201
vessels crossed the Strait of Hormuz in March 2026.
In peacetime, 120 cross every single day.
The Gulf crisis didn't break Lloyd's.
Cookie-cutter transformation did.
Digitisation was the right move. Leaving out the systemic intelligence wasn't.
Quick transformations build efficient systems. The world doesn't do efficient. It does unexpected. The Gulf just sent the invoice.
In the days after February 28, 2026, the world's most critical energy chokepoint froze.
When US and Israeli forces launched Operation Epic Fury, marine risk models hit their thresholds and panicked. This was not a black swan. Hormuz has been identified as a chokepoint risk for decades. Iran's military capability was known. The scenario was entirely foreseeable. But the algorithmic systems had never been designed to handle it — only to price efficiently in stable conditions. The market generated astronomical go-away pricing — up to 10 percent of hull value for a single transit. The system's only available response was to exit.
Washington stepped into a role London had held for three centuries — not through a treaty, but through a Truth Social post. The US Development Finance Corporation created a $20 billion reinsurance facility with Chubb. When the market said hull and cargo coverage wasn't enough, they expanded it on March 20 to include catastrophic liability. The financial model is now complete.
And yet. Western ships still aren't moving.
Instead, Iran did something the model never anticipated. It didn't just exploit the vacuum the insurance market left. It filled it.
The IRGC established a selective blockade, repurposing the strait into a geopolitical sorting mechanism. Chinese ultra-large container ships and Indian LPG carriers transit safely under IRGC escort. At least two vessels have paid $2 million per crossing — settled in Chinese yuan. Iran's parliament is advancing legislation to formalise transit tolls payable in rials.
This is not a temporary disruption. It is a structural shift. The toll isn't just a revenue stream — it is a currency event. Energy trade through the world's most critical chokepoint is moving from dollars to yuan. The insurance market didn't just create a physical vacuum. It accelerated the de-dollarisation of global energy. China is transiting. The West is stranded. And the architecture that made this possible was a risk model built in London.
You can indemnify a ship on a spreadsheet.
You cannot convince a human captain to sail into a drone barrage.
You cannot model for the moment a sovereign state turns your withdrawal into their own revenue stream.
Every finance and transformation leader watching this should feel something shift. Because what just happened is not primarily a story about war.
Because I have watched this exact pattern play out before.
And when the shock arrives, discover that nobody owns the consequence.
I have watched this pattern play out in every transformation programme I have sat inside for two decades.
The Gulf just gave it a name.
In 1688, Edward Lloyd opened a coffee house on Tower Street in London. Ships' captains, merchants, and financiers gathered to share intelligence — weather patterns, piracy reports, the creditworthiness of trading houses. From this exchange, men began writing their names under descriptions of voyages, committing to cover a share of the risk.
What made this architecture extraordinary wasn't the maths. It was the design.
338 years. Two world wars. Pandemics. Catastrophes. The architecture held — because the judgment was built into the room.
But what made it work wasn't just the intelligence. It was the trust architecture. When an underwriter wrote his name under a risk, he was personally accountable. He sat across from the people who would know if he was wrong. Reputation was the operating system — not compliance, not a control framework, not an audit trail. The human wasn't just in the loop. The human was the loop.
Lloyd's Coffee House concentrated multi-domain intelligence in one physical room. It fostered long memory: reputations, track records, and patterns across wars and trade cycles that lived in the room for decades. And it forced underwriters to confront what they didn't know — because the person sitting across from them might know it.
The original AI in finance was a human being. And it was extraordinary at the one thing modern systems cannot do: it knew what it didn't know.
War Risks Insurance Act 1939 — jointly designed by Lloyd's and UK government. Pool Re after IRA bombings (1993). TRIA after 9/11 (2002). Market and state built the response together.
Market exited. Waited for the US to build the DFC/Chubb facility after the fact. Lloyd's wasn't at the design table. It was a bystander.
1984 Tanker War — underwriters priced individual voyages. This vessel, this route, this convoy, this date. Actual intelligence distinguished 2% risk from 15%.
Models applied a blanket regional threshold. Everything inside Hormuz got the same treatment: exit.
Lloyd's Names had unlimited personal liability. When your fortune is on the line, you don't hit a threshold and walk away. You think harder.
No personal consequence. No judgment required. The algorithm carried no cost for being wrong.
In the crises where the principles held, the architecture survived. Where they failed, Lloyd's reformed and rebuilt. In 2026, they failed and nobody rebuilt.
In every previous crisis, something existed between individual rational action and systemic catastrophe. A war risk pool. A convoy system. A government partnership. Collective risk architecture.
In 2026, for the first time in 338 years, the market exited and waited for someone else to build the solution.
