China-Europe Rail Passes 140,000 Trains With 96 Domestic Lines and 237 EU Cities

The Analysis

The announcement is built around a departure. A train left Shenyang's Puhe station for Moscow, and that particular departure is the one that took the China-Europe rail network past 140,000 cumulative services and more than 560 billion dollars of goods carried since the first one. Ninety-six scheduled domestic lines running at 120 kilometres an hour, 132 Chinese cities on them, 237 cities in 26 European countries and more than 100 cities in 12 Asian countries at the far end, spread across a northern, a central and a southern corridor. May alone ran 1,965 trains and 197,000 TEU, a monthly record. Two numbers in that list describe a road. The rest describe a timetable.

Put the pin in the map where it belongs. Shenyang, Puhe station, north-east China, a train loaded and cleared and sent north-west towards Moscow. That is the northern corridor, and from Shenyang the natural exit is Manzhouli, which is the crossing that has carried this traffic since the corridor was a hopeful line on a slide rather than a timetable. The central corridor leaves through Erenhot and the southern one through Alashankou and Horgos, and the three of them do not converge until they are west of the Chinese border. So when the announcement says 237 cities in 26 countries, it is describing three roads that share a starting region and almost nothing else. A shipper in Ningbo and a shipper in Chengdu are not on the same product even when the destination city is the same city.

Now do the division, because the division is where the story actually is. May: 1,965 trains, 197,000 TEU. That is 100.3 TEU a train, call it a hundred, which is about fifty FEU. Fifty forty-foot boxes is one train, and one train is the unit of capacity on this road. That single figure reorganises how you should read everything else in the release. A shipper moving five FEU a month is a tenth of a train, which means the shipper is not buying rail, the shipper is buying a slot inside somebody's block train and the somebody is a consolidator in Chengdu, Zhengzhou, Yiwu or Shenyang. The allocation risk does not sit with the operator. It sits one layer down, with the party who promised you the slot before the train was full.

Do the other division. Five hundred and sixty billion dollars over 140,000 trains is four million dollars of goods a train. Divide by the hundred TEU and you get 40,000 dollars of value per TEU as a network average. That number is the reason this corridor exists in the shape it does. At 40,000 dollars a box, freight is a single-digit share of the goods value and the economics work; at 4,000 dollars a box, the same freight is a double-digit share and the buyer ships by sea and waits. Rail between China and Europe is not a cheaper sea route. It is a different product for cargo whose value density can carry the rate, which is why the trains are full of machinery, electronics, auto parts and specialty chemicals and empty of furniture and floor tiles.

Forty thousand dollars a box also changes what a train is worth losing. A fifty-box train at that average is two million dollars of cargo on one path with one crew, and that single fact explains most of the operational behaviour you see on this corridor: the seals, the inspections, the escorts in some stretches, the insistence on knowing exactly which box is where before the train is closed. It also explains why a single incident on a corridor reprices the whole corridor for a quarter, because the underwriters do not average across trains, they look at the concentration. If your cargo is above the average, you are the reason the train is worth insuring, and you should be buying the insurance yourself rather than trusting a carrier's limit.

The choice of Shenyang for the milestone departure is not neutral, either. The north-east is an industrial base with its own export mix and its own distance problem, and a train leaving Puhe is a statement that this node has enough volume to fill a scheduled path without borrowing cargo from a thousand kilometres west. Node competition on this corridor is real and it is quiet: cities compete for scheduled lines the way ports compete for calls, and the ones that win the path get the consolidator, and the one that gets the consolidator gets the volume. When you read that a milestone train left a particular station, read it as that station telling you it has arrived.

Then the number that everybody quoted and nobody read correctly. 140,000 is a cumulative count. It is a stock, not a flow, and it includes every train since the first one, so it can only ever go up. Set it against the flow: 1,965 in a record month, which annualises to 23,580, which means the entire 140,000 is worth about six years at the current record pace. Read that again. The whole history of the corridor, at its best month repeated twelve times, is six years of running. That is not a criticism of the milestone. It is a warning about what the milestone measures. A stock number tells you the road exists. A flow number tells you whether you can get on it this month, and only one of those two questions matters when you are trying to book.

