Decisions that gave something up to stay credible

Retiring the Johor coal station four years before its contract ended

In 2022 our 1,000-megawatt coal station in Johor had eleven years left on its power purchase agreement and a major overhaul due in 2028. Commissioned in 2008, it earned about US$110 million of EBITDA a year, almost all from capacity payments that Malaysia's single buyer makes for availability. It also emitted close to 6 million tonnes of carbon dioxide a year, a third of the group's total, and our lenders had begun asking how we would refinance it when fewer banks would lend to coal.

I proposed negotiating closure at the end of 2029. Our chief financial officer opposed it: the station was the steadiest cash flow we owned, and losing it early tightened the cover on our dividend. The two directors nominated by Kelawari Holdings Berhad raised the 410 people employed at the site, most of them from nearby towns. Both objections were sound. On our model the four lost years cost about US$140 million of net present value.

Over two years we built a package that gave each party something. The single buyer would stop paying capacity charges after 2029 and keep the connection point for replacement capacity. A multilateral development bank and two commercial lenders refinanced the station's remaining US$310 million of debt on concessional terms tied to the closure date. Avoiding the 2028 overhaul and later sustaining spending saved about US$60 million. We committed in writing to offer every permanent employee a new role, retraining or voluntary retirement. The board approved the agreement in August 2024, and government approval followed that December.

The refinancing recovered about US$35 million, so our shareholders bore roughly US$45 million. In 2025 we won a competitive tender, never promised to us, to build 400 megawatts of four-hour battery storage at the site using the station's connection. By September 2026, 180 staff had moved to the battery project or to other plants and 70 were in retraining.

The board now treats a retirement date as something to negotiate rather than a liability to defer. What I would do differently: I told the lenders before I told the station's workforce, and staff learnt of the plan from a public statement three weeks after the banks did. I would brief the station's managers and union representatives first, under confidentiality, and design the redeployment with them from the start.

Leaving a Vietnamese LNG-to-power project after three years of development

In 2019 Lintasari had spent three years and US$31 million developing a 1,500-megawatt LNG-fired power project in central Vietnam with a provincial partner. The plan needed a floating import terminal, a long-term LNG supply contract and a power purchase agreement with Vietnam's state utility. I had joined eighteen months earlier and reviewed the project in each quarterly portfolio round.

The numbers would not close. The buyer's latest draft offered a minimum contracted output of 65 per cent of the plant's annual energy. The three LNG suppliers who had bid wanted take-or-pay commitments of 85 to 90 per cent of contracted volumes. The draft also lacked a government undertaking on converting the buyer's dong payments into dollars, which our lenders treated as essential. In a wet year, when hydropower displaced gas, we would pay for fuel we could not burn. The next phase, front-end engineering and financing, would cost a further US$120 million before any of this was resolved.

In November 2019 I recommended that we leave. Our chief executive and our head of gas disagreed. They argued that Vietnam's LNG market would be large, that early entrants would hold the best coastal sites and that a negotiating draft was not a final position. They were right that the market would grow. The board compromised: six more months, with three written conditions on minimum offtake, currency conversion and the terminal's permits. None was met by May 2020, as the pandemic slowed every ministry, and in July 2020 the board approved our exit.

Spending had reached US$38 million. In 2021 we sold the site rights and studies to another developer for US$9 million, so the net write-off was US$29 million. To my knowledge the project had still not reached financial close by mid-2026.

The lasting change is our stage-gate rule. No project may spend more than US$5 million of development money until the buyer has agreed heads of terms on how volume, price and currency risk will be shared, and every project receives written exit conditions when it enters the pipeline. I would do one thing differently. I should have proposed those conditions in my first review in 2018, not eighteen months later; the six-month extension alone cost US$7 million that earlier conditions would have saved.

Taking 40 per cent of the Selat Hijau export consortium instead of control

In 2022 we decided to bid in Singapore's process for importing low-carbon electricity. Our proposal was solar and battery storage in Indonesia's Riau Islands, exporting 1,000 megawatts to Singapore through a subsea cable of about 70 kilometres. The full project, with roughly 2,000 megawatts of solar panels, 4,000 megawatt-hours of storage and the cable, was estimated at US$5.8 billion.

Our chief executive wanted 60 per cent, the operator's role and full consolidation of the earnings. I recommended 40 per cent, with PT Lingga Daya Kepulauan Tbk, a fictional listed Indonesian renewables and construction group, holding 35 per cent and Harrowgate Infrastructure Partners, a fictional Singapore fund, the remaining 25. Part of my reasoning was financial: at 60 per cent our share of the equity would have approached US$1 billion, close to a third of our market value. Most of it was political. Indonesia expects export projects to strengthen its own manufacturing and serve domestic needs, and an Indonesian owner with real equity would answer for that over thirty years far better than we could. Our chief financial officer supported me. The chief executive accepted the proposal after a board discussion in which his 60 per cent option was presented with its own numbers.

We gave up the operator's role, which went to our Indonesian partner, and about a third of the equity value we would have held. In return our development spending is capped at US$18 million until financial close, of which we had spent US$14 million by mid-2026.

The consortium received conditional approval from Singapore's energy regulator in 2024. Our partner has signed supply agreements with two Indonesian module and battery assemblers for a minimum share of the equipment. Financial close is planned for 2028, and nothing about it is certain: the cable route, the financing and the final export permits remain open.

What changed is a group rule that cross-border projects begin with a host-country partner holding meaningful equity. What I would do differently is survey the cable route before choosing the landing site in Singapore. A rerouting to avoid an anchorage area added nine months and US$3 million to the development budget, and a seabed survey costing a fraction of that would have shown the problem at the start.

Pricing carbon into investment cases before any regulator required it

By 2022 our Investment Committee was approving assets that would run into the 2040s using cases that priced carbon at zero wherever no tax applied, which meant nearly everywhere we operated. I proposed a shadow price of US$40 a tonne of carbon dioxide from 2023, rising in equal steps to US$80 by 2030, applied to every new investment and every life extension, and reported beside the unpriced return.

Our head of thermal generation objected most strongly. His team was preparing two 300-megawatt open-cycle gas peakers in Malaysia, with an unpriced return of 11.5 per cent against our 9 per cent hurdle; with the shadow price the return fell to 7.9 per cent. He argued that competitors who ignored carbon would win the contracts we gave up. Our chief financial officer worried that lenders would read a lower internal return as a weaker project. Both concerns were reasonable. My answer was that the price was a test of resilience rather than a forecast, and that our buyers and lenders were already asking how our assets would fare under a carbon tax.

The board adopted the policy in December 2022. In its first three years it changed the ranking of three out of eleven proposals. We withdrew the peakers and bid a 250-megawatt, four-hour battery instead, which won a contract in 2024 at lower revenue. A competitor won a peaker contract we might otherwise have contested, so the policy had a visible cost. We also declined a US$95 million retrofit to co-fire ammonia at our East Java coal plant, because even with the shadow price it cut too little carbon for the money. And a Philippine wind repowering that had looked marginal cleared the hurdle once the emissions it displaced were valued.

The board now sees two returns for every proposal and asks which one the decision depends on, and our lenders request the priced case in their own reviews. What I would do differently is use a range rather than a single price. Low, central and high paths would have shown which decisions were sensitive to the assumption and which were not, and spared us long arguments about a number nobody could know.