Chief Executive Officer / Singapore / Southeast Asia
Remedies, rural sites, a shared radio network and a lower dividend
- Mobile subscriptions, June 2026
- 13.6 million
- About 68 per cent prepaid, alongside 410,000 fixed-wireless homes
- Revenue in 2025
- RM6.18 bn
- EBITDA of RM2.51 billion, a 40.6 per cent margin
- Net debt to EBITDA, December 2025
- 2.2x
- Including leases; down from 2.7x when the merger completed
Decisions & consequences
01Offering a spectrum remedy before the regulator asked for one
When Altirra and Lembayung signed their merger agreement in March 2023, the combined company would hold about 31 per cent of the country's mobile subscriptions, and more than 45 per cent in Sabah and parts of the east coast. The merger needed the approval of the communications regulator, which also reviews competition in our sector. Every month of review delayed savings worth about RM40 million a year, and left 5,000 employees unsure of their future.
The usual approach is to file, wait for the regulator's concerns and negotiate the remedies down. Lembayung's chair and the government-linked shareholder's nominees favoured it. Lembayung's chief technology officer objected to giving up any spectrum, because the 2.6 GHz band carried much of our capacity in the Klang Valley. I argued that we should file our own package with the application: return 2 × 10 MHz in the 2.6 GHz band, publish a five-year wholesale reference offer for virtual operators, freeze our cheapest prepaid pass for 24 months and keep every rural site that served only one of the two networks for three years. The regulator would ask for something similar anyway, and we would design it better ourselves.
We gave up spectrum that our capacity model valued at about RM380 million in avoided network spending, and we accepted competition from virtual operators on terms we had set ourselves. The regulator accepted the package in February 2024 with one addition, an independent annual audit of the wholesale offer, and the merger completed on 1 March, ten months after filing.
Replacing the capacity cost RM210 million in small cells and re-farmed 1800 MHz spectrum. The reference offer proved more attractive than I had judged: virtual operators had taken about 640,000 of our subscriptions by June 2026, against our estimate of 300,000, although the wholesale revenue they pay us now exceeds RM95 million a year.
Remedy compliance now has its own team and its own committee report, and nobody at Seralis describes the undertakings as the regulator's.
What I would do differently: I would have tied the wholesale price to a floor derived from our own retail cost of serving a prepaid customer, which I believe the regulator would have accepted. And I would have told our employees about the remedies in our own words before they read about them.
Keeping 900 rural sites the integration plan had marked for closure
After completion we had 14,600 radio sites, and our integration advisers identified 4,300 of them as duplicates: sites from the two former networks close enough that one could be switched off and its traffic carried by the other. Decommissioning all 4,300 would save about RM260 million a year in rent, power and maintenance, and that saving formed part of the RM750 million figure we had given investors.
Our regional engineers in Sabah, Sarawak, Kelantan and Pahang pointed out that about 900 of those duplicates were not duplicates in the way that mattered. In rural districts each pair of sites drew power from different feeders and backhaul from different routes. When one failed in a flood, the other usually kept working. On paper we were paying twice for coverage. In practice we were paying for resilience that neither former company had ever priced.
I decided to decommission 3,400 sites and keep the 900, upgrading their batteries and reassigning them as capacity and standby sites. Our chief financial officer opposed the decision, reasonably: it removed about RM55 million a year from the savings plan and obliged us to explain a shortfall to investors. A nominee director asked why rural resilience should be a cost to shareholders rather than a matter for the universal service programme. I did not have a complete answer to that question then, and I still do not.
The decommissioning programme finished in March 2026. During the north-east monsoon floods at the end of 2024, 61 of the retained sites carried traffic for communities whose main site had lost power or backhaul for more than a day. Monthly prepaid churn in the rural districts where we kept duplicates averaged 1.9 per cent during the integration, against 2.6 per cent in comparable districts where we did not. Run-rate savings reached RM610 million by June 2026, and we now expect about RM690 million by the end of the year rather than RM750 million.
