2025 Investor Day: Singtel mentioned that the asset monetisation target is much larger than S$9bn, the scaling up of its data centre as Singapore capacity doubles, finalised its GPU-as-a-service business model, and the need to adopt AI at scale to secure the competitive edge in cost and service levels.
- Read this at SGinvestors.io -
Key Highlights from Singtel Investor Day 2025
CEO: AI is to use it or lose it to competitors
There is no change in FY26 outlook, such as S$9bn monetisation, S$2bn share buyback. Adopting AI at scale to raise productivity by 20-30%. Not working on AI initiatives exposes Singtel to greater risk.
- Read this at SGinvestors.io -
Optus is sharpening its B2B enterprise offering. The MOCN partnership with TPG will reduce costs and generate revenue from TPG as the 5G network is rolled out.
In Singapore, Singtel must participate in the lower prices. The new 5G plus tiering of the network is for premium customers who want a differentiated experience.
CFO:
The capital intensity of the group is declining and Singtel’s operations have no first order impact from tariffs;
Cancelling the S$2bn shares will boost earnings per share (~1%);
Need to grow underlying earnings to offset the absence of the 3 to 6 cents value realisation dividends (VRD) p.a. planned for medium term;
Value realisation dividends (VRD) of S$1.4bn (or total 8.5 cents) has been paid over past 2 years. See Singtel's dividends. Paying 6 cents (or S$1bn) for the next 5 years is plausible as associates improve dividends by S$500mil and the pipeline of assets to be monetised is larger than S$9bn (S$4bn already completed). Other monetizable assets include more real estate, reducing stake in subsidiaries, disposing of infrastructure assets and equalisation 6.5-7% stake in Bharti is worth S$12- 13bn. There is no intent to exit associates.
Singtel's subsidiaries:
Digital InfraCo: Scaling and earnings growth round the corner
RE:AI will launch GPU as a service (GPUaaS) and will cater to Singapore sovereign and other governmental agency needs. It will be the first few to operate Nvidia GB200 and GB300 GPUs. These GPUs require a power of 200 kWh per rack vs 8-10 kWh per rack for conventional servers. Such GPUs need liquid cooling and strong operating systems. For instance, conventional data centres' terminal event is 22 minutes reaction for a 10 kWh rack, but 200 kWh is only 2 minutes, or the expensive GPU will burn out. Therefore, it requires predictive machine learning capabilities.
The business model of GPUaaS is to contract for 3 to 5 years. The GPUs are largely underwritten by customers or redeployed into other countries at the end of the contract. Target returns are in the mid-teens and above IRR. Nvidia’s cloud partner (NCP) is to avoid excessive reliance on the four major hyperscalers. The AI alliance is also to partner with other data centres around the world to resell GPUaaS.
Singapore has limited power. The next CFA for data centres in Singapore is likely by this year's end, with completion in 2028-30. The likely allocation is 200MW rather than 300MW. The upcoming 58MW data centre in Tuas will commence in January 2026. It is 50% sold and will double Singapore's capacity to 120MW. Half of DC Tuas will be liquid-cooled GPUs with a connection to five subsea cables. Existing capacity is full, and revenue growth is only from the reservation fee or price escalation. The new 26MW Thailand data centre is 80% sold, with the 38MW due in 2Q27 already one-third sold.
Nxera will focus on the needs of Singapore customers. Johor and Batam are special zones and connected to Singapore, where certain workloads can be passed on. Entering the Tokyo market for GPUaaS because demand is strong.
Singtel Singapore: Star Wars Episode IV – A New Hope
Simba took on S$1.1bn to acquire M1. There will be pressure for mobile price repair too. Mobile price repair occurs if less bandwidth is offered at the cheap S$8 plans. Hundreds of gigabit plans are given, but most customers' average bandwidth usage is only 15GB per month.
Roaming is another area of competition where more roaming data is being bundled into mobile plans. It does make roaming more affordable, and using Singtel’s network offers security as it taps into key partners. With its scale, Singtel can negotiate more competitively with partners on roaming.
