Reward 500 Tiger-coinsSpeaker: @Kenny_Loh (SGX Academy Trainer, Wealth Advisory Director, Founder of REITsavvy.com)
Live Date: July 29, 2026 (Review Link>>)
In this livestream, Kenny Loh mapped the global AI value chain and showed exactly where SGX-listed names sit within it — from semiconductor hardware to data centre REITs — before widening the lens to how investors can structure global AI exposure across public markets, private funds, and ETFs.
Want a deeper dive? We broke this session down into 4 full recap articles, each covering a different piece of the puzzle>
Prefer to watch the highlights? Catch these key moments from the live session in short clip form>
🐯💬 Join the discussion: Share your market view or questions below. Every useful and thoughtful comment will receive Tiger Coins!
🎯 5 Key Takeaways
SGX's AI exposure sits almost entirely in infrastructure and hardware, not high-growth software, offering a more income-oriented way to play the theme versus US-listed names.
-
The "picks-and-shovels" approach favours the equipment, testing, and infrastructure suppliers behind the AI build-out (AEM, UMS Integration, Frencken, Micro-Mechanics) over flashy end-products.
-
Singapore data centre REITs ( $Keppel DC Reit(AJBU.SI)$, $DIGITAL CORE REIT MGMT PTE LTD.(DGTCF)$, $NTT DC REIT MANAGER PTE LTD.(NTDUF)$, $Mapletree Industrial Trust(MAPIF)$, $CapLand Ascendas REIT(A17U.SI)$) offer both AI-linked growth and dividends broadly in the 4%–7% range.
-
Non-US holders of US-listed AI stocks or ETFs face a 40% US estate tax on death — a material factor when structuring long-term AI exposure.
-
Kenny Loh's own approach leans on funds/ETFs over individual stock-picking, given how quickly AI winners and losers can change.
🏭 Why AI Isn't the Next Dot-Com Bubble
Every AI tool sits on top of a long infrastructure chain — telco fibre, data centres, power and water for cooling, semiconductor chips, and the equipment used to build them. Kenny Loh's view: AI isn't a dot-com repeat because that broad industrial build-out has to happen first, and the world is still in its infant stage — healthcare, finance, manufacturing, transportation and retail all have much further to go.
🔧 The Picks-and-Shovels Strategy
Rather than chasing high-profile, high-volatility AI end-products (mostly on Wall Street), Kenny Loh's framework favours the infrastructure, equipment and services layer enabling the build-out — which is where nearly all of SGX's own AI-linked names sit, offering income alongside growth.
🇸🇬 SGX's AI Hardware Names
UMS Integration (SGX: 558) and Frencken Group (SGX: E28) sit in front-end precision — mechanical modules and structural mechatronics. Micro-Mechanics (SGX: 5DD) supplies consumable tools and dies. AEM Holdings (SGX: AWX) focuses on back-end system-level test handlers. Each check needs to be run against the company's own management commentary to gauge real AI exposure.
🏢 Data Centre REITs: Singapore's Digital Real Estate Play
Five SGX-listed REITs carry AI-linked data centre exposure: Keppel DC REIT (AJBU) and Digital Core REIT (DCRU) are close to pure-play; NTT DC REIT (NTDU) is Japan-anchored and GIC-backed; Mapletree Industrial Trust (ME8U) carries over 50% data centre exposure; CapitaLand Ascendas REIT (A17U) around 8–9%. Distributions range broadly from 4% to close to 7%, though several trade in USD.
🌐 Two Markets: Public Equities vs Private Unicorns
The regulated public market already includes Nvidia, Microsoft, Apple, Alphabet and Amazon — but valuations can be rich. The private market includes pre-IPO names like Anthropic (~US$965B) and OpenAI (~US$852B), where the appeal is capturing value before an IPO re-rating, at the cost of a fragmented, harder-to-call landscape.
⚖️ Public, Pre-IPO, or ETF — Which Vehicle Fits?
Direct public equities offer the highest liquidity and lowest entry barrier but carry single-stock volatility risk. Pre-IPO funds (accredited investors only) offer potentially higher returns but lock up capital until IPO. AI ETFs spread risk across a basket of names in exchange for a more market-average return.
🧾 Don't Forget the 40% US Estate Tax
Non-US holders of US-listed AI stocks or ETFs face a 40% US estate tax on death. On a hypothetical US$1 million portfolio, that could mean an executor paying roughly US$400,000 to the IRS before a beneficiary can inherit the remainder — a real factor in structuring long-term AI exposure through US-listed vehicles.
🔍 Q&A Highlights
-
Limited capital — ETF or REIT? Dollar-cost averaging into an ETF suits a long horizon aiming for growth; data centre REITs suit those closer to retirement wanting income alongside AI exposure, though performance still correlates with tech broadly.
-
Spotting a fundamentally sound AI firm: no AI company is safe forever, even established leaders can be disrupted — Kenny Loh's checklist is growing revenue, growing profit, and positive operating cash flow.
-
Utility, storage and memory stocks: AI's power and water needs make utilities relevant, and memory names benefit from rising processing demand — but valuations can run ahead of fundamentals.
💬 Words from Kenny Loh
"AI will not be the dot-com bubble."
"Look at the foundation — the manufacturing and the infrastructure — not just the end product."
"I don't back individual AI stock picks; I'll go with the fund, or I'll go with the ETF."
🐯 Your Turn: Join the Discussion
Share your view on one of these questions:
-
Do you hold any SGX-listed stocks or REITs with AI exposure? Which ones?
-
Would you rather own the AI "picks-and-shovels" infrastructure names, or the end-product tech giants?
-
For long-term AI exposure, which vehicle fits you best — public equities, pre-IPO funds, or ETFs?
-
Does the 40% US estate tax change how you'd structure exposure to US-listed AI stocks?
🎁 Every useful, thoughtful, and well-explained comment will receive Tiger Coins.
Let's compare different views and learn from one another.
Comments
Not exciting, but steady and safe. For retirement money or funds that cannot afford to lose big, I feel a bit of investment via CPF OA is one of the easiest way to get some AI exposure without too much drama. The bet is against 2.5%. If drop, juz stack but do not over commit.
AI产业仍在快速变化，很难提前判断最终赢家。ETF可以分散单一公司的估值、技术路线和监管风险；个股则可选择已经拥有订单、现金流和基础设施优势的“卖铲子”公司。
美国遗产税风险也会影响我的配置方式。我会优先研究非美国注册的ETF或本地上市产品，在保留AI敞口的同时，降低跨境税务和集中持仓风险。
数据中心REIT、半导体设备、精密工程、电力和冷却系统，都能受益于长期算力需求，而且不需要判断最终哪一个AI模型胜出。投资工具方面，我更倾向用ETF作为核心仓位，再少量配置具备订单和现金流的个股。
美国遗产税风险也会影响我的部署。我会优先考虑非美国注册的ETF或本地上市工具，并在投资前确认税务结构，避免只关注收益、忽略跨境持有成本。