Oil, Gas and Mining Are Forty-Three Percent of the New ICSID Docket. Most Cases That Ended This Year Produced No Money at All

Watch · The story in brief

What the numbers say

On 14 August 2026 ICSID published its caseload statistics for the financial year ended 30 June. Sixty new arbitration and conciliation cases were registered. The cumulative total since the Centre opened in 1966 reached 1,118. During the year it administered 363 cases, which it describes as the highest number in any year of its history. That last figure is the one to hold onto: registrations are one measure of appetite, but the administered caseload is the measure of load, and it is at a record.

The sectoral split of the new filings puts extractives at the centre. Oil and gas accounted for 25 percent and mining for 18 percent. Electric power and other energy added 10 percent. Set against that, finance was 10 percent, transportation 8 percent, construction 7 percent, and services and trade 5 percent. Take oil, gas and mining together and 43 percent of everything registered in the year came out of the ground.

The respondent map is equally concentrated. South America accounted for 22 percent of new cases, Eastern Europe and Central Asia 20 percent, and Sub-Saharan Africa 20 percent. Western Europe took 13 percent, the Middle East and North Africa 10 percent, North America 7 percent, Central America and the Caribbean 5 percent, and South and East Asia and the Pacific 3 percent.

The outcome figures are the ones nobody quotes

Of the cases concluded during the year, 37 percent settled or were otherwise discontinued and 63 percent were decided by a tribunal. Within those decisions, 40 percent rejected all claims, 37 percent upheld claims in whole or in part, 18 percent found the tribunal had no jurisdiction, and 5 percent were dismissed for manifest lack of legal merit.

Read that as a claimant. Reaching a tribunal decision at all leaves you slightly more likely to have every claim rejected than to win any part of one. Add the jurisdictional exits and almost three in five decided cases end with the state paying nothing.

The damages distribution reinforces it. Across concluded cases, 70 percent produced no damages award, 14 percent produced under 10 million dollars, 8 percent fell between 10 and 49 million, and 8 percent exceeded 50 million. That is a heavily skewed distribution with a long thin tail, and the tail is what gets written about.

One caution before anyone puts these figures in a pitch. That 70 percent is measured across concluded cases, which bundles settlements and discontinuances together with jurisdictional exits and outright losses. A case that settles confidentially produces no damages award in the statistics while producing money in reality. So the honest reading is not that claimants lose 70 percent of the time. It is that only a minority of ICSID cases end in a public, quantified transfer, and the ones that do are unrepresentative of the whole.

Why extractives keep filling the docket

The dominance of oil, gas and mining is structural rather than cultural. A mine cannot be moved when the fiscal regime changes. Its value sits in a licence, a concession or a permit granted by the state it is arguing with, and the capital is sunk long before the revenue arrives. That combination is the textbook profile of an investment that needs treaty protection, and it is why the sector has been overrepresented in this docket for two decades.

What has changed is the character of the measures. The claims that formed in the last few years have less to do with outright seizure and more to do with instruments that look administrative: a concession cancelled for failure to meet investment thresholds, a mineral reclassified as strategic, an export prohibited pending domestic processing capacity that does not yet exist, a permit annulled on the ground that the original approval lacked legal justification. Those measures are harder to characterise as expropriation and easier to defend as regulation, which is one plausible reading of why so many decisions this year rejected claims outright.

The regional concentration follows the same logic. South America, Central Asia and Sub-Saharan Africa are where the commodities are, where the fiscal pressure is sharpest, and where the last four years of resource nationalism have been most active. Two thirds of the new docket sitting in those three regions is not a coincidence, and anyone holding ground there should read the number as a base rate rather than a headline.

Why it matters for dispute formation

The practical use of a statistics release is calibration. A claim is a multi-year, multi-million-dollar commitment made under uncertainty, and the reference points most boards carry are the handful of very large awards they have read about. This dataset says those are the exception. If a board is being asked to approve a claim, the right comparison is not the largest award in the sector but the distribution of every case that ended this year.

It also changes what leverage looks like. If 37 percent of concluded cases settle or discontinue, then filing is often a way of forcing a negotiation rather than a route to an award, and the value of the claim is in the pressure it creates before a hearing rather than in the judgment at the end of one. That is a different instrument, and it should be resourced and timed differently.

For states the reading is more comfortable than the filing numbers suggest, and that has its own consequence. A government that knows most claims fail, and that most of the rest settle, is a government with less reason to soften a measure in advance. The record administered caseload and the low payout rate point in the same direction: more claims, filed against measures that are harder to characterise, resolving more often in the state's favour. That is the environment the next few years of mining disputes will be argued in.

Who's exposed

What to watch

• Whether the extractives share holds above 40 percent next year, or whether the 2026 figure is a peak driven by the wave of concession cancellations and export measures since 2022.
• Whether the proportion of decisions rejecting all claims stays near 40 percent, which would suggest tribunals are treating reclassification and permit annulment as regulation rather than taking.
• The jurisdictional exit rate, at 18 percent this year, and whether denunciations and treaty terminations push it higher.
• Whether the record administered caseload of 363 translates into longer timetables, since capacity at the Centre and in the arbitrator pool is finite.
• How many of the settled and discontinued 37 percent involved payment, which is the figure the published statistics cannot show and the one that would most change the calculus.