Singapore's AI Boom Warning Is About Concentration, Not Hype

AI-related investment accounted for most of US investment growth and about half of US GDP growth this year, according to Chia Der Jiun, managing director of the Monetary Authority of Singapore. That scale helps explain why markets are excited about AI. It also explains his warning: growth can become fragile when too much depends on one investment cycle.

This is not a claim that AI has no economic value. Chia’s point is narrower and more useful. The race to build AI is increasingly tied to the same handful of assumptions: that leading firms can keep funding enormous build-outs, that chips and power remain available, and that future productivity gains will justify the cost.

For Southeast Asia, that is a signal to watch. The speech does not predict a regional downturn or quantify ASEAN exposure. But it raises a practical question for governments, investors and businesses across the region: are local AI gains becoming durable capabilities, or are they mainly an indirect bet on a global spending boom?

In remarks delivered in Singapore on 25 May, Chia described the AI boom as a major force behind global growth and market optimism. He said AI-related shares had driven much of this year’s market gains, while Taiwan and Korea had seen strong export growth linked to the semiconductor boom.

The scale of the spending matters. Chia said hyperscalers had announced data-centre investments worth several hundred billion dollars for the year, with strong projections continuing into the years ahead.

But the same race is creating pressure points. He said companies are competing for scarce chips and energy, which is raising the unit cost of compute—the processing capacity used to train and run AI systems. The build-out is supported by relatively easy financing and the large cash flows of major cloud companies.

That combination can work for a long time. It can also change quickly if funding becomes more expensive, if expected returns disappoint, or if a technological or regulatory shift changes the economics of the race.

Chia’s central concern is not simply a possible slowdown in AI spending. It is concentration.

He warned that global growth could be high but unbalanced if it is driven by AI, a small group of companies and a narrow set of sectors. The benefits may also be uneven if productivity gains are not widely shared across countries, workers and employers.

Put plainly: an AI boom can lift headline growth while leaving the economy more dependent on a few big companies, specialised suppliers and expensive infrastructure. That does not make the technology unimportant. It makes the distribution of risk more important.

A BIS working paper published in July offers a useful analytical lens, but it should not be treated as a forecast for Singapore or Southeast Asia.

The paper models AI investment as a winner-take-most contest: firms spend heavily because getting ahead could secure a dominant share of future revenue. In that kind of race, each firm can have an incentive to commit more capital than is best for the economy as a whole.

Under a conservative model baseline, the authors estimate over-investment at around 50% above the socially efficient level. They also find that debt, circular equity stakes and specialised AI hardware can make a downturn harder to absorb. If one company fails, financial links can spread stress to others.

Those are model-based findings, not a prediction that an AI bust is imminent. Their value is in clarifying why concentration matters: when investment, financing and expected returns all become tightly linked, a disappointment can travel further than it would in a more diverse economy.

Singapore is a financial centre and has been building AI capability in its financial sector. Its warning therefore carries regional relevance even though the speech does not claim a specific ASEAN outcome.

Southeast Asia benefits when AI investment expands trade, skills, infrastructure and business adoption. The open question is whether those benefits will spread widely enough to last if global compute costs rise or the pace of investment slows.

That is why the most useful measure is not the size of any one announcement. It is whether the region is building capabilities that remain useful beyond a single global capital-spending cycle: skilled workers, reliable power and connectivity, practical AI adoption, and businesses that can use the technology productively.

Three signals are worth watching:

None of these signals alone predicts an outcome. Together, they show whether the AI boom is becoming broader and more resilient—or more dependent on the same narrow bet.

Singapore’s warning is not anti-AI. It is a reminder that a powerful technology cycle can still create economic risk when growth, finance and infrastructure all lean on the same assumptions.

For Southeast Asia, the better question is not whether to participate in the AI build-out. It is whether participation is creating lasting regional capability before the next change in funding, supply or expectations arrives.