What Is Actually Happening to AI Stocks

The narrative that the "AI trade is over" is too simple. What is happening is a repricing — from unbounded optimism to demand verification. Investors who bought Nvidia at $236 in anticipation of infinite AI infrastructure spending are now asking whether the capex commitments from hyperscalers will generate the returns that justify the valuations.

Nvidia fell 7.1% in one week, briefly entering correction territory after dropping from $211.34 to $178.91, driven by a general risk-off in the market and OpenAI CEO comments on financing $1.4T spending commitments that spooked AI investors. The stock has since recovered to around $207 — but the episode illustrates the volatility profile: sharp drops on sentiment, partial recoveries on earnings data.

Investor enthusiasm for the AI boom has been waning, with AI stocks falling broadly as investors question Jensen Huang's announcements — despite new production authorisations for SK Hynix, Samsung and Micron. This is not a demand collapse. It is a valuation rationalisation. Nvidia still has strong revenue growth, strong margins, and a dominant market position in AI accelerators. What has changed is the market's willingness to pay 40x forward earnings for it.

What This Means for Singapore Stocks

The STI is a different animal from the Nasdaq. Its top holdings — DBS, OCBC, UOB, SIA, Singtel, CapitaLand — have minimal direct AI chip exposure. Singapore's banks earn their income from loans and wealth management. Singapore Airlines earns from passengers. The STI's all-time high of 5,041 in February 2026 was driven by different forces entirely: declining interest rates, strong Singapore GDP growth of 4.8% in 2025, and recovering ASEAN trade flows.

The STI reached a new high of 4,579 in December 2025, extending gains supported by US rate cuts and China's pledge of broad economic support for 2026. The STI hit an all-time high of 5,041 in February 2026, gaining 43% year-on-year. That is a materially different return profile from the AI-driven Nasdaq, and it is built on a fundamentally different set of earnings drivers — bank net interest margins, aviation recovery, REIT distributions and regional trade.

The implication for Singapore investors: you do not need US tech exposure to capture equity returns. But neither the STI nor the Nasdaq protects you from the core problem both share — mark-to-market daily volatility that can evaporate 10% of a position in a week.

Why Singapore Property Holds Up When Markets Wobble

Property is not a return maximiser. It is a volatility dampener. The case for Singapore residential property as a hedge against equity market volatility rests on four specific structural characteristics — not general "property is safe" sentiment.

• 1
  No Daily Mark-to-Market Nvidia can drop 7% in a week and your portfolio shows the loss immediately. A Singapore condo bought in 2024 does not update its valuation because of an AI stock selloff. The psychological benefit of not watching your net worth fall in real time is undervalued. For many Singapore families with their primary wealth in property, this stability is not passive — it is structural.
• 2
  Low Correlation to Tech Sector Cycles Private residential prices in Singapore continued to rise in Q1 2026 despite softer transaction activity, supported by firm take-up at selected new launches, resilient local buyer demand, and still-contained unsold inventory. The drivers of Singapore property — domestic employment, HDB upgrading cycles, school zone demand, infrastructure upgrades — are structurally disconnected from Nvidia's revenue guidance or Jensen Huang's keynote announcements.
• 3
  Inflation Protection via Land Scarcity Real estate remains the most reliable hedge against inflation because land is Singapore's most limited resource. Singapore has no ability to expand its land bank meaningfully. When input costs — construction, labour, materials — rise with inflation, replacement cost of existing buildings rises with them. Land values tend to outpace inflation over time in a supply-constrained city-state.
• 4
  TEL + NSC Infrastructure Tailwind Is Confirmed, Not Speculative Nvidia's premium over book value is a bet on future AI spending. Thomson Reserve's premium over JadeScape is a bet on infrastructure that is confirmed, completion-dated and partially operational. The CRL Bright Hill interchange in 2030 is not a revenue forecast — it is an LTA project with a published timeline. That distinction between speculative and structural growth matters when markets are repricing risk.

Where Is the Singapore Government Heading — and What Does It Mean in 3–5 Years?

Two signals from 2026 policy are relevant to the property-as-hedge argument.

Signal 1: The government is managing property prices, not suppressing them. A total of 4,575 private residential units will be released via the Confirmed List in the first half of 2026, about 50% above the average supply per GLS programme over the past decade. This is supply management — the government using land release to prevent a repeat of 2021's 10–15% annual price surge, not to deflate values. The message: moderate, sustainable growth is the policy objective.

Signal 2: Infrastructure investment is accelerating the value of specific corridors. The TEL, NSC, CRL and JRL are not stock market speculation. They are government-funded infrastructure projects with published completion dates. Every MRT station on the Thomson-East Coast Line has been built. Every NSC viaduct section has a tender date. The appreciation that follows confirmed infrastructure cannot be reversed by an AI stock selloff or a rate shock.

In 3–5 years, the picture: Nvidia will either have justified its valuation through AI revenue or continued to correct. Singapore private property on the TEL corridor will have the Bright Hill CRL interchange operational, the NSC fully open, and the first Thomson Reserve, Lentor Gardens and Dunearn House cohort approaching TOP. Those are not forecasts. They are scheduled events.

The Honest Case Against Property as a Hedge

Intellectual honesty requires the other side. Three reasons Singapore property is not a clean hedge.

Illiquidity is a double-edged sword. No daily mark-to-market means no daily loss visibility — but it also means you cannot exit in a week if your circumstances change. A Singapore condo typically takes 3–6 months to sell. During the 2008 crisis, transaction volumes collapsed and sellers who needed liquidity were forced to accept significant discounts.

ABSD makes the second property a high-hurdle investment. A 20% ABSD on a $2M property is $400,000 upfront. That is not a hedge — it is a significant committed capital outlay that needs to be recovered through appreciation before any real return is generated. For buyers who already own one property, the maths requires a clear 5–10 year horizon and specific corridor conviction to justify the entry cost.

The property-equity correlation is not zero in a severe downturn. In 2008, Singapore private home prices fell approximately 25%. In 2020, they wobbled before government support stabilised them. Property is less correlated to equities in normal volatility periods — it is not decorrelated in systemic crises where employment collapses and credit markets freeze.

The property hedge argument is strongest for buyers who: have a long horizon, have the income base to carry the mortgage without relying on returns, and are buying in a corridor with confirmed structural tailwinds. It is weakest for investors seeking short-term capital preservation from equity drawdowns.

FAQ — AI Stocks, STI and Singapore Property