AI Bubble?

Picture generated by Meta AI

By now, one must have read a lot of financial news, social media posts and videos, and of course blogposts about the inflating bubble in the artificial intelligence (AI) sector and its related industries. Not missing out on the bandwagon, I decided to write a short post on my thoughts.

Some analysts and fund managers were warning of a potential AI bubble coming like the dot-com bubble that happened at the turn of the century, and recent news reports are fuelling this viewpoint. To name a couple, a recent MIT study had found 95% of enterprise AI projects failed to turn a profit¹, and there were the reported interconnected business deals between AI-related companies (e.g. the Nvidia-Oracle-Open AI circular deal) not dissimilar to those made in the late 1990s.

Should the bubble explode, the notion is that not just AI stocks are going to get hit the hardest, but there would be a ripple effect across the entire equities market as seen in 2000 and 2008. Every sector and industry would be hit at least for that moment as market participants are chasing for liquidity at the first instance (which I had explained here before), so expect a counter that has nothing to do with AI would go down temporarily.

It is beyond our pay grade (and divinity grade) to know whether the whole thing is going to burst, but here are some of my takes going forward and how best to prepare for the eventuality occurring.

#1: AI Is Not New

There were at least two investors whom I had spoken to have the impression that AI started in November 2022, which was around the launch date of Chat GPT and thus sparking off the entire revolution. However, AI already had its roots back in the 1950s and slowly evolved over the decades. The advances of processing power and large amounts of data and information available provided the growth catalysts needed, especially in the domain of generative AI via large language models (LLMs) that most layman people know about.

#2: Sectoral Sensing

Using November 2022 as the cut-off time, we shall look at a proxy counter that could be considered a part of the AI-boom: Keppel DC REIT (KDC). After listing in December 2014, its first steep rise was not during 2022, but a bit further back in the second half of 2019, reaching its twin peaks in 2020 to 2021. This was the COVID-19 and its immediate aftermath period, when working from home was seen as the way-to-go and the importance of platforms supporting it, thus the need for more server space.

After the launch of Chat GPT and the sensing of everything AI has heightened, KDC managed to go up gradually, likely due to the tampering effects of interest rate hikes, now at price levels not reached since late 2021, so in a way the AI boom is probably not as strong as the upshot brought about by COVID-19.

While it may sound prejudiced to use just one counter for comparison, the message is that data centres had shown its relevance in our already digitally connected world. AI just served to add on a huge demand for them.

#3: Cusp Of A New Era?

One of the main reasons for the dot-com bust was the insatiable appetite of everything and anything internet, even on companies that had no revenue or products to show for. In other words, what the investors did were ploughing into potentials, hopes and dreams, or worse, castles in the air.

This parallel was echoed by the people who were convinced of the bubble, and the 95% MIT report did not help in alleviating the thought. With the “all-things-internet” and a slew of accompanying features like social media, e-commerce and “everything” platforms (e.g., WeChat, Grab, etc.), one could say that the 2000 crash brought about the beginning of a new period, and investors back then were “too early” in their calls.

Coming back to the present, though there are not many useful use cases for AI, we are, by my reckoning, still in the tail end of the early adoption stage. Sure, there were news of companies firing staff due to being replaced by AI, only to recall them back when their AI initiatives failed (likely the 95%), but learning, like the AI technology, is evolving, and companies and people would adapt them holistically, sooner or later. So, we may be seeing the cusp of a new era unfolding.

#4: Stay Diversified

Coming from me, this is a must-have in my blog posts about bubbles and crashes. In the very short run, everything seems to be going down in a crash, but eventually some would rise faster than others due to their distance from the epicentre of the bubble explosion, i.e., correlation.

Some had advocated going into different sectors and industries for diversification, but I would prefer to go on a higher degree and that is asset classes, because their correlation is more varied due to the different behaviour in different market conditions. While there is undoubtedly damage felt in the portfolio from the burst if it happens, it would probably be lighter as different asset classes react differently to the situation.

Disclosure

The Bedokian is vested in Keppel DC REIT, Nvidia and Grab.

Disclaimer

 

1 – Estrada, Sheryl. MIT report: 95% of generative AI pilots at companies are failing. Fortune. 18 Aug 2025. https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/ (accessed 11 Oct 2025)

We’ll Get There Fast And Then We’ll Take It Slow

This blog title was taken from a line in the 1988 song Kokomo by The Beach Boys that described two lovers taking a trip to a Caribbean island.

