Three funding decisions and what each one cost

Selling most of our towers while keeping the sites our network still needed

In 2023 a specialist tower company, PT Menaraksa Infrastruktur, offered Rp 3.9 trillion for all 2,380 towers left from our fixed-wireless years. The towers hosted about 3,330 mobile-operator tenancies. With net debt near four times EBITDA and a large dollar loan outstanding, the offer was attractive. The buyer also wanted us to lease back space on 900 towers for fifteen years.

The founding family's representative on the board of commissioners favoured selling everything and paying a special dividend. Our network director wanted to keep 1,100 towers. My concern was narrower. Some 520 towers in Sumatra carried microwave links for enterprise customers along routes our fibre would not reach for years, and 220 sites had unresolved land titles that the buyer would discount heavily. I recommended selling 1,640 towers and keeping 740. Our equipment would stay on the 520 Sumatra sites under a ten-year leaseback, rather than a fifteen-year arrangement covering 900 towers.

The directors adopted it and the commissioners approved. We sold 1,640 towers for Rp 2.79 trillion; the narrower leaseback reduced the price per tower slightly. We gave up about Rp 1.1 trillion of immediate proceeds, and the family accepted the regular dividend instead of a special one. We also kept a small tower team costing Rp 36 billion a year.

Proceeds repaid US$120 million of the dollar loan. Net debt fell from 4.0 to 2.4 times EBITDA before lease liabilities, or 2.8 times including them. I reported both figures, because leaseback rentals fall below EBITDA under the lease standard and would otherwise flatter our results. Only the gain relating to rights transferred to the buyer could be recognised.

What I would do differently: I accepted the buyer's power-charge clause too readily; it cost Rp 14 billion more than planned in the first year. I would also ask network colleagues for a five-year fibre map before deciding which towers to keep; 160 of the retained Sumatra sites have since been reached by fibre.

Funding a dollar-priced fibre programme through a weakening rupiah

In early 2024 we approved a Rp 2.4 trillion programme to bring fibre to twelve secondary cities in Java and Sumatra over two years. About 55 per cent of the cost, including optical line terminals, home routers and splicing equipment, was priced in dollars. We had budgeted at Rp 15,600 to the dollar. By mid-year the rupiah had weakened by about six per cent, and the commercial director wanted all twelve cities launched together before a competitor arrived.

I proposed three changes. Treasury would hedge 60 per cent of committed dollar orders for the next twelve months with forward contracts. Procurement would move home routers to a Batam assembler pricing in rupiah at a five per cent premium. And the cities would open in two waves: seven first, then five more only if paying connections in the first wave reached 30 per cent of installed ports within nine months.

The objections were fair. A commissioner on the audit committee called the forward premium, about Rp 33 billion, an insurance policy we might not need. Procurement worried about an unfamiliar supplier. The commercial director warned that delay would hand two cities to a rival. The board of directors accepted the staged plan and the hedge.

Wave one cost Rp 1.38 trillion against a budget of Rp 1.33 trillion. The unhedged share absorbed Rp 29 billion of currency loss; without cover we estimated about Rp 74 billion. The first lot of Batam routers failed at 1.9 per cent against 0.7 per cent for imported units, until the assembler changed a component. Paying connections reached 27 per cent of ports at nine months, short of the condition. We released three of the five remaining cities and deferred two, where a competitor launched first.

What I would do differently: I matched forward contracts to the budget calendar rather than to delivery dates. When shipments slipped, rolling the contracts cost Rp 6 billion that a delivery-based schedule would have avoided.

Replacing our last dollar bank loan with rupiah bonds at a higher stated cost

By 2025 US$140 million remained of the syndicated loan we had raised in 2018, due in June 2026. Every rupiah of our revenue came from Indonesian customers. Since drawdown, the rupiah's depreciation had added roughly Rp 300 billion to the local value of the outstanding principal.

The banks offered to extend the loan in dollars for five years at an all-in rate near seven per cent. Our president director favoured that option because the headline cost was lowest. I compared it with a rupiah alternative: Rp 1.5 trillion of bonds in three-year and five-year series, plus a Rp 800 billion facility from two domestic banks, at a blended cost near 8.3 per cent. Fully hedging the dollar extension would have cost more than the rupiah route. Leaving it unhedged meant betting our interest cover on the exchange rate.

I recommended the rupiah route and accepted that reported interest would rise by about Rp 85 billion a year. The directors agreed after two discussions, and the commissioners endorsed it.

The bond offering required a public rating, a trustee and a prospectus that set out our subscriber figures. Due diligence showed that our count still included homes suspended for non-payment for up to ninety days. I proposed counting only homes suspended for thirty days or less. The commercial director argued that rivals used the looser measure. We changed it anyway, and the December 2024 base fell from 1.01 million to 962,000 connections, with a published reconciliation.

The bonds were issued in October 2025 and were 1.7 times subscribed. We repaid the loan in December 2025. Apart from US$22 million of hedged vendor credit, we now carry no dollar debt. Interest cover moved from 5.2 to 4.6 times.

What I would do differently: I should have tightened the subscriber definition a year earlier, on its own merits. Making the change inside a prospectus invited the suspicion that we did it only because lawyers were reading.