Using the 700MHz to improve coverage and offering 5G plus plans to consumers in May for priority access, 4x faster speed and packet inspection security. Will work with gaming companies on 5G plus plans. Network slicing is also being used for enterprise customers such as PSA, Changi and hospitals.
Cost savings will come from modernising the network and the use of AI. Other areas of savings will be shifting to new data centres and on-premise cloud. Singtel Singapore still has the globally leading EBITDA margins of 39%. Consumer and enterprise margins are similar, with capex intensity for enterprise being lower. More than 80% enterprise revenue is real telco connectivity.
Optus: Rising profitability post reorganisation
Optus has undergone a major revamp of senior management over the past 12 months. The new CEO arrived in November 2024, and thereafter, a new marketing officer, Chief Operating Officer, and Chief Technology Officer.
Tier 1 and Tier 2 mobile operators have been raising prices. Optus’s MVNO amaysim has raised prices to mid A$20 and has seen the other major MVNO brands, such as Woolworths and Aldi, following suit. Tier 1 operators set wholesale prices with MVNOs on monthly and even annual price changes. Every contract is negotiated differently. Supporting growth has been the 1 to 2% p.a. population growth in Australia, improving consumer confidence and low 3% telco expenditure as a percentage of household income.
In the enterprise, Optus is strong in the government, finance and healthcare sectors. But there is a low market share in the mid-market enterprise and will work with Dicker Data to penetrate this segment.
On cost savings, operating expenses have been flat despite high inflation in Australia. Exiting unprofitable business lines are one-off cost savings. Other areas are modernising systems and improving processes but this will not occur within a year. Room for the current 27% EBITDA margin to expand with peers Telstra and TPG at mid 30% EBITDA. Big costs saving is IT and network system. No provision has been made on the civil case brought by the Australian Information Commissioner (OAIC) on the September 2022 cyberattack.
Singtel's associates:
Bharti Telecom: Gift that keeps on giving
On mobile prices, expect increases over the next 3 to 4 years. India has a price per GB and ARPU that is the lowest in the world, and the opportunity to increase is high. There can be more price tiering in India, but it cannot do it alone because of the risk of losing market share. A possibility is price increases before Jio IPO. The quantum of the increase in price is the issue, not when a price hike occurs. The Indian mobile market is better served with three private players due to job losses and foreign investor perception. If Idea does collapse, there will be revenue coming and Bharti has the capacity to serve these customers. Indus Towers' cost may rise and be passed on to operators.
Apart from competitive, there is an organic path to raising ARPU, such as customers switching from feature phones to smartphones, prepaid to postpaid and more international roaming. Excess data charges have been another source. Organic is a 5 to 8 rupees a year ARPU lift.
With the growth in fee cash flows and capital intensity reducing, Bharti will step up dividends and keep trending up. There are also monetisation opportunities in disposing Robi, Dialog and Airtel Payments Bank IPO.
Not using the FCF for additional growth will be a risk and a missed opportunity. The three identified sources of growth are –
Data centres: currently 200MW (half used internally) and will add 500MW in the next five years in key cities such as Mumbai, Bangalore and Chennai;
Cloud business: Bharti already operates one of the most significant clouds in India with 10,000 servers. Moving to the public cloud would have raised its cost by 40-50%. Hyperscalers are useful during occasional spikes in workload. Bharti already has a cloud platform for data management, workflow (buy, build, pay) and channels (customer interface);
Lending: Using non-bank partners and already lending US$600-700mil with its own credit models.
Advanced Info Service (AIS): Competition has stabilised
Mobile tariffs have been stable for the past year, with unreasonable data prices being removed. 2Q25 Blended ARPU up 3.9% y-o-y to Bt442. Without unlimited data plans, customers shift to higher-priced plans, such as Bt200-250, with fixed data. The competitor is also sending a message that it is being rational because, with scale, it can now drive profitability. In postpaid, there is also less device subsidy. 5G penetration rising from the current 35% will also drive up ARPU.
