Singapore’s position as a “strategic gateway” for foreign companies looking to expand their operations in Asean, as well as the China+1 strategy, is likely to serve the city-state well in the ongoing artificial intelligence (AI) revolution, say OCBC Investment Research (OIR) analysts Donovan Tan and Ada Lim.
However, the country’s lack of available land, power and water could hinder its growth in the data centre space, which is seen as the “bedrock” of AI, they add in their Sept 23 report, while noting that Singapore is the largest data centre market in the region with a capacity of 1.4 gigawatts (GW).
With this in mind, Singapore has set out a green data centre roadmap that pivots towards nurturing talent and expertise in AI instead.
“By prioritising the development of broader AI capabilities, Singapore can ensure long-term sustainability, growth and innovation in its data centre industry, even within the constraints it faces,” the analysts write.
How do investors position themselves to benefit from the AI revolution, then? Tan and Lim take a closer look at Singapore equities and the sectors that are likely to reap positive returns from the growing demand for AI. Communication services is one such sector, as are real estate, financials and industrials, they note.
Communication services
See also: Horse racing stocks fairly valued, but gamblers can take a bet on Evolution
Tan and Lim have identified Sea and Singapore Telecommunications Z74 (Singtel) as companies that stand to benefit from the growing demand for AI.
While Sea’s founder Forrest Li has “warned of a difficult transition to AI”, he has “remained reticent on the company’s plans for developing AI tools”, Tan and Lim say. Although Sea AI Lab shared its findings on the latest AI technologies, the lack of a plan to leverage AI by Sea keeps Tan and Lim cautious about the disruption that AI could bring.
In line with Singapore’s National AI Strategy (NAIS) 2.0 initiative, Singtel has collaborated with Nvidia to build accelerated computing data centres using Nvidia’s graphic processing units (GPU) with a view to introduce GPU-as-a-Service (GPUaaS), enabling enterprises to gain wider access to AI.
See also: Singtel is one of the most compelling data centre plays in region: RHB
Furthermore, Tan and Lim note that Singtel has continued to hone its data centre segment. Singtel has plans to operationalise over 200 megawatts (MW) of data centre capacity by the end of 2026, first starting in Singapore before expanding to Thailand, Malaysia and Indonesia. Singtel also signed a memorandum of understanding (MOU) in July with SK Telecom to work on creating differentiated innovative solutions, such as Edge AI infrastructure.
“As Singtel focuses on reinvigorating its core business, its up-and-coming data centre segment forms an important scalable growth engine for the business in the long run,” Tan and Lim state.
Real estate
Tan and Lim note that as the demand for AI increases, businesses require their properties to be equipped to support these advancements. According to them, Singapore REITs (S-REITs) stand to benefit from implementing AI within properties through improved tenant retention, decreased operating costs and increased bargaining power for rent.
The OIR analysts cite data centre owners and operators as the main beneficiaries of the AI revolution. Among the REITs that own or will potentially own AI-ready data centres, CapitaLand Ascendas REIT A17U , CapitaLand India Trust CY6U and Mapletree Industrial Trust ME8U are key beneficiaries, they add.
Meanwhile, Keppel DC REIT still has “overhang from rental arrears for its Guangdong data centres on its share price”, Tan and Lim add.
The analysts suggest that other REIT sub-sectors can thrive by adapting to AI. Retail REITs can adopt an omnichannel strategy that combines both physical stores and online platforms, using data analytics to personalise offerings and optimise store layouts.
For more stories about where money flows, click here for Capital Section
Meanwhile, office REITs can adapt to hybrid working arrangements by revitalising offices and considering changing underutilised spaces to co-working hubs.
Tan and Lim also note that in the travel and hospitality sector, AI has significantly impacted customer service with targeted advertising and predictive maintenance. In the logistics and industrial space, companies can improve operations by including automation, robotics and Internet of Things technologies.
The duo also notes that AIMS APAC REIT is working on upgrading and redesigning its logistics warehouses. According to the REIT’s results briefing for 1QFY2024 ended June 30, this will accommodate their tenant’s requirements, optimise the building’s potential and secure a long-term lease agreement.
Financials
Tan and Lim have identified DBS Group Holdings as a leader in the industry, having invested early and strategically in the right areas. DBS has made “notable” progress in deploying AI, with over 800 AI models across 350 use cases, with a measured economic impact exceeding $1 billion by 2025.
DBS has revamped its entire technology stack, providing a foundation that supports application programming interfaces and microservices. This ensures that DBS can deploy, maintain and upgrade applications seamlessly without affecting its core businesses, OIR’s report says.
