This article first appeared in Forum, The Edge Malaysia Weekly on November 10, 2025 - November 16, 2025

The big push to build capacity to cater for artificial intelligence (AI) applications has morphed into a bubble of technology stocks and related infrastructure, data centres included.

From the US to South Korea, companies in the AI supply chain are seeing their shares trade at historic highs. The valuations are equivalent to levels last seen in 2000, when the dotcom bubble burst.

For instance, the Korea Exchange is the best performing securities exchange in the world this year on account of the euphoric share price rise of SK Hynix and Samsung Electronics. These companies manufacture memory chips for AI applications and supply directly to Nvidia Corp, which manufactures the high-powered graphics processing units that add value to AI-driven data centres.

The stock exchanges of Hong Kong and Taiwan have also benefited from the AI boom. Companies like e-commerce champion Alibaba Group Holding Ltd, smartphone manufacturer Xiaomi Corp — both from China — and Taiwan Semiconductor Manufacturing Co Ltd (TSMC) have anchored the bullish runs in their respective bourses this year.

The hype around AI-related stocks and infrastructure is due to the large spending by technology companies to build infrastructure. Google, Meta, Amazon and Microsoft are spending billions to procure servers, processors and high bandwidth memory chips to build data centres. The four tech giants are spending US$700 million (RM2.9 billion) on data centres this year alone, and the forecast is that data centre spending will reach US$3 trillion by 2029.

The data centres cater for a wide variety of uses. However, in the AI era, they are seen as essential for AI start-ups to power and train their large language model (LLM) applications.

The AI start-ups, which include the likes of OpenAI and Anthropic, get their money from investors who believe their applications will be the foundation for the next technology revolution. Up to the third quarter of this year, venture capitalists have poured almost US$200 billion into AI start-ups, which is more than 50% of the total money invested by the VCs.

However, there has not been a successful model for profitability so far. Billions are being invested in AI start-ups but none have given any returns to investors. Even OpenAI, the owner of ChatGPT, is burning through billions every year.

Which brings us to the question of whether AI start-ups will continue to draw new money from investors. And consequently, will the demand for data centres continue to grow?

Although the Malaysian stock exchange has remained rangebound at the 1,600-point level, the country has benefited from riding the AI wave. This is on account of the country being the leading hub for data centres in the region for two years in a row.

According to a research report, Malaysia will be a leading data centre hub in Asia-Pacific in the next few years. On this score, the government has approved 143 data centres to be built in Malaysia since 2021.

But how many will actually be built eventually, considering that the valuations of AI start-ups are said to be in a bubble now? Will the start-ups continue to draw money from investors?

Even now, the infrastructure built to cater for data centres is more than the actual demand. The latest disclosure in parliament reveals that as at end-June, the uptake of electricity by data centres is only 47% of the declared demand of 1,276mw.

Deputy Minister of Investment, Trade and Industry Liew Chin Tong told parliament last week that there had been a review of the energy demand requirements for data centres with the view to divert resources elsewhere and ensure that Tenaga Nasional Bhd (KL:TENAGA) does not end up with “stranded assets”.

Unlike other stock markets in Asia, the AI bubble has not had an impact on the Malaysian stock market as a whole.

Some technology companies saw their share prices recover this year. But it is mainly because the stocks were deeply oversold due to the tariff issues in the first half of this year. The closest to a direct AI play is YTL Power International Bhd (KL:YTLPOWR), which has built a data centre powered by Nvidia chips and whose share price is currently off its peak.

Construction giants Gamuda Bhd (KL:GAMUDA) and IJM Corp Bhd (KL:IJM) have won contracts to build data centres. But the data centre job wins merely add on to their already huge order books and have not had much impact on their share prices.

One of the few winners of the data centre play is MN Holdings Bhd (KL:MNHLDG), which specialises in building the electricity infrastructure, such as substations, for data centres. Its share price is at an all-time high, with investors anticipating more job wins related to data centres.

In the US, even the most bullish of tech investors admit that the AI play is in a bubble. They feel that the billions being poured into AI start-ups are not sustainable.

Jeff Bezos of Amazon.com Inc, however, describes the hype over AI as a “good bubble” because it leaves behind infrastructure that allows for continued development of the technology. Those who echo Bezos’ optimism point to the sprouting of e-commerce platforms following the dotcom bust in 2000 that revolutionised the business-to-business and business-to-consumer sectors by allowing products to be delivered at a fraction of the prevailing cost.

The emergence of Facebook and Google disrupted the advertising and media industries by giving companies easier access to their target customers. However, the naysayers point out that the tech revolution following the dotcom bust came with early profitability for those who still remained in the space. Within a few years, the profits generated from e-commerce activities were more than the cash burn.

The same is not happening with the AI hype. They feel that too much money is being poured into building expensive platforms in the hope that the technology will eventually be used by the general public.

So far, AI applications have not cut down the operating costs of companies and have not been able to replace people. Even in call centres, where AI is used to perform repetitive responses, the human touch is still required. The most frequent users of AI are software programmers.

The fear is that the AI platforms will eventually turn into other technology platforms such as blockchain, virtual reality and the metaverse, which has not seen overwhelming demand.

When there are no profitable applications or a super app that draws demand, the question of sustainability comes into play. In this respect, if the hype around AI fades, the need for data centres will come into question.

M Shanmugam (m.shanmugam@bizedge.com) is a contributing editor at The Edge

Save by subscribing to us for your print and/or digital copy.

P/S: The Edge is also available on Apple's App Store and Android's Google Play.