Capital Is a Living Discipline: The Case After Sanction

A final investment decision looks like the end of one process and the start of another. The economics get modeled, the sensitivities get stressed, the contractors get selected, the board approves, and the notice to proceed goes out. Everyone shakes hands. The organization shifts from evaluating the project to building it, and the case that was argued for months slides into a shared drive and mostly stops being touched.

That is the moment the case starts to expire. Not because anyone did anything wrong at sanction. Because the world the case was priced in does not stay still for the four, five, six years the build is going to take. Steel moves. Labor moves. Interest rates move. Permitting timelines move. The regulatory backdrop moves. The commodity the plant is going to sell moves. On a two-year software program the case can survive being left alone. On a six-year construction program it cannot. By the second annual review, the case that was approved is not a description of the project any longer — it is a photograph of what the project looked like the week the board voted.

Rio Grande LNG reached FID in mid-2023 at $18.4 billion for its first phase, with $2.3 billion set aside for owner costs and contingencies. Rapidan Energy Group promptly said the contingency was likely too low given labor shortages, and estimated total costs closer to $20 billion including financing and interest across a six-year build. That is not a criticism of the sanction case. It is a description of how quickly a case that was tight on the day of the vote stops being tight once the market moves against it. The sanction number is a snapshot. The capital position is a live thing.

On the construction programs we watch finish inside their case, there is a set of practices that runs from the day after sanction to the day of commissioning. They are not glamorous, and they do not appear in a stage-gate slide. They are the ordinary operator work of keeping the capital position honest.

The case gets reopened on a cadence — monthly at the program level, quarterly at the executive level — and every reopening asks the same three questions. What has moved since we last looked. What is the new best estimate for cost to complete, in the world we are actually in and not the one we sanctioned in. What is the delta between our current authorized capital and what the program now needs, and what are we going to do about it. The reopening is not a request for more money. Sometimes the answer is that scope needs to come out. Sometimes it is that the sequence between phases needs to change. Sometimes it is that a contingency draw is prudent now because a draw in eighteen months will be twice the size. The discipline is that the question gets asked, on a schedule, by someone whose job is to protect the case rather than defend a specific line in it.

Contingency is treated as a working balance, not a rainy-day fund. It is held at the program level rather than distributed into every package, because a contingency parceled out to twenty package managers is a contingency that will be fully drawn by month eighteen regardless of what the program actually needed. Draws are approved against a shared view of what the program looks like at completion, not against the pressure of the current invoice. The programs we see burn through contingency in the first third of the build are almost always the ones that pushed it down into the packages at sanction and lost visibility into it by year two.

Escalation is priced continuously, not once. Steel, alloys, transformers, high-voltage cable, skilled trades — each of these has its own curve, and each of those curves has moved sharply in the last five years. A program that priced escalation once in the sanction case and never reprised it is running on a number that stopped being useful the quarter after it was signed. The programs that hold their case have someone whose job is to keep the escalation view current and to feed it back into the cost-to-complete every month.

In a multi-phase program — two trains, three units, four fabs — capital strategy is also about reallocation between phases while the build is running. Phase one is producing information the sanction case did not have. Productivity is coming in above or below plan. A particular subcontract is going to cost more than the estimate said, or less. A supplier lead time has doubled. The commodity market has moved in a way that changes what phase three should look like. The question is whether the capital plan for phase two and phase three is being rewritten in light of what phase one is teaching, or whether each phase is running against the plan it was sanctioned against and nothing else.

Operators who reallocate between phases treat the program as a portfolio rather than a stack. They will pull scope out of phase three to protect a productivity win they want to lock into phase two. They will move a long-lead order forward to hedge a supplier constraint that phase one exposed. They will accelerate a phase whose commodity window is opening and defer a phase whose window is not. Each of those moves requires the capital position to be live — not a spreadsheet built for the board deck last quarter, but a working document that a small number of people are actually keeping current.

The programs that do not do this end up in the position of the Golden Pass contractor, which by its own account submitted change order requests approaching $2 billion and watched them subjected to reviews long enough that the contractor’s balance sheet ran out before the reviews finished. There is a version of the same story on the owner side of many programs, where the capital request that should have been made in year two gets deferred, then compressed, then made in year four as an emergency ask against a case that no longer resembles the project. The market did not force those decisions. The absence of a discipline for making them earlier did.