According to the OIR analysts, this “agile approach” to deploying AI “sets DBS apart from its competitors and positions the bank as a front-runner in harnessing the power of AI technology”.
Further to the report, United Overseas Bank U11 (UOB) is the first Singapore bank to trial Microsoft 365 Copilot, a Generative AI-powered productivity tool. This tool aims to increase the productivity and efficiency of workers, improve their access to key information and enhance output by making messages more concise for stakeholders.
Industrials
According to the OIR analysts, Keppel remains one of the few companies worldwide that is able to provide end-to-end services across the data centre value chain.
“Keppel has an established track record in the development and operation of power plants,” Tan and Lim say. Keppel’s assets include Keppel Merlimau Cogen, a 1,300MW combined cycle gas turbine (CCGT) power plant, and a 600MW CCGT power plant designed to operate on fuels with 30% hydrogen content.
Furthermore, Keppel has expanded into renewable energy and has exposure to wind farms in Europe and a solar portfolio in Germany via Keppel Infrastructure Trust A7RU .
According to Tan and Lim, “Keppel, together with Facebook and Telin, is currently building the Bifrost cable system, the world’s first and largest high-speed transmission cable system spanning over 15,000km across the Pacific Ocean.” They note that the subsea cables will cater to increasing connectivity needs in Southeast Asia and enable AI in the region.
In data centre development and operations, Keppel DC REIT has experience designing and developing data centres locally and overseas. The company is developing a floating data centre park in Singapore that relies on seawater to cool its servers and operates on a green energy grid using hydrogen, Tan and Lim add.
The OIR analysts have also identified Sembcorp Industries U96 as a leading player in the renewable energy sector, both a user and an enabler of AI. They state that data centre operators have increased demand for renewable energy to reduce their environmental footprint, which would add to Sembcorp’s customer base in the region.
Tan and Lim report that data centre operator Equinix has signed a power purchase agreement with Sembcorp to power its upcoming data centres in Singapore with renewable energy.
“Singapore Technologies Engineering [ST Engineering] is a company that is actively monetising AI by incorporating it into scalable solutions that are relevant across many end-industries,” Tan and Lim say. Some tools that the company has introduced include its intelligent water management system built for Singapore’s Public Utilities Board (PUB) to collect information on Singapore’s water resources and Einstein.AI, which helps validate information authenticity.
Is AI sustainable?
Despite the positive use of AI, it is undeniable that data centres are highly resource-intensive.
An analysis by Goldman Sachs found that data centre power demand is poised to grow 160% by the end of the decade as compared to 2023, doubling carbon dioxide emissions as compared to the 2022 baseline. The International Energy Agency is even more pessimistic, expecting global electricity demand from data centres to double towards 2026.
Additionally, Tan and Lim note that a mid-sized data centre uses around 300,000 gallons of water daily to cool its servers, equivalent to the water consumption of 100,000 homes. It is estimated that water consumption from data centres has grown globally from 738 million litres in 2015 to over 840 million litres in 2021, posing a threat to water availability, add the analysts.
The impact of AI on sustainability is particularly pertinent to Singapore. While Singapore is one of the three top-tier data centre markets in Asia Pacific and enjoys strong demand from global hyperscalers, Tan and Lim also recognise that Singapore lacks power, water and land.
Data centres contributed to 7% of Singapore’s total energy consumption in 2020 and this number is expected to grow to 12% by 2030. The increase in data centre capacity leads to an increased carbon footprint, threatening Singapore’s commitment to achieve net-zero emissions by 2050, according to Tan and Lim.
Singapore is among the most water-stressed countries globally due to the lack of natural water sources and limited land area for water collection. Tan and Lim recognise that this has led to spillover investments into neighbouring Southeast Asian countries, such as Johor in Malaysia, given its proximity to Singapore and cheaper costs.
However, the analysts affirm that Singapore still stands to gain as the proposed Johor-Singapore Special Economic Zone is expected to bolster the Singapore Economic Development Board’s “SG+” initiative, which encourages multinational corporations to set headquarters in Singapore as a gateway for regional expansion. “Such collaboration allows Singapore and its neighbours to mutually benefit from the AI revolution.”
These challenges also present an opportunity to innovate, Tan and Lim add. The National University of Singapore’s College of Design and Engineering signed an MOU with Equinix in 2022 to research technologies that allow the use of hydrogen as a green fuel source for mission-critical data centre infrastructure.