Picture generated by Meta AI

Though it sounds a bit oxymoronic as in going there fast and then reduce the pace, but from a honeymoon trip’s point of view, it is an incentive to hurriedly go to the vacation spot and just chillax upon reaching the destination.

On the economic front, this phrase was used to describe the United States Federal Reserve’s strategy in combating high inflation back in 2022. The “get there fast” referred to the rapid interest rate hike to bring inflation under control, while the “take it slow” was the gradual lowering of interest rates as inflation began to cool down.

From a portfolio management perspective, this could mean “rushing” one’s portfolio to a set value before “calming” it down. In other words, the portfolio would be heavily laden with high-growth/high-yielding/high-risk securities, and once the portfolio hits the mark, it would be pivoted towards more prudent income generating/safe-haven assets/securities.

However, it is not necessary to adopt two extreme approaches to do the “get there fast, take it slow”, i.e., accelerating full speed and then stepping on the brakes. Both ways could be done in a deliberate and measured manner, considering one’s comfort level on the portfolio mix and risk appetite and tolerance. 

A good example would be our pivot (which I had described in detail here) that we took six to nine months shifting from a balanced Bedokian Portfolio (35% equities, 35% REITs, 20% bonds, 5% commodities and 5% cash) to a slightly aggressive make-up (15% bonds, 5% commodities, 5% cash, and with the remaining 75% equities/REITs in a free-float from 40/60 to 60/40). It was a slight shift as we still wanted to have the important element of diversification, balancing between a slower growth with some degree of capital preservation. For the slowing down, we had planned to pivot with a less aggressive portfolio depending on the prevailing market and economic situation, with the transitioning taking probably over a period of nine months to a year.

If one thinks about this whole fast-slow scenario, it is akin to the principles of age-based portfolios, with an aggressive structure for young investors, a balanced mix during middle-age, and finally a conservative one at conventional retirement age. The main difference is just that an objective amount is the catalyst for the change in asset class proportion, rather than be dictated by one’s lifetime stages.

 

Disclaimer

The Trading Portfolio Handbook eBook (and it’s free by the way)

In my first publication The Bedokian Portfolio, I had a chapter on the Portfolio Multiverse, where an individual would plan, manage and organise multiple portfolios based on each portfolio’s objective(s) and characteristics, and the individual’s risk appetite, risk tolerance, knowledge and allowable time for the portfolios, asset classes and/or financial instruments used in it. The idea of a trading portfolio was briefly introduced in the Portfolio Multiverse. 

While there were suggestions on including a section on trading in The Bedokian Portfolio, I would prefer it served as a basic and broad introduction on investing for beginners, and wanted to remain as it is. Thus, resulting in this spinoff companion publication. 

The Trading Portfolio Handbook: Companion Guide to The Bedokian Portfolio is not an instruction on the strategies and methods of trading derivative instruments like futures and options, nor it is a comprehensive manual to look at charts, drawing lines and recognising patterns. Rather, as the title contains, this handbook acts as a guide from my perspective. Trading is seen as frequent transacting of counters over the course of seconds to years, but I also categorise some sophisticated and speculative financial instruments such as derivatives and cryptocurrencies for inclusion in the trading portfolio.

This (very) short eBook guide is available for download, either scan the QR code on the right or click on the "Download the free eBooks and paper" tab above and look for the download link. It is free by the way.

Enjoy the guide and hope you would find some useful tips and information in your trading journey, if you decide to embark on it.

Low Risk, Low Returns. High Risk, High Returns. Period.

It is known that low risk assets and securities brings about lower returns, while high risk assets and securities would generally yield higher returns. This is the tenet of the risk-return trade-off known in investment circles.

Picture generated by Meta AI

If someone comes up and tells you that there is a low risk, high returns financial product, you would really need to take a step back and think thrice about it. 

Fraud aside, there are such things exist legitimately, but it does not exist permanently.

A good example of a legit low-risk-high-returns financial instrument is our own almost risk-free treasury bills (T-Bills). When short term interest rates began to rise back in mid-2022, for three years, we had enjoyed high yields from 6-month and 1-year T-Bills, whose average buying rates reached an all-time high of 4.36% and 4.24% respectively in Dec 2022. However, in Jul and Aug 2025, the numbers came back down to below 1.8% for both, the last time of which were that low was back in May 2022¹.

Thus for low-risk, high-returns, it is a simple case of “enjoy while it still lasts”.

As for the other end of the spectrum, i.e., high risk, low returns, any sane investor should be avoiding this scenario at all costs.

