the sharper warnings in AMRO’s report concerns younger workers.
Entry-level professional jobs often involve routine research, coordination, documentation and analysis. These tasks have historically helped graduates learn the basics before moving into higher-value work. If AI absorbs more of that junior work, employers may need fewer fresh hires, or expect them to contribute at a higher level from day one.
AI could also widen competition for some service-sector roles. By reducing language barriers and making remote delivery easier, it may allow companies to source certain tasks from outside Singapore more easily.
Also Read: Malaysia wants 300,000 AI jobs by 2030. Talent will decide if it gets there
AMRO was careful not to claim that AI is already driving graduate employment weakness. It noted that the proportion of tertiary graduates in the labour force who were employed declined in 2025, while youth unemployment stayed at 6.6 per cent in both 2024 and 2025. The share of young people not in education, employment or training improved to 3.8 per cent in 2025 from 4.1 per cent in 2024.
Graduate hiring also slowed across advanced economies during the same period, particularly in technology and professional services. Still, the report’s warning is clear: the transition costs may be concentrated among those with the least workplace experience.
From AI access to AI productivity
AMRO identified three main channels through which AI could raise potential output: higher total factor productivity as tasks are automated, additional physical capital investment as output rises, and improved effective labour input if AI helps less experienced workers learn faster.
Most of the expected long-term gain comes from productivity and capital, which together contribute 0.98 percentage points to potential output. The human-capital learning channel adds 0.08 percentage points under the baseline scenario.
The overall gain could range from 0.47 per cent to 2.10 per cent, depending on productivity assumptions. Speed also matters. If diffusion is completed by 2030, AMRO estimates a long-run gain of 1.13 per cent. Under the baseline 2035 timeline, the gain is 1.06 per cent. Under a slower 2040 timeline, it falls slightly to 1.01 per cent.
For Singapore, the policy implication is straightforward but difficult to execute. Support cannot stop at encouraging firms to “adopt AI”. It must help them apply the technology to real operational bottlenecks.
That means sector-specific use cases, implementation help for SMEs, better data systems, cybersecurity, cloud access, computing capacity and managerial capability. It also means tracking outcomes such as productivity gains, deeper integration and worker transitions, rather than counting AI users.
Also Read: Seoul tops global AI implementation as Singapore enters top tier
Existing programmes such as SkillsFuture AI courses, the SkillsFuture Workforce Development Grant and the National AI Impact Programme could help. But AMRO said training and job-redesign support should focus more closely on novice and entry-level workers in roles where AI is likely to complement human tasks.
Singapore’s AI challenge is therefore less about ambition than execution. The country has the strategy, infrastructure and policy machinery to move early. The harder question is whether enough firms, especially smaller ones, can turn experimentation into everyday productivity before demographic pressures intensify.
The post Singapore’s AI dividend will depend on what happens after the pilot phase appeared first on e27.