The principles that held for 338 years didn't fail. They were never carried into the digital architecture. Here is what needs to be built.
Government-market scenario desks. Shared intelligence frameworks. Pre-agreed backstop triggers before a crisis arrives.
1939: Lloyd's and the UK government built it together. 2026: the market waited to be rescued.
AI-powered risk distinction. Vessel-level, route-level, real-time intelligence. Not blanket regional thresholds.
1984: priced each voyage individually. 2026: one threshold, one response, entire region.
Systemic impact checkpoints. Human sign-off before portfolio-wide exits. Judgment built into the architecture.
Names had personal liability. The architecture forced thinking. 2026: the algorithm had no consequence for leaving.
Not replacing digitisation.
Completing it.
I have spent more than two decades at the intersection of finance and technology. I have sat in the rooms where these operating models are designed. I have watched strategic judgment get descoped from a transformation programme because it couldn't be quantified, measured, or put into a sprint.
In every finance transformation I have worked on, the pattern is the same. The business requirements capture the process. The workflows capture the decisions. The controls capture the thresholds.
"What happens when the whole market hits this threshold at the same time?"
So it never gets built.
A senior underwriter asked:
What is the actual probability?
What do I know about the trajectory?
What does the other side do when I withdraw?
The model hit a threshold and did the only thing it was designed to do:
Exit.
Finance and transformation leaders have developed a dangerous habit of believing that what they are building is insulated from the world.
Consider what the 2026 Iran crisis did to the global technology sector. When Iranian missiles struck Qatar's Ras Laffan LNG complex, they wiped out roughly a third of the global helium supply overnight. Helium is irreplaceable in semiconductor manufacturing for cooling silicon wafers.
South Korean chipmakers like Samsung and SK Hynix import nearly 65 percent of their helium from Qatar. While they hold four to six months of buffer inventory, they are now paying exorbitant emergency premiums to secure alternative supplies.
A war. An insurance panic. A gas field. A helium shortage. A massive margin squeeze on the AI chips every technology company in the world is betting its future on.
On March 1, Iranian drones struck three AWS data centres in the UAE and Bahrain — the first military attack on a hyperscale cloud provider in history. A drone costing tens of thousands hit infrastructure worth tens of millions. The nature of warfare has changed. Our risk models are still pricing for the old one.
The AI systems we are building to make finance smarter depend on the same supply chain this crisis is squeezing — and run on the same physical infrastructure that drones just hit. We are asking AI to solve the problem while everything AI depends on is under threat.
Seven steps. Nobody's model connected them.
Yet each link was foreseeable.
I use AI every day. I use it to research, to stress-test arguments, to find connections across domains I couldn't cover alone. AI is not the problem.
Same training data. Same risk models. Same blind spots. Same decision — made faster, at scale, with no human in the loop.
AI that integrates geopolitics, supply chains, and shipping data in real time. AI that sees second-order consequences. Intelligence.
AI without systemic thinking doesn't fix the problem. It makes it all the same — same data, same models, same blind spots — but faster, and at scale.
The Bank of England calls it correlated positions.
The plain English is: automated stampede.
The Gulf crisis is a design brief. Five specifications. Each one rebuilds a first principle that held for 338 years.
Formal review before portfolio-wide model changes. A system can be individually correct and collectively catastrophic.
The question the 1984 underwriter asked. Now build it into the system.
REBUILDS: ACCOUNTABILITYA team mapping chains across geopolitics, supply chains, and finance. AI can power these — but only if someone builds them.
What connects this system to the world outside it?
REBUILDS: CO-DESIGNQuarterly tabletop exercises under real stress scenarios. Not a risk register — a feared outcome.
Have we practised when the feared outcome actually happens?
REBUILDS: DIFFERENTIATERed-team perspective review for every cancellation committee.
How could a competing state weaponise our rational response?
REBUILDS: DIFFERENTIATECapture tacit knowledge. Build it into the architecture.
Are we transferring judgment — or just compliance?
REBUILDS: CO-DESIGN + ACCOUNTABILITYThese are the specifications.
The US stepped up when London couldn't.
Three centuries of institutional knowledge was made structurally fragile by two decades of efficiency-driven transformation.
A sovereign state walked into the space the market left behind and built a new architecture — a state-controlled toll system, settled in yuan, that sorts the world's shipping by geopolitical alignment. China transits. The West is stranded. The models didn't just fail to predict the crisis. They handed the infrastructure to restructure global energy trade to someone else entirely.
The question now is not whether to transform.
It is whether to transform with strategic intelligence embedded in the architecture — or to keep building systems that perform perfectly right up until the world stops cooperating with the roadmap.
The Gulf just showed us
which path we have been on.