Resilience now has its own line in our capital plan, and site decisions in flood-prone districts go to the Board Risk and Sustainability Committee.
What I would do differently: I would have tested the duplicates against our flood records before we announced a savings figure, not after. The shortfall was defensible. The surprise was not.
Sharing a radio network with the rival we were trying to beat
In 2016 Veyara Mobile was the third-largest of Cambodia's mobile operators, with 3.1 million subscriptions and a 4G network that reached 38 per cent of the population, almost all of it in Phnom Penh, Siem Reap and the larger provincial towns. Our capital budget was about US$30 million a year. At that rate, reaching the rural provinces would take most of a decade, and our larger competitors were already there.
One of those competitors proposed that we share the radio equipment on rural sites, each of us keeping its own spectrum and core network. We would host the shared equipment in the north-west and they would host it in the south-east, 2,100 sites between us. The group marketing team at Corvantis opposed the idea, arguing that rural coverage was the one advantage we could build over time and that sharing would give it away. My own chief technology officer worried about relying on a rival's engineers to repair half of our rural network.
Both concerns were legitimate. I supported the agreement because building alone meant not reaching those customers for years. We gave up any claim to better rural coverage and handed control of fault repair on about 1,050 sites to a competitor. The arrangement needed the regulator's approval, which it gave after we showed that each operator kept its own spectrum, core network and customer data.
The shared network went live in stages from late 2017. By 2019 our 4G coverage reached 81 per cent of the population, and we avoided about US$52 million of capital spending over four years. Subscriptions grew to 4.3 million and EBITDA margin rose from 21 to 34 per cent. In the first year, however, faults on sites hosted by our partner took about 40 per cent longer to restore than faults on our own, and rural customers noticed before our reports did.
Veyara and its partner later set up a joint fault centre, and the revised terms were written into the sharing template I had drafted for Corvantis years earlier.
What I would do differently: I would have put restoration penalties in both directions, and the joint fault centre, into the original contract. We relied on goodwill for a year, and goodwill does not answer a customer whose site is down.
Lowering the dividend with a government-linked shareholder in the room
Lembayung had paid out more than 90 per cent of its profit as dividends for a decade, and its last full-year dividend before the merger was 21.5 sen a share. The government-linked investment company that holds 24 per cent of Seralis, and many individual shareholders, had come to treat that payout as a fixed feature of the company. At completion our net debt, including leases, was RM5.8 billion, 2.7 times EBITDA, and we needed about RM1.1 billion over three years to connect 6,000 sites with fibre backhaul and add capacity for fixed-wireless homes.
There were two ways to fund it. We could keep the payout and borrow, which our banks were willing to support, or lower the payout and fund the programme mostly from cash. The government-linked shareholder's nominee directors preferred the first, pointing to the income on which their beneficiaries depend. Teguh Permata had hoped for a special dividend after completion. I recommended the second: a payout of 60 to 70 per cent of normalised profit for 2025 to 2027, an aim of net debt at 2.0 times EBITDA by the end of 2027 and a commitment to review the policy then.
The board debated it across two meetings. The chair asked me to set out the borrowing alternative in full, with its interest cost and its likely effect on our credit rating, before the vote. The board approved the lower payout. Our share price fell 9 per cent on the day of the announcement in February 2025, and a number of individual shareholders wrote to me directly.
The dividend for 2025 was 15.0 sen a share, a payout of 65 per cent. Net debt fell to 2.2 times EBITDA by December 2025, and by September 2026 3,700 of the 6,000 sites had fibre backhaul. The share price had recovered its February 2025 level by the end of that year.
The capital programme now reports to the board every quarter against the milestones we set when we lowered the dividend.
What I would do differently: I would have published those milestones on the day we announced the change, so that shareholders could measure what they had given up against what we were building. We published them a quarter later, and the delay cost us trust we then had to earn back.