Singapore’s recent stock-market sell-off may look like a banking story. But beneath the decline in DBS, OCBC and UOB is a much bigger global debate: how much are investors willing to pay for growth when bond yields are rising and the AI boom is demanding enormous amounts of capital?
Singapore’s stock market has spent much of 2026 rewarding investors who owned the right themes.
Banks rallied.
AI-related technology surged.
Data-centre demand strengthened.
Infrastructure and utilities gained attention.
The Straits Times Index climbed from around 4,650 points at the beginning of the year to a record 5,828.5 on Sept 4.
Then the mood changed.
DBS, OCBC and UOB suffered heavy losses on Oct 7 and Oct 8, dragging the STI sharply lower.
At first glance, this looks like a banking-sector correction.
But the timing suggests something bigger may be happening.
Across global markets, investors are confronting a combination of higher bond yields, higher financing costs, elevated equity valuations and enormous AI-related capital expenditure.
That creates a much more important question for Singapore investors:
Is the market simply rotating away from expensive bank stocks — or is 2026 turning into a broader valuation reset?
AI is not necessarily the problem
The easiest interpretation of the current market turbulence would be to say:
“The AI bubble is bursting.”
That may be too simplistic.
The underlying AI investment cycle remains powerful.
Hyperscalers continue to build data centres.
Demand for AI chips, networking equipment, memory, power and cooling infrastructure remains strong.
S&P Global recently noted that AI investment is spreading far beyond chipmakers into data centres, power, industrial equipment, connectivity and cloud computing.
That is particularly relevant for Singapore.
The country is positioned at several points in this infrastructure chain.
It has data centres.
It has telecommunications infrastructure.
It has industrial companies.
It has utilities.
It has engineering and infrastructure businesses.
And it has REITs with exposure to digital infrastructure.
So an AI investment cycle can create opportunities for Singapore companies even if they are not building AI models themselves.
The problem is not necessarily AI.
The problem is the price investors are paying for AI-related growth.
The real issue is the cost of capital
This is where bond yields become critical.
The global bond market has been under significant pressure.
On Oct 8, Reuters reported that the US 10-year Treasury yield remained close to a 24-year high as investors worried about inflation, oil prices and government debt.
Higher bond yields change the mathematics of equity investing.
When a government bond offers a much higher return than before, investors can demand a higher expected return from stocks.
That means valuations can fall even when corporate earnings remain healthy.
This is particularly damaging to companies whose valuations depend heavily on earnings expected many years into the future.
In simple terms:
Higher yields → higher discount rates → lower present value of future earnings → lower acceptable P/E multiples.
That is why a company can report good results and still see its share price fall.
The market is not necessarily saying:
“The company is bad.”
It may simply be saying:
“The price was too high.”
And AI itself may be contributing to higher yields
This is the really interesting part.
AI is not merely another technology trend.
It is becoming one of the largest capital-spending cycles in modern corporate history.
Hyperscalers and technology companies are spending enormous sums on chips, servers, data centres, electricity and networking.
And some are increasingly turning to debt markets to fund that expansion.
Reuters reported on Oct 8 that companies including Broadcom, SpaceX and Oracle were seeking very large amounts of financing for AI-related investment, adding pressure to corporate credit markets.
That creates an unusual feedback loop.
AI optimism → more investment → more borrowing → higher demand for capital → potentially higher financing costs → pressure on valuations.
In other words, the AI boom can eventually become a victim of its own success.
The more capital the industry requires, the more important the cost of that capital becomes.
This is why Singapore banks matter
At first this may seem unrelated to DBS, OCBC and UOB.
It isn’t.
Banks are extremely sensitive to interest rates, funding costs, loan demand and credit conditions.
But there is another connection.
Singapore banks have enjoyed a remarkable period of earnings strength.
DBS generated S$6.01 billion of net profit in the first half of 2026, while OCBC generated S$4.19 billion. UOB also reported strong first-half results.
Investors therefore started paying higher multiples for those earnings.
The recent sell-off is partly a question of whether that earnings strength can continue at the same rate.
That is why Citi’s OCBC downgrade was so damaging.
The argument was not that OCBC was suddenly a weak bank.
The argument was that expectations had become too demanding.
CNA reported that Citi expected OCBC’s third-quarter earnings to be flat year on year and downgraded the stock to “sell”, with a S$27.50 target price.
That is a classic valuation-reset dynamic.
The same thing is happening globally
The Singapore bank sell-off therefore fits into a broader pattern.
US banks have also come under pressure as Treasury yields climbed.
Reuters reported that the KBW Bank Index had fallen 13% from its August peak as investors focused on the impact of higher yields on lending, funding costs and capital-market activity.
