Risks no experiment can retire
Four outcomes no study retires, each with the structural answer we would choose instead of a test.
Four, with the treatment we would choose for each. These are outcomes, not estimates: no amount of work retires them, so each needs a structural answer rather than a study.
The Singapore pilot is a pilot. Nothing obliges its extension beyond 300 MW, the quota is a volume rather than a rate, and the February 2026 award has not been published. Make the strategy survive it: contract a term of at least eight years so that exposure ends with the contract rather than with the scheme, and price against the contract rather than against the pilot's continuation.
Malaysian transit is a single point of failure on the recommended route. Build an escape hatch: site the hub within reach of Songkhla port so that tank-container export remains physically available, while recording plainly that the escape hatch returns less than the hurdle and exists to recover value rather than to earn a return.
Regional supply is building faster than Singapore demand. Straits Bio-LNG alone is designed for 80.5% of the sandbox burn, Indonesia is targeting 100 projects by 2030, and the first physical deal has gone to a Malaysian supplier. Take this bet with eyes open, and price the offtake to a floor rather than to a premium that competing supply will erode.
Thai renewable energy policy is being restructured. The July 2026 move on legacy tariffs covers solar and wind and the direction of travel is downward, which cuts in this project's favor by lowering what a mill forgoes when it sells its gas. Contract feedstock now, while the mills' alternative is falling, and by indexing the feedstock price to the biogas power tariff rather than to a fixed number.
Section added · Pass 5 separates testable hypotheses from risks that can only be treated; the two need different sections because they need different responses