The 96 lines are the real operating fact in the release, and they are buried under the big number. Ninety-six scheduled lines at 120 kilometres an hour, touching 132 cities. Scheduled is the word doing the work. A scheduled line means a path is reserved in the national timetable, and a reserved path is the scarce thing on this corridor, not wagons and not locomotives. Ninety-six paths across 132 cities is 1.4 cities per line, which means most of those cities are a stop or a collection point on a line rather than the origin of their own train. If your factory is in one of the 132 and your product is on a line that runs twice a week, your booking window is two slots a week whether the release says 140,000 or 14 million.

And 237 cities in 26 countries is nine cities per country on average. That average is the most misleading number in the whole announcement, because it sounds like reach and it behaves like a hub map. Nine cities per country means the corridor calls at a handful of consolidation nodes in each country and distributes onward by truck, which is fine and normal and completely different from a direct service to the city on your purchase order. Ask a simple question of any quote: is this city a rail node or a truck leg from a rail node. The answer changes your transit, your cut-off and your damage claim, and it is almost never in the rate sheet.

Here is the thing the coverage missed, and it is the thing I would put at the top of the internal note. On this corridor the product is not the country pair, it is the corridor plus the origin station plus the destination node, and the same destination can be three products. Moscow from Shenyang through Manzhouli is one thing. Moscow from Zhengzhou through Erenhot is another. Moscow from Chengdu through Alashankou is a third, and the three do not share a transit time, a cut-off, a border queue or a winter risk profile. A rate quoted as China to Moscow is therefore not a rate, it is an average of three products with different variances, and averages hide exactly the variance that ruins a delivery date. When someone sends you a China-Europe rail quote with no corridor on it, you have been sent a marketing document.

The 120 kilometres an hour deserves a line too, because people read it as speed and it is not. It is a timetable class. Freight on this corridor does not average 120; it averages whatever the slowest segment allows, and the slowest segment is not the running line, it is the border. Two crossings in a journey, each with its own queue, its own documentation and its own weekend, and the timetable is written around the assumption that both behave. When one does not, the train does not slow down, it waits, and a wait at a border costs a full day and cannot be recovered by running faster afterwards because the path ahead is already allocated to somebody else. That is why the arrival spread on this corridor is wider than the transit time suggests, and why the operators who publish a range are more honest than the ones who publish a number.

Winter is the other timetable assumption nobody prices until it fails. The northern corridor runs through a climate where equipment behaves differently, where a brake test takes longer, where a delay outdoors compounds, and where the same published transit quietly becomes a different product for eight weeks. The timetable does not print a winter version. It prints one number and lets winter be your problem, which is why the shippers who have run this road for years build their fourth-quarter commitments on a different transit than their second-quarter ones and simply carry the difference. If your contract has one transit figure in it, it has a summer figure and a dispute in it.

Layered, the corridor looks different depending on where you stand. A large electronics manufacturer with a dedicated block train and a twelve-month allocation is buying a schedule and treats the 96 lines as irrelevant, because its train is one of them. A mid-size shipper moving eight FEU a month is buying a slot from a consolidator and should care about nothing except that consolidator's fill rate on the specific line, which is knowable and rarely asked. A small shipper with two pallets is buying groupage and is really buying a promise about which train its pallet gets on, and that promise is the weakest link in the whole chain because a pallet can always be rolled to next week at almost no cost to anyone except the owner of the pallet. Three shippers, one corridor, three completely different risks, and one press release written for none of them.

The May record is also worth a sceptical look, because records have seasons. A monthly high in May is a monthly high in the window when European buyers are restocking for the second half and Chinese exporters are pushing out the orders that missed the spring. Annualising it into 23,580 is arithmetic, not a forecast, and the release does not say what the other eleven months ran. If the rest of the year is 1,500 and May is 1,965, the trend is flat with a spike and the 140,000 keeps climbing on history rather than momentum. If the second half is running above May, the number is a floor. The release gives you one month and asks you to infer a year, and the honest answer is that you cannot.