At the same time, global equity markets have become increasingly concentrated around AI-related winners.
That concentration is important.
If investors continue believing in AI earnings growth, capital flows towards AI leaders.
Everything else can struggle.
A recent Wall Street Journal analysis found that almost 80% of S&P 500 stocks declined in September even though the Nasdaq-100 gained 3%, illustrating how narrow the market leadership had become.
That creates a strange market environment:
The headline index can look healthy while a large part of the market is already correcting.
Singapore has experienced a similar phenomenon.
The STI remained extremely strong through September, but much of that strength was concentrated in its largest constituents — particularly the banks.
Singapore’s AI opportunity is broader than technology
This is where Singapore investors need to be careful.
If the AI investment cycle continues, it would be a mistake to assume that only semiconductor companies will benefit.
AI requires an enormous physical infrastructure ecosystem.
It requires:
- electricity;
- data centres;
- cooling;
- networking;
- fibre;
- engineering;
- construction;
- industrial equipment;
- cybersecurity;
- telecommunications;
- and increasingly, specialised real estate.
This creates several potential Singapore beneficiaries.
Sembcorp: AI becomes an energy story
One of the most interesting examples is Sembcorp Industries.
The company has increasingly positioned itself around energy and infrastructure.
That matters because AI data centres require enormous quantities of electricity.
Sembcorp itself has described the AI race as increasingly becoming an energy race, citing projections that electricity consumption from AI facilities could reach roughly 1,000 TWh annually by 2030, compared with around 415 TWh today.
This creates a fascinating investment chain:
AI models → data centres → electricity demand → power generation and infrastructure.
Investors therefore do not necessarily need to buy an AI chip company to participate in the AI boom.
They can potentially own the companies supplying the infrastructure that makes AI possible.
Keppel: another way to play AI without buying AI
Keppel is another example.
Its transformation towards infrastructure, asset management and data centres gives investors exposure to the physical infrastructure required by digitalisation.
That could become increasingly valuable if AI-related capital expenditure remains strong.
But there is an important caveat.
Investors should not automatically assume:
AI growth = Keppel stock goes up.
The company still has to execute projects, recycle capital effectively and generate attractive returns on investment.
The stock price also matters.
A good company can become a poor investment if investors pay too much for its future growth.
That is precisely the lesson being delivered by the current bank correction.
Data-centre REITs: the ultimate AI income trade?
Singapore’s data-centre REITs offer another fascinating angle.
Business Times reported in August that Singapore’s pure-play data-centre REITs were showing resilient operating performance, supported by cloud computing and AI demand.
On paper, the investment thesis is straightforward.
AI requires computing capacity.
Computing capacity requires data centres.
Data centres require physical real estate.
Therefore:
AI growth → data-centre demand → rental demand → potential REIT income growth.
But there is a major complication.
REITs are highly sensitive to interest rates.
So the same AI boom that increases demand for data centres can coexist with higher financing costs that put pressure on REIT valuations.
That creates an unusually interesting investment trade-off.
AI is bullish for the asset.
Higher bond yields can be bearish for the security.
Investors need to decide which force will dominate.
This is where the bank-to-AI rotation becomes interesting
Suppose the following happens.
DBS, OCBC and UOB continue to de-rate because investors believe their earnings growth has peaked.
Where does the money go?
One possibility is that it simply leaves Singapore.
But another is that investors rotate into companies with different growth drivers.
That could mean:
Banks → industrials
Banks → utilities
Banks → infrastructure
Banks → technology
Banks → data-centre assets
This would be a very different market from the one Singapore investors have become accustomed to.
And there are early signs of this rotation.
SGX reported that institutional investors were net sellers overall in September but directed their largest sector-level inflows towards utilities and technology. Sembcorp Industries attracted approximately S$106 million of net institutional inflows, while AEM attracted about S$81 million.
That does not prove a permanent rotation.
But it is an important signal.
The valuation question becomes more important than the AI story
The biggest mistake investors can make at this stage is to ask:
“Is AI still growing?”
Of course it is.
The better question is:
“How much AI growth is already reflected in the share price?”
That is the question investors are now asking across global markets.
And it is increasingly relevant to Singapore.
Take two hypothetical companies.
Company A has strong AI exposure, but investors already price it at 40 times earnings.
Company B has indirect AI exposure through power infrastructure, data centres or industrial equipment, but trades at 15 times earnings.
If AI investment continues strongly, both companies can benefit.
But Company B may offer the better risk/reward.
This is why the next stage of the AI investment cycle may be less about finding the next Nvidia and more about finding the companies supplying the infrastructure behind AI.
Singapore could actually benefit from this shift
There is an interesting strategic advantage here.