The visible cost of a case that stopped being managed is the overrun. Golden Pass added roughly $2 billion between sanction and 2022, then more once the EPC contract broke. Rio Grande’s outside observers expect it to add another billion or two before commissioning. Those are the numbers that make earnings calls and get written up in the trade press.

The quieter costs are harder to see and larger in aggregate. There is the cost of the decision the owner did not make in time — the scope that could have been trimmed cleanly in year one and had to be cut painfully in year three. There is the cost of the contingency that was spent early because no one was watching the balance, so when a real event landed there was nothing left to absorb it. There is the cost of the sequencing that could not be adjusted because the plan had ossified, so a supplier constraint in one phase forced a compression in the next. There is the cost of the counterparty relationships that were damaged when overruns had to be handled as disputes rather than as jointly managed events, and the follow-on cost of the next program being priced with a wider risk premium because the last one ended in litigation.

And there is the strategic cost, which is the hardest to price and often the largest. A program that runs over by twenty percent and delivers eighteen months late does not just cost more — it lands into a different market than the one it was sanctioned for. The commodity price has moved. A competing project has come online. The customer contracts that anchored the case have repriced. The capital that was committed to this program was capital that could not be committed to something else. Every quarter a case is left un-revisited is a quarter in which the option to reallocate quietly closes.

We watched a Gulf Coast owner run a multi-year processing build across two phases without an emergency capital ask. Nothing about the market cooperated. Steel moved. A key alloy went on allocation. Two of the specialty subs missed their productivity numbers in the first year. The program still finished inside its authorized envelope, six weeks later than the sanction date and a small percentage above the sanctioned cost.

What that owner did is unremarkable in isolation and rare in combination. There was one person, senior, whose job was to hold the capital position for the program end to end — not the CFO’s deputy, not the project controls lead, a named principal on the hook for the case. The case got reopened every month at the program level and every quarter at the executive level, with the same three questions each time. Contingency was held centrally and drawn against a shared view of cost to complete, not against the pressure of any single invoice. Escalation was repriced quarterly against a small basket of the commodities and trades that actually mattered to the program, not the general PPI. When phase one taught the program that a certain subcontract was going to cost more than estimated, the phase-two plan for that subcontract was rewritten within the quarter rather than left to be rediscovered when phase two was mobilized.

Two things changed that were harder to see. The relationship with the primary contractors was structured so that change orders got adjudicated on a weekly rhythm, in a room, against the shared view of the case — not filed, batched, and litigated. And leadership treated the quarterly capital review as a decision meeting rather than an information meeting. When the review said scope should come out, scope came out. When it said a phase should be resequenced, the sequence changed. The review had authority to move the plan, not just to describe it. That is the difference between a governance ritual and a capital discipline.

The technical question — how to build the plant — is largely resolved by the time a program is sanctioned. The engineering firms know how to design it. The EPC contractors know how to construct it. The commissioning teams know how to bring it up. What is not resolved, and what only the operator can resolve, is whether the capital position underneath the build is going to be managed as a discipline or handled as a series of milestones.

That decision does not sit with the EPC. It does not sit with the lenders. It does not sit with the board’s capital committee, which sees the program four times a year at most and cannot manage a live case from that cadence. It sits with the operator that owns the program, and it is made not once at sanction but every month for the length of the build. The operators who make it consistently are the ones whose programs finish inside a recognizable version of the case they signed for. The operators who do not are the ones whose programs surprise them, usually in year three, usually against a market that has already moved.

The reason the surprise arrives in year three rather than year one is that the drift is quiet. A single quarter of escalation slippage is invisible. Twelve of them stacked on top of a contingency that was pushed down into the packages at sanction is a program that has moved half a billion dollars away from its case without anyone naming the moment it happened. The discipline is what makes the drift legible — what turns it from a story told at the end into a series of decisions taken along the way, each one small, each one made against a case that is still a description of the project rather than a picture of what it used to be.