MAS says 32% of Singapore-listed companies could become financially vulnerable under a severe AI-investment downturn. But the more important number is 16% — and it changes what investors should worry about.
Singapore’s artificial-intelligence boom has created an obvious investment narrative.
AI is driving demand for semiconductors. Data centres need more electricity. Semiconductor equipment orders are rising. Infrastructure spending is expanding.
For investors, that has created a natural question:
Which Singapore-listed companies stand to benefit?
The Monetary Authority of Singapore’s latest Financial Stability Review, however, suggests that investors should ask another question first:
Which companies can survive if the AI investment cycle slows?
MAS’s stress test found that 32% of Singapore-listed companies would be classified as “at risk” under a severe downturn in AI-related investment.
That sounds alarming.
But another number is arguably more important.
Those companies represented only 16% of overall corporate debt.
That distinction tells investors something important about where the vulnerability lies — and where it does not.
The 32% headline needs context
MAS’s exercise was a severe stress test, not a forecast that an AI crash is coming.
The scenario applied revenue shocks of up to 30% and differentiated interest-rate shocks of up to 400 basis points, designed to capture wider credit spreads under stress.
A company was considered at risk if its interest coverage ratio fell below one, or if it had negative cash flow and insufficient cash to cover six months of the shortfall.
Even under that severe scenario, MAS said most firms were able to weather the shock, supported by their earnings and cash reserves.
That distinction matters.
“At risk” in the stress test does not mean a company would default, become insolvent or fail.
It means the company’s financial buffers would become strained under the assumptions of the scenario.
Investors should therefore resist the tempting but inaccurate interpretation that:
32% of Singapore stocks could fail if AI spending falls.
MAS did not say that.
What it did was expose where the corporate sector’s financial sensitivity is concentrated.
The number that matters: 32% of companies, 16% of debt
The fact that the 32% of at-risk firms account for only 16% of corporate debt is revealing.
The stress-test vulnerability is disproportionately located among companies that are smaller or financially more sensitive.
MAS says highly leveraged, capital-intensive and working-capital-dependent firms are more exposed. Smaller firms are also disproportionately represented, reflecting their generally thinner margins and cash buffers.
That makes the story less about an imminent Singapore-wide corporate credit crisis and more about dispersion among individual companies.
For equity investors, that is a much more useful distinction.
Two companies can benefit from the same AI investment cycle but have radically different downside characteristics.
One might have:
- substantial cash reserves;
- modest debt;
- diversified customers;
- strong free cash flow.
Another might have:
- significant borrowing;
- heavy working-capital requirements;
- thin margins;
- concentrated customers;
- and a valuation that already assumes years of rapid growth.
They may both be called “AI beneficiaries”.
They are not the same investment.
MAS isn’t saying the AI boom is unhealthy
This is another point that gets lost in the headline.
MAS continues to recognise that AI investment is supporting economic activity and corporate earnings.
The concern is not that AI investment has no economic value.
The concern is what happens if the returns on that investment fail to justify the enormous amount of capital being deployed.
That is a much more subtle risk.
And it matters because AI infrastructure is increasingly capital intensive.
The hidden issue is the cost of financing AI
MAS says the AI infrastructure buildout has become an important driver of global demand for capital.
Hyperscalers’ capital expenditure has outpaced internally generated cash flow, increasing reliance on bond markets, private credit and special-purpose vehicles to finance data-centre and semiconductor investment.
At the same time, record bond issuance by hyperscalers is competing with sovereign borrowers for institutional investors’ capacity to absorb long-duration debt.
That creates a less obvious transmission mechanism.
The AI story is usually presented as:
more AI demand → more capex → more orders → higher revenue.
But the financial system sees another chain:
more AI capex → more financing → greater sensitivity to interest rates and credit spreads → higher required returns on AI investment.
And that changes the economics of the marginal AI project.
The most important question may not be whether AI demand falls
This is where the MAS report becomes particularly relevant to equity investors.
MAS says higher interest rates, rising semiconductor and electricity costs, and greater reliance on market financing have raised the hurdle rate for AI investment.
At the same time, current equity valuations require sustained revenue growth and substantial eventual profitability from large investments in data centres and advanced semiconductors.
If earnings or expected returns fall materially short, MAS warns that AI-related valuations could be reassessed, with losses potentially spreading across public equities, corporate bonds and private credit.
That creates an important distinction:
AI demand can remain strong while AI investments disappoint.
A data centre can remain busy.