Singapore does not need to become the world’s largest AI-model developer to benefit from AI.
It can benefit from being a place where AI infrastructure is financed, powered, connected and housed.
That plays into existing strengths:
finance + infrastructure + power + data centres + connectivity + industrial engineering.
This is one reason the country’s AI investment story is broader than its technology sector.
And it creates an important distinction between Singapore and markets dominated by pure-play AI technology companies.
Singapore may have fewer companies capable of producing explosive AI software growth.
But it has companies positioned to provide the physical infrastructure required by the AI economy.
But higher rates can still ruin the party
There is no guarantee that the AI infrastructure trade will outperform.
Higher interest rates remain the biggest risk.
DBS itself highlighted “AI capex, surging bond yields and Fed unpredictability” as major market themes for the fourth quarter of 2026, while arguing that investors should look beyond technology for AI opportunities.
That is a useful framework.
If yields stabilise, infrastructure and data-centre companies could benefit from continued AI investment.
If yields continue climbing sharply, valuation compression could overwhelm the underlying earnings growth.
In other words:
Good AI economics do not automatically produce good stock-market returns.
The price paid still matters.
What should Singapore investors watch?
There are five indicators worth watching over the next few months.
1. US Treasury yields
This is arguably the most important macro variable.
If long-term yields stabilise, pressure on equity valuations could ease.
If they continue climbing, expensive growth stocks and rate-sensitive assets may remain under pressure.
2. AI capital expenditure
Investors should monitor whether hyperscalers continue increasing AI infrastructure spending.
If capex keeps accelerating, the infrastructure opportunity remains powerful.
But investors should also ask how those investments are being financed.
3. Singapore bank margins
For DBS, OCBC and UOB, the key question is whether net interest margins can stabilise.
If margins deteriorate while non-interest income normalises, bank earnings could face additional pressure.
4. Data-centre demand
For Keppel and data-centre REITs, occupancy, rental reversions, development pipelines and power availability will be critical.
The AI story ultimately has to translate into actual demand for physical infrastructure.
5. Market breadth
This may be the most revealing indicator of all.
If banks continue falling while industrials, utilities and technology stocks rise, Singapore is experiencing a rotation.
If banks fall and almost everything else falls with them, the market is experiencing a risk-off event.
Those are two very different situations.
The biggest misconception about the current sell-off
Investors should resist two extreme conclusions.
The first is:
“Banks are falling, so Singapore’s bull market is over.”
The second is:
“AI is still booming, so just buy anything related to AI.”
Both are too simplistic.
The market may instead be entering a third phase.
Phase one: bank-led rally
Strong earnings, dividends and capital returns drive DBS, OCBC and UOB higher.
Phase two: valuation reset
Investors begin questioning how much further earnings can grow and how much they should pay for them.
Phase three: broader stock selection
Capital begins moving towards companies with different earnings drivers.
That third phase could ultimately be much healthier for Singapore’s market.
The next Singapore bull market may look very different
The previous Singapore market rally was heavily associated with banks.
The next one could be much more diversified.
Imagine a market where:
DBS provides banking and wealth-management growth.
Sembcorp benefits from rising power demand.
Keppel benefits from infrastructure and data centres.
ST Engineering benefits from aerospace and defence.
Singtel benefits from digital infrastructure and asset monetisation.
Selected REITs benefit from lower rates and stronger property demand.
Technology and semiconductor companies benefit directly from AI investment.
That would be a much broader investment ecosystem.
And importantly, it would reduce the dependence of the STI on three banks.
The bottom line
The current Singapore stock-market sell-off should not be viewed simply as an OCBC story or even a DBS-OCBC-UOB story.
It is increasingly part of a much bigger global debate.
How much should investors pay for growth when the cost of capital is rising?
AI remains one of the most powerful structural investment themes in the world.
But AI is also becoming one of the biggest consumers of capital.
That means the next stage of the AI investment cycle could be very different from the first.
Investors may increasingly distinguish between:
AI companies
and
companies that enable AI.
For Singapore, that distinction could be particularly valuable.
The country has relatively few pure-play global AI giants.
But it has companies exposed to the infrastructure that AI requires — power, data centres, telecommunications, engineering, industrial equipment and real estate.
That could create a new group of Singapore investment winners.
But only if valuations remain sensible.
And that is ultimately the lesson from the DBS, OCBC and UOB sell-off.
The market can be completely right about the long-term growth story — and still be wrong about the price.
For Singapore investors, that may be the most important investment lesson of 2026.
The question is no longer simply whether AI will reshape the economy.
It is:
Which Singapore companies can capture the AI boom without requiring investors to pay an AI-bubble price?