There is a structural fact about this corridor that never appears in a milestone release and always appears in a rate sheet: the two directions do not carry the same thing. Westbound trains run loaded with goods Europe wants; eastbound trains have less to bring back, and the return leg is where the corridor's economics are decided. That imbalance is why a westbound slot and an eastbound slot are not the same price even when the distance is identical, and it is why the operators who control the return leg can afford to be aggressive on the outbound. If you ship westbound only, you are a convenient customer and you will be priced like one. If you can offer return volume, even intermittently, you have something to trade, and the trade is worth more than the discount you would get from shopping the rate.

The buying chain is worth knowing before you argue about a rate, because the rate is set four steps away from you. The path belongs to the operator. The platform company in the city sells the block train. The consolidator sells you the slot and takes the fill risk. You buy the slot and own the arrival risk. Between those four parties the only one with an incentive to overpromise is the one closest to you, because the consolidator's margin is the difference between what it paid for the train and what it collected from shippers like you. That is not a criticism, it is a structure, and structures are predictable. Ask the consolidator how many boxes it is short of a full train this week, and you will learn more about your transit than from any published timetable.

A word on the box itself. On this corridor a great deal of the equipment is carrier or platform owned and pooled, which means the container you are promised is not necessarily the container that shows up, and the substitution can change the tare, the internal height and whether your pallets fit two across. Shippers who load to the millimetre get burned by this once and then start asking for the box number a week before cut-off. Ask for the box number. It is a small request and it is the difference between a plan and a hope.

The counterfactual I would argue with is the one the milestone invites, which is that a bigger cumulative number means a better road. It does not, and the two can move in opposite directions. A corridor can add trains every month while adding congestion at the same crossings, and the 140,000th train can be the most delayed train in the sequence. The test is not the count, it is the spread between the published transit and the delivered transit on your own shipments, tracked monthly. If that spread is widening while the cumulative number climbs, the road is getting busier and not better, and you should be pricing the difference rather than celebrating the milestone.

The other divergence worth naming is between the three corridors and their border risk. They are not interchangeable substitutes, because each has a single dominant crossing and each crossing has its own queue, its own inspection regime and its own winter. A shipper who routes everything through one corridor because the rate is two hundred dollars cheaper has bought a single point of failure and called it a saving. The professional version is to hold a primary and a secondary corridor per destination and to know, in advance, what the switch costs in transit days, because you will need that number on a Friday afternoon and you will not have time to build it then.

What it comes down to for a booking desk this week is unromantic. Take every live rail quote you have and add three fields to it: corridor, origin station, destination node. Any quote that cannot fill all three is not a quote. Then ask the consolidator for its fill rate on your specific line over the last eight weeks, and if it will not give you a number, assume the worst and add a week to the promise. Then check whether your destination is a rail node or a truck leg, because that single fact decides who is liable when the box arrives at the ramp instead of the door.

There is one more number that will matter more than all of them by the end of the decade, and it is not in the release: the number of paths. Ninety-six scheduled lines is a fixed supply in a system where demand is growing, and fixed supply with growing demand is a queue. Every additional train on this corridor is competing for a path that was allocated months ago, and the reason the release leads with a cumulative count is that a cumulative count never has to mention the queue. Watch the line count, not the train count. Ninety-six going to a hundred is a capacity story. Ninety-six staying flat while the cumulative number doubles is a pricing story, and it is your story.

So, does this road still run. It runs, and it runs better than it did when the first train left, and it is still the only land bridge between two manufacturing systems that mostly do not like each other's weather. What has changed is that the road is now a network with nodes, corridors and queues, and a network rewards the shipper who knows which node it is standing on and punishes the one who bought a country pair. The 140,000th train left Shenyang on a schedule that somebody had reserved months earlier. That is the whole lesson. The train was not the achievement. The reservation was.

↑ Back to the key points

— By Ömer Kaya