 

1 – SGS Prices and Yields – Benchmark Issues. Monetary Authority of Singapore. https://eservices.mas.gov.sg/statistics/fdanet/BenchmarkPricesAndYields.aspx (accessed 28 Sep 2025)

More On The REITs Front: UI Boustead And Keppel DC

Proposed UI Boustead REIT

Boustead Singapore has announced the proposed listing of the UI Boustead REIT (UIB REIT) on the Singapore Exchange (SGX). The REIT will focus on investing in logistics, industrial, high-specifications industrial, and business space assets in the Asia Pacific region, with an initial emphasis on Singapore and Japan. The initial public offering (IPO) portfolio is expected to include 23 properties, comprising 21 leasehold properties in Singapore and two freehold properties in Japan, with a total gross floor area (GFA) of 5.9 million square feet and a net lettable area (NLA) of 5.3 million square feet. The total agreed property value of the IPO portfolio is estimated to be S$1.9 billion.

Boustead will hold up to 16.9% of the IPO units post-listing and is expected to recognize a gain on disposal of approximately S$52.6 million. The REIT will be sponsored by UIB Holdings Limited, in which Boustead has a 20% interest. The listing is subject to regulatory approvals and market conditions.

Keppel DC REIT Preferential Offering

Keppel DC REIT is set to raise approximately S$404.5 million through a preferential offering to fund its acquisition of the freehold Tokyo Data Centre 3 for 82.1 billion yen (S$707 million). The preferential offering will allow entitled unitholders to subscribe to 80 new units for every 1,000 existing units held, at an issue price of S$2.24 per unit, which is a 5.2% discount to the volume-weighted average price of S$2.3622 on September 22, 2025.

The proceeds from this placement will not only fund the Tokyo acquisition but also support an asset enhancement initiative at Keppel DC Singapore 8, cover associated costs for a 30-year land lease extension for Keppel DC Singapore 1, and assist in debt repayment. The acquisition of Tokyo Data Centre 3 is expected to be completed by the end of 2025.

The preferential offering is slated to open on 3 Oct 2025 and close on 13 Oct 2025, with the new units listed on 22 Oct 2025.

 

The Bedokian’s Take

It is noteworthy that 100% of Tokyo Data Centre 3 is contracted to a leading global hyperscaler for 15 years with annual rent escalation (Amazon? Microsoft? Google Cloud?). The rise of cloud services and use of artificial intelligence had made data centres good proxies to invest in the former fields. With an accretive distribution-per-unit of +2.8%, and a still healthy gearing of 34.5% post-acquisition, we would at least be buying up our entitlement.

As for UIB REIT, I would wait for the issue of the prospectus before deciding.

Disclosure

The Bedokian is vested in Boustead Singapore and Keppel DC REIT.

Disclaimer

References

UI Boustead REIT

https://boustead.sg/sites/boustead.sg/files/2025-09/2025-09-18_Proposed-Divestment-of-Stakes-In-Certain-%20Singapore-Logistics-and-Industrial-Assets-And-Other-Potential-Transactions-In-Connection-With-The-Proposed-Listing-of-UI-Boustead-REIT-On-The-SGX-ST_1.pdf

 

Keppel DC REIT

https://www.keppeldcreit.com/en/file/investor-relations/presentations/2025/kdcr-acquisition-of-tokyo-dc-3-and-preferential-offering-presentation.pdf

https://www.businesstimes.com.sg/companies-markets/reits-property/keppel-dc-reit-raising-s404-million-unitholders-help-fund-82-billion-yen-tokyo-data-centre-buy

 

Two IPOs Coming Up

Two initial public offerings (IPOs) got the investing town talking nowadays: one is a bond exchange traded fund (ETF) and the other a real estate investment trust (REIT), namely the LionGlobal Short Duration Bond Fund (Active ETF SGD Class) (LionGlobal Bond ETF) and the Centurion Accommodation REIT (Centurion REIT). 

I will briefly go through these two IPOs and will give my very short take.

Screenshots of LionGlobal Short Duration Bond Fund brochure cover and the 2024 Centurion Corporation annual report cover

LionGlobal Bond ETF

The LionGlobal Bond ETF is an active bond ETF and is the listed version of an existing unit trust fund with the same name. It is an actively managed ETF that has a wide exposure of short-term duration bonds across various countries (primarily Singapore with about 40% of its net asset value or NAV) and sectors (76.5% NAV from financial and real estate). The weighted yield to maturity, duration and credit rating stood at 3.18%, 2.25 years and A-, respectively.

The main selling points of the ETF is the advent of falling interest rates, to which short duration bonds are less correlated to, and the stability of the Singapore Dollar (SGD) from which it would be hedged with against non-SGD bonds in the ETF. 

