Dr. Mehdi Snène
Senior Advisor, United Nations Office for Digital and Emerging Technologies
As a stronger El Niño arrives, AI forecasting and compute power are rewriting global trade and governance.
El Niño is back. The WMO confirmed a clear shift in the equatorial Pacific, with sea-surface temperatures rising rapidly and high confidence in the onset of El Niño and further intensification.
Some forecasters have drawn comparisons to the 1877-78 event, considered one of the strongest, when Pacific warming contributed to widespread droughts, triggering famines across Asia, Africa and South America.
The 2023 episode exposed how brittle commodity-dependent trade agreements had become. Cascading shortfalls in rice and palm oil across Southeast Asia revealed frameworks built on historical averages rather than dynamic risk.
This time, with a larger disruption forecasted, policymakers have less excuse for being caught unprepared.
Power of modelling and real-time action
Digital transformation has changed what’s possible. AI-driven forecasting models now integrate atmospheric data, shipping telemetry, commodity futures and geopolitical signals into real-time risk assessments. What once required months of inter-agency analysis can be compressed into hours.
The power isn’t just in prediction; it’s in design: running thousands of scenario models to stress-test trade corridors, tariff structures and supply chain dependencies before disruption lands. When a signal crosses a threshold, policy instruments need to move with it.
And national digital strategies must treat compute capacity as a strategic reserve, equivalent to energy stocks or grain buffers
Compute: a tradeable asset
Geopolitically, the stakes are asymmetric. Nations with advanced computational infrastructure hold a structural advantage in forecast-driven decision-making.
Access to high-performance infrastructure and trained models is reshaping who anticipates disruption and who merely absorbs it. Agreements over AI infrastructure, data-sharing protocols and model interoperability are becoming as consequential as traditional trade deals.
Building policy frameworks that keep pace
Trade agreements need dynamic adjustment clauses that trigger renegotiation when AI-validated disruption signals cross defined thresholds. Multilateral bodies need shared forecasting infrastructure, not just shared data. And national digital strategies must treat compute capacity as a strategic reserve, equivalent to energy stocks or grain buffers.
The gap between what AI systems can detect and what policy instruments can respond to is widening. Closing it means moving from episodic summits to continuous, model-informed policy cycles, where the forecast and the framework evolve together.