A semiconductor supplier can continue growing revenue.
A technology company can continue increasing AI spending.
And yet shareholders can still lose money if the returns generated by that spending fall below what investors had priced in.
That is the difference between AI growth and AI economics.
The feedback loop investors should watch
The more worrying scenario is therefore not necessarily an AI collapse.
It is a deterioration in the economics of the AI investment cycle.
Consider the potential chain:
AI investment slows
↓
Semiconductor and data-centre orders weaken
↓
Supplier revenue and margins come under pressure
↓
Cash flow weakens
↓
Highly leveraged companies face greater financing pressure
↓
Investors demand higher risk premiums
↓
Equity valuations fall
↓
The cost of capital rises further
↓
Marginal AI projects become less attractive
This is an analytical scenario, not an MAS forecast.
But it explains why leverage and liquidity matter so much when analysing AI beneficiaries.
A company with a strong balance sheet can potentially absorb a temporary slowdown.
A company that needs continuous growth to service debt, fund working capital and justify its valuation has much less room for disappointment.
Why this is not yet a Singapore credit crisis
There is an important counterargument.
Singapore’s corporate sector currently has substantial buffers.
MAS says corporate debt-servicing capacity improved over the past year as borrowing costs declined and earnings remained stable.
Corporate debt stood at 117% of GDP in Q1 2026, while the median interest coverage ratio of SGX-listed firms was 4.2x in Q2 2026.
Refinancing risks also remained manageable. Short-term debt represented 32% of total corporate debt, below the 10-year average, while bonds due by the end of 2027 represented 19% of outstanding corporate bonds as of August 2026.
Bank credit quality was similarly strong.
The corporate NPL ratio fell to 1.2% in Q2 2026, an 18-year low. MAS also said the banking sector continued to benefit from strong capital and liquidity buffers.
So the evidence does not support a conclusion that Singapore is heading into an AI-driven credit crisis.
The more defensible conclusion is:
Singapore’s corporate sector is resilient in aggregate, but that resilience is uneven across companies.
What about DBS, OCBC and UOB?
The MAS report does not turn the AI story into a straightforward banking crisis either.
In fact, MAS says its stress testing affirms that Singapore banks have sufficient capital buffers to withstand an adverse scenario involving a downturn in the AI-led global growth cycle.
For bank investors, the more interesting question is therefore not:
“Will an AI crash destroy Singapore’s banks?”
It is:
“Which parts of the corporate credit book would become more vulnerable if AI-related investment retrenched?”
That is a different question — and one that deserves company-level analysis.
The investment lesson: AI exposure is not enough
This is perhaps the most important takeaway from the MAS exercise.
Investors should stop thinking about AI beneficiaries as one homogeneous group.
Instead, consider five variables:
AI exposure × operating leverage × financial leverage × liquidity × valuation
AI exposure
How dependent is the company on AI-related spending?
Operating leverage
How sharply would profits change if revenue growth slowed?
Financial leverage
How much debt does the company carry?
Liquidity
How much cash and free cash flow does it have to absorb a downturn?
Valuation
How much future AI growth is already reflected in the share price?
This framework produces a much more useful investment question than simply asking:
“Is this an AI stock?”
What investors should watch next
For Singapore’s AI beneficiaries, investors should monitor:
- order intake and backlog;
- customer concentration;
- gross margins;
- free cash flow;
- working-capital requirements;
- capital expenditure;
- debt and refinancing requirements;
- interest coverage;
- and whether earnings growth is keeping pace with valuation expectations.
The key question is not simply whether revenue continues rising.
It is whether the economic return generated by each additional dollar of AI-related investment remains attractive.
Bottom line
MAS has not predicted an AI crash.
It has done something more useful for investors: it has stress-tested the assumptions underlying Singapore’s AI-linked corporate boom.
The result is reassuring at the system level.
Most companies can withstand the severe scenario. The 32% of firms classified as at risk represent only 16% of corporate debt, while Singapore’s corporate and banking sectors retain substantial financial buffers.
But the report also exposes an important vulnerability.
AI beneficiaries are not equally protected against an AI slowdown.
The companies most dependent on continued investment, most exposed to leverage and working-capital requirements, or most reliant on high future growth to justify their valuations could face much greater downside than the headline AI narrative suggests.
And that leads to the more practical investment question:
Which Singapore-listed AI beneficiaries have enough financial resilience to benefit from the boom without becoming dangerously dependent on it?
That is where the stock-level analysis begins.