The LionGlobal Bond ETF IPO application period is from now till 23 Sep 2025 and will be listed on 29 Sep 2025, with an issue price of SGD1.00 per unit.

Centurion REIT

In the first of its kind in Singapore, the upcoming Centurion REIT is focused on worker and student accommodation (hence the word in their official listing name), with 14 properties across three countries (Singapore – 5 worker dormitories, United Kingdom – 8 student dormitories and Australia – 2 student dormitories), and a possible one more student dormitory coming up in Australia. The sponsor, listed company Centurion Corporation, is a known player in the dormitory business.

Yield wise, the projected distribution for 2026 is 7.47% and 2027 is 8.11%. The gearing ratio is around 20.9% at IPO, increasing to 31% after the acquisition of the 15th property mentioned in the previous section. Though 262 million plus units are issued in the IPO, only 13.2 million units are available for public offering in Singapore.

Centurion REIT is now open for IPO application till 23 Sep 2025 and will be listed on 25 Sep 2025, with the issue price of SGD0.88 per unit.

The Bedokian’s Take

I would view the LionGlobal Bond ETF as a corporate bond since the bond issuers are mostly in the commercial side. For our Bedokian Portfolio, it sits somewhere in between the cash portion and bond component, since the maturity (2.25 years) is too short for a corporate bond (five years minimum based on guideline) and too long as a cash-equivalent (one year maximum). The 0.25% management fee for an actively managed ETF is reasonable relative to the Amova SGD Investment Grade Corporate Bond ETF (Amova IG Corporate Bond ETF) which is at 0.26%, and not actively managed.

The LionGlobal Bond unit trust version had provided a 3.7% per annum performance since its inception in 1991, which is impressive given its bond status. While as an ETF, the risks are somewhat distributed, the big one in my opinion would be a systemic risk affecting the entirety of the financial and real estate sectors, since 76.5% of the bonds are from them.

As for the Centurion REIT, with accommodation as its primary driver, the income deriving from workers and students would have to depend on which they are supporting, that is the construction/manufacturing sectors, and education sector respectively. According to the prospectus, the increasing demand for workers is there with Singapore’s ongoing development and expansion of infrastructure and industrial capabilities. For students, particularly international ones, the United Kingdom and Australia remained as one of the top three study destinations for them. 

Both IPOs do have their unique propositions in terms of the potential; low correlation of short-term bonds to interest rates, and a REIT on dormitories. It would be a go for me though I find the Centurion REIT might be harder to get due to the small Singapore public offering numbers vis a vis the total number of units offered.

All figures are from the respective IPO and fund documentation unless otherwise stated.

Disclosure

The Bedokian is vested in the Amova IG Corporate Bond ETF.

Disclaimer

A Sign Of Market High?

A few weeks ago, I was taking a private hire vehicle home when the driver struck a conversation with me, which (somehow) drifted to the topic of investing and trading, and I was asked this question, “What do you think of (insert asset/security name)?”.

Picture generated by Meta AI

He claimed that he knew nuts about the workings of the market, and he picked this asset/security up from a recent passenger who did day trading on it, and apparently became quite successful. While I enjoy discussions like these, but taking into his investing/trading background (which I gave him the benefit of doubt of his stated zero knowledge), I treaded carefully the dialogue and advised him to read up more, since the asset/security he brought up is volatile.

For those who are aware, and/or had seen a couple of Hong Kong TV series on the stock markets, whenever anyone who has had no investing/trading experience suddenly engage with you on the markets, it is a sign of exuberance and hubris. The reason for this occurrence is simple: when people make money from the markets, the beneficiaries tend to talk about it, and if there are a lot of them around, the network effect gets bigger. Adding into the effects generated by news and social media, the outreach becomes greater to the point that those who were initially not interested would become so.

So, does this imply that the bubble is about to burst? My “think fast” instincts might tell me it is going to happen soon, and I needed to do some extreme hedging actions on my portfolio (e.g., sell all and run for the hills). My “think slow” brain, however, may be telling me to sit back and analyse on the asset/security in question, e.g., whether its bull case is justified. A balance is needed between the two, but I admit it is difficult to keep it this way.

Our remedy is simple: if we are vested in this asset/security and has future potential, we would just hold or rebalance when it goes up, or buy more when it is down. Following diversification and its related factors like correlation, this holding would not be the only thing in our portfolio, but only form part of it.

Therefore, whenever someone brings up a market tip in the car, just listen, evaluate and enjoy the ride.