Singapore AI adoption rose 60% in a year. The next constraint is not technology.

Singapore’s digital economy grew to S$144.1 billion in 2025 and now makes up 19.3% of GDP, up from 18.8% a year earlier. It grew 5.6% against nominal GDP growth of 3.1%. The digital economy is expanding about twice as fast as the economy that contains it. That is the headline from IMDA’s Singapore Digital Economy Report, released on 5 October 2026. Behind that headline, Singapore AI adoption moved further and faster than the economy it sits in.

Among all enterprises, adoption went from 14.7% in 2024 to 23.5% in 2025 — up 59.9% in a single year. Among large local enterprises it is now 70.4%, from 62.5%.

Measured against the OECD, Singapore is well ahead rather than catching up: 34.6% of its enterprises with ten or more staff use AI, against an OECD average of 20.3% — 14.3 points above the pack. And the growth is arriving from outside the technology industry. Tech employment grew 3.8% to 222,200, but employment in non-information-and-communications sectors grew 5.3% against 1.8% inside tech. The people adopting AI are, increasingly, not technologists.

The engine and the brakes

On 26 September 2026, Singapore’s Foreign Minister, Dr Vivian Balakrishnan, took the same subject to the United Nations General Assembly, in a speech otherwise about the rule of law in a fragmenting world. His metaphor for AI is worth borrowing: “to drive a high-performance car quickly and safely, you need a powerful engine. But equally, you need good brakes.” The engine, in his telling, is investment in science, talent and infrastructure — “especially developing our people, equipping them with the skills necessary to master and deploy AI quickly, widely, at scale and safely.” The brakes are safeguards: “rigorous testing and evaluation before deployment”, “clear limits on what autonomous systems can do”, and “mechanisms to intervene when they behave in ways we do not intend.”

The sentence that matters most commercially is not about models at all. It is about management:

“The real productivity gains from AI comes not from the IT department, but from our businesses and workers having the skills, the tools, and the wherewithal to re-engineer their work processes from workflow to workflow. That is where the cutting edge of value creation arises.”

Dr Vivian Balakrishnan, UN General Assembly, 26 September 2026

He then names the binding constraint, and it is not a technology either: “The scarce resource today is not legal architecture or technology. It is trust — trust that the risk will be disclosed, that testing will be comprehensive and credible, and that cooperation will be inclusive.”

Where the numbers thin out

Go back to the report and the optimism acquires an edge. Almost nine in ten AI-adopting enterprises in Singapore are using commercial off-the-shelf tools — ChatGPT, Claude, Gemini or Microsoft Copilot. Only 28.3% are customising anything, with AI agents built on Claude Cowork or Google AI Studio given as the examples. Two in five are expanding their deployment or have integrated AI fully across their operations. The rest have not.

That is not a criticism — a spreadsheet is also off-the-shelf. But it locates the frontier precisely. Getting an AI tool into a business is now cheap, fast and unremarkable. The gain Balakrishnan describes is the next step: re-engineering a workflow around the tool, with somebody who owns the change. Most adopters have not started it.

The training numbers say the same thing from the other side. 86% of workers now use AI at work, up from 78% a year earlier, and the reported benefits are consistent: 73% cite productivity, 69% better work quality, 68% stronger problem-solving. Among those who saved time, about six in ten saved up to an hour a day. Yet 68% of workers say they need to upskill or reskill in AI, and only 37% attended any AI-related training in the past twelve months. Of those who did not, around 35% were not nominated by their employer, a similar share said they lacked the time, and about one in four did not know which course to take.

IMDA reads that as a signposting problem. It is also a management one. The hour a day is already being saved. The capability that would compound it is not being funded.

Why the direction is still up

Three numbers say the promise is real rather than promotional.

First, the returns are not marginal. SMEs that adopted AI-enabled digital solutions through the Productivity Solutions Grant reported average cost savings of 49.5% in 2025. Whatever the measurement behind that figure, it is not a rounding error.

Second, the pay signal is unambiguous. The median monthly wage for resident tech workers is S$8,000 against an overall resident median of S$5,000 — 60% higher — and the fastest-growing roles are in AI, data and cybersecurity. Demand is concentrated exactly where the report says it is.

Third, the pipeline is filling. The share of enterprises that have not adopted AI but intend to within twelve months rose from 5.6% in 2023 to 12.1% in 2025. IMDA is expanding the TechSkills Accelerator under the National AI Impact Programme to upskill 40,000 technology professionals over three years, and to train 100,000 more in domain-specific AI skills by 2029, with accountancy and law already under way.

Put those together and the picture is not that Singapore is behind. It is that the country has passed the adoption question and arrived at the harder one.

What it means if you run a business

The report suggests a sequence, and it is short. Adopt. Then redesign a workflow — not a job title. Then make certain that a named person is accountable for the outcome. That last step is the same principle Balakrishnan applies to autonomous systems at the UN, and it applies just as well to a five-person finance team:

“Most importantly, human beings must ultimately remain in control and accountable. Let me give you an extreme thought experiment. Do you want the nuclear button to be in the hands of a human being or an AI?”

Dr Vivian Balakrishnan, UN General Assembly, 26 September 2026

Singapore’s AI story is genuinely upward. The digital economy is outgrowing the economy, adoption rose almost 60% in a year, wages are following the demand, and Singapore is making its own governance frameworks and testing tools available to others while it joins the call for binding oversight of frontier models. As the Minister put it: “Rules do not stop us or slow us down. They give us assurance and confidence so that we can move faster and more effectively for the benefit of all.”

The constraint has simply moved — from whether a business can get AI, to whether it can re-engineer how it works around AI, and whether the people doing that work have been trained to. Both are cheap to start. Neither can be faked, which is exactly why they are worth doing now, while the numbers are still moving this fast.

Frequently asked questions

How much did Singapore's digital economy grow?

To S$144.1 billion in 2025, or 19.3% of GDP, up from S$136.5 billion and 18.8% in 2024. It grew 5.6 per cent year on year, against nominal GDP growth of 3.1 per cent.

What is Singapore's AI adoption rate?

23.5 per cent of all enterprises in 2025, up from 14.7 per cent in 2024 — a rise of 59.9 per cent in a single year. Large local enterprises reached 70.4 per cent, and among firms with ten or more staff Singapore sits at 34.6 per cent.

Does Singapore's AI adoption beat the OECD average?

Yes. Among enterprises with ten or more staff, Singapore's adoption rate was 34.6 per cent in 2025 against an OECD average of 20.3 per cent — 14.3 percentage points above it.

Are Singapore SMEs adopting AI?

Yes. Adoption among small and medium enterprises rose from 14.5 per cent in 2024 to 23.4 per cent in 2025, and SMEs that adopted AI-enabled digital solutions through the Productivity Solutions Grant reported average cost savings of 49.5 per cent in 2025.

What did Vivian Balakrishnan say about AI at the UN?

In Singapore's national statement to the 81st UN General Assembly on 26 September 2026 he described AI as arguably the most urgent frontier, argued that a powerful engine still needs good brakes, and said the scarce resource today is not legal architecture or technology but trust.

What did Balakrishnan say about where AI productivity comes from?

That the real productivity gains from AI come “not from the IT department, but from our businesses and workers having the skills, the tools, and the wherewithal to re-engineer their work processes from workflow to workflow”.

What is the gap in AI training in Singapore?

68 per cent of workers say they need to upskill or reskill in AI, but only 37 per cent attended AI-related training in the past 12 months. Of those who did not, about 35 per cent were not nominated by their employer, a similar share said they lacked the time, and about one in four did not know which course to take.