New York had the best port in the world. Here's why it lost it

When the container appeared, the biggest ports in the world spent their money improving what they already had, and they lost. The ports that won redesigned everything around the box. The same thing is happening with artificial intelligence: almost every company uses it, but very few have redesigned their processes around it.

In my last column I told the story of the Ideal-X, which sailed from Newark in 1956 with 58 containers on board. According to Marc Levinson, loading a tonne of loose cargo cost $5.83 at the time; on Malcolm McLean’s voyage it cost about 16 cents. That’s only half the story. The other half, the one that stayed with me after the panel at TOC Americas in Cartagena, is about the ports: what the ones that had everything did when the box turned up, and why nearly all of them chose badly.

There’s nothing brilliant about a twelve-metre steel box. What decided the next thirty years was who rebuilt their infrastructure around it and who preferred to carry on as before.

Why did New York lose its port when it had everything?

In the early 1950s, the port of New York handled roughly a third of America’s seaborne trade in manufactured goods. It had 283 piers spread across Manhattan and Brooklyn, and half a million people made their living from it. It was the most important port in the world, and it knew it.

In December 1955 the Port Authority announced a 450-acre terminal at Port Elizabeth, across the river in New Jersey. Over the following years it would become the first major terminal designed from the ground up for containers. The city’s answer was a plan from Mayor Robert Wagner: $130 million to rebuild its own piers. The same old piers. Bigger and newer, but designed for loose cargo and for the longshoremen who moved it by hand.

They were right about everything they could measure. Their piers were the busiest in the world and their workforce was the most experienced. In the mid-fifties, the container was a trucker’s experiment with an old ship.

In The Box, Levinson makes a point I haven’t been able to shake since I first read it: McLean’s real innovation wasn’t the box, it was understanding that the whole cargo system had to be redesigned around it. New York’s numbers bear him out. Longshore workdays in Manhattan fell from 1.4 million in 1963 to around 127,000 in 1976, a 91% drop in thirteen years. Today it’s called the Port of New York and New Jersey, and the order of the names is pure courtesy.

London, Felixstowe and the advantage of having nothing to protect

London did exactly the same. In 1961 its docks had the best year in their history, handling more than 60 million tonnes. Between 1967 and 1981 every one of them closed. Container ships couldn’t get far enough upriver, and nobody was willing to move the port until it was unavoidable. Rotterdam took the traffic.

Felixstowe, on the other hand, was so insignificant that it wasn’t even part of the National Dock Labour Scheme. It became the UK’s leading container port, partly because it had nothing to protect.

That last sentence is the one that matters. New York and London had it all: the infrastructure, the unions, the contracts, the reputation and a century of accumulated know-how. That’s precisely why they read the container as a threat to be contained rather than a tool to adapt to. They did try, in their own way, which was the worst way possible: they took the new technology and squeezed it into the old operation. They rebuilt piers for a world that no longer existed.

What New York’s lost port has to do with AI

Swap the container for artificial intelligence and the story tells itself.

McKinsey’s The state of AI in 2025 survey found that 88% of organisations surveyed already use AI regularly in at least one business function. But only 39% can attribute any EBIT impact to it, and just 6% qualify as what McKinsey calls high performers: companies that attribute more than 5% of their EBIT to AI and say they get significant value from it.

What separates that 6% from everyone else isn’t the model they use. According to the same study, high performers are nearly three times as likely to have fundamentally redesigned their workflows, rather than layering AI on top of the process they already had. It’s the difference between building Port Elizabeth and rebuilding the Manhattan piers.

I see it every week with companies in Chile and across Latin America. The conversation arrives with the technology already chosen: let’s build a chatbot, let’s use AI. It rarely arrives as “let’s look at the problem and how we could transform it”. That’s substitution, not transformation. It’s Mayor Wagner’s plan all over again: it improves today’s operation and leaves it just as incompatible with what’s coming. We’ve seen this before with electricity in factories, where the new motor went in exactly where the steam engine had stood and the productivity gains took decades to show up.

New terminal or old pier? Redesigning processes for AI

In logistics there’s an extra irony. The industry that lived through this lesson is the one with the most to protect today: legacy systems, regulated processes, decades of tacit knowledge and delicate labour relations. That’s exactly where New York stood in 1955. And like New York, it’s right about everything it can measure today. Current AI makes mistakes, doesn’t understand the context of a port and can’t replace a planner with thirty years on the job. All true. And all beside the point for the question that matters: ten years from now, will your operation be where the new infrastructure is, or where the old piers used to be?

The question isn’t only for ports. If you run operations, take the process that eats the most hours and ask why each step exists, not which of them a bot could do. If you’re in IT, count how many of your AI projects connect a model to a system nobody has dared to change. If you own the business, ask yourself which part of it you’re protecting the way New York protected its piers.

Redesigning doesn’t mean betting the company in one go. You can start with one well-bounded process, test it with real data and measure whether the redesign moves the number before you scale it. That’s what a proof of concept is for: it’s the small terminal you build before deciding whether to move the port.

McLean didn’t invent a box. He invented the obligation to redesign everything else. Artificial intelligence is doing the same, and the decision facing executives today isn’t whether to adopt it, because according to McKinsey almost all of them already have. It’s whether they’ll build a new terminal or pour more money into the old pier. New York chose the pier. The order of the names on its port tells you how that went.

FAQ

Why did the port of New York lose out to New Jersey?

Because the city spent its money rebuilding piers for loose cargo while the Port Authority developed a terminal at Port Elizabeth that ended up designed for containers. Container traffic moved to the New Jersey side.

What percentage of companies get real value from AI, according to McKinsey?

In The state of AI in 2025, 88% of organisations use AI in at least one function, but only 39% attribute any EBIT impact to it. Just 6% qualify as high performers, meaning companies that attribute more than 5% of their EBIT to AI and get significant value from it.

What’s the difference between substituting and redesigning a process with AI?

Substituting means putting AI where a person used to do the same task, in the same workflow as always. Redesigning means asking why each step exists and rebuilding the process around what AI does well, with a person reviewing wherever judgement is needed. According to McKinsey, the companies getting the most value from AI are the ones that redesigned their workflows.