The virtual bank license is partly to strengthen the core business. Mobile banking can bring AIS ecosystem on board and not lose customers. All three partners in the bank (AIS, PTTOR, KTB) have a large existing customer base, resulting in low customer acquisition costs. A virtual bank is less costly to operate because it has no branches and fewer staff. 1H26 will start commercialisation. Capital requirement of Bt5bn by June 2026 and Bt10bn in 5 years.
The data centre is a new source of revenue to expand into the enterprise segment. There is demand from sovereign data, social media and enterprise customers. Gulf can provide security of power with technical and business expertise from Singtel. Singtel also has a relationship with hyperscalers where data needs to be hosted in Thailand.
The dividend payout ratio is 80-90% past 4 years. Reserving some cash for the next spectrum auction and expansion into the virtual bank.
Globe Telecom: Creating a parallel banking eco-system
The 3rd mobile operator, Dito, has an 8% revenue market share. It was mainly at the expense of PLDT. 3Q24 saw price aggression from PLDT as it reacted to gains from Ditto. There is a natural cap to how large Dito can grow. It has not solved their funding. China Telecom is unlikely to inject funding if it cannot increase its stake.
Myint (or GCash) net income grew 79% y-o-y in 2Q25 and accounts for 26% of Globe's net income before tax. Myint has not filed an IPO application and is preparing the next phase of growth. The company has no real needs for funds. Revenue has been more diversified, where 3-4 years ago it was e-wallet (telco reload), revenue currently includes lending, wealth management, GStock, and adtech. Tapping into overseas foreign worker remittances is another opportunity. The GCash wallet can be opened in other countries for Filipinos.
Globe is expanding its base station sites with deeper granularity to meet demand. It has closed too many physical stores, and many customers are not digital. The average waiting time has now declined from 45 minutes to 10 minutes with the store expansion. Also, expanding the retailer distribution network to avoid being absent in certain areas.
The Open Access Network Bill (Konektadong Pinoy Bill) is a positive in allowing network investments in areas missed by incumbents, primarily geographically isolated areas, and increasing the utilisation of existing networks. The purpose of the Bill is universal access. There are loopholes in terms of looser vetting of new players and cybersecurity access. Rather than regulators mandating of lease rates of the network for new entrants, it should be market rates. Incumbents have spent huge sums and need certain returns. There is a risk of investments falling off if incumbents are impacted. As the 3rd player in fibre, it opens up the network of competitors.
Telkomsel: Mobile price repair is fragile in a weak consumer environment.
Consumer confidence has been declining, especially in the middle to low segment. There was a festive boost in 1Q25. Government spending has surged in June and will take several quarters to materialise in the economy. Mid to high customers are still resilient, and with a good customer experience, there is a willingness to pay. However, mid to low customers are looking at more sachet price points, such as 7,14 to 21-day prepaid top-up packages and are more careful with spending.
Telkomsel is simplifying its starter pack with one starter pack across brands. Peers have followed with reduced bonus quota data and shorter maturity period. Raising prices has led to some customer rejection and minor churn in the lower-end market. It shows Telkomsel needs to be careful with price repair, and it remains fragile.
Broadband is a clear engine of growth. Only 15mil households out of 70mil have broadband. There is a need for capital investment in fibre-to-the-home to boost the market. EZnet prepaid plans started at 10MBPs for Rp150k and now offer 20MBPs.
On Telkom spinning off its fibre optic business, it did sell TowerCo in the past. The fibre access was exclusive to Telkomsel and cannot last forever. There are still 5000 towers under Telkomsel. Due to Telkomsel’s high volume of orders, it will be eligible for a larger discount. Telkom has a strategic objective to diversify and not rely solely on Telkomsel for its fibre optic mobile and fixed revenue.
Maintain ACCUMULATE with higher Singtel's target price.
Read more at SGinvestors.io.
Above is an excerpt from a report by Phillip Securities Research. Clients of Phillip Capital may be the first to access the full PDF report @ https://www.stocksbnb.com/.
