Los Angeles Sets 118-Year Record With 1,042,652 TEU as Empties Hit High

The Analysis

A million and change containers in one month at one port is a number you should never read on its own. Los Angeles moved 1,042,652 TEU in September, its busiest month in 118 years, and the first thing I did was look for the empties. There were 371,206 of them. Better than a third of everything that port lifted last month was air.

Here are the published figures before I start putting my own spin on them. The Port of Los Angeles handled 1,042,652 TEU in September, the busiest month in its 118-year history and roughly 20,000 TEU clear of the record set in July 2025. Loaded imports came in at 545,696 TEU, up 19 percent year on year. Loaded exports were 125,751 TEU, up 10 percent. Empties set a monthly record at about 371,206 TEU. Nearly 3 million TEU moved in the third quarter, and close to 4 million across June to September. Long Beach, right next door, set its own record at 997,177 TEU, with imports up 31.3 percent to 509,390 and empties up 22.1 percent. That is the news.

The number I cannot get past is the empties, so let me take it apart. Loaded imports plus loaded exports is 545,696 plus 125,751, which is 671,447. Take that away from 1,042,652 and you get 371,205, which is the empty figure to within rounding. So 371,206 divided by 1,042,652 is 35.6 percent. Better than one lift in three at Los Angeles in September was an empty box. A third of the crane time, a third of the gate moves, a third of the chassis hours went to moving air, and somebody is billing for all of it.

That is not a footnote about trade statistics, it is the machinery behind your drayage quote. Assume a drayage carrier running a fleet of thirty trucks in the basin. If roughly 35 percent of the moves in that market are empty repositioning, then ten or eleven of those trucks are hauling nothing on any given day and only about nineteen are doing revenue loaded work. Assume a truck can manage four loaded turns in a normal day and three when the terminals are tight. The fixed cost of the truck has not changed, so going from four turns to three means every move has to carry about a third more of that fixed cost.

Four divided by three is roughly a 33 percent uplift. That is the mechanism behind a rate increase that has nothing to do with fuel, wages or the port itself. When you next ask your drayage provider for a rate, ask them what turns they are actually achieving this month. That answer tells you more than the number on the quote.

Now the imbalance in the other direction, which is the part that decides how long the pressure lasts. 545,696 loaded imports against 125,751 loaded exports is a ratio of about 4.34 to one. For every loaded box that leaves Los Angeles, roughly 4.3 arrive. The surplus is 545,696 minus 125,751, which is 419,945 TEU of boxes that came in full and must go back empty. But only 371,206 empties actually moved. The gap between those two is 48,739 TEU. That is not a rounding difference or a reporting lag. It is forty-eight thousand boxes sitting somewhere: stacked in a terminal, parked on a chassis, waiting in an off-dock empty yard, or queued for a vessel with the space to take them. Forty-eight thousand boxes of slack is what the machine is absorbing right now, and it is the buffer standing between normal operations and the first morning you cannot get an appointment.

Exporters should read the same table differently, because the record empties number cuts both ways. If you are shipping out of Southern California, 371,206 empties leaving the basin means the box supply is being swept back to Asia at record speed. Assume your empty pickup lead time stretches from next day to two days. Assume a two-day slip on a booking costs you a vessel on a seven-day rotation, so seven days. Assume your customer's contract carries a late delivery penalty of half a percent a week on a 60,000 dollar container, which is 300 dollars a week. At ten containers a month that is 3,000 dollars a month of exposure created by a two-day slip at the street end. The defence is procedural and cheap: book the empty pickup and the vessel space on the same day, and hold a two-day buffer in your booking calendar through January.

Running the month as a daily rate makes the congestion argument concrete. 1,042,652 over thirty days is about 34,755 TEU a day at Los Angeles. 997,177 over thirty days is about 33,239 a day at Long Beach. Together the San Pedro Bay complex moved roughly 68,000 containers a day in September. The record itself tells you something smaller and more useful: 20,000 TEU above the previous peak, spread across a month, is about 667 extra moves a day against the 1,022,652 TEU that the July 2025 record implies. An extra 667 moves a day sounds trivial next to 68,000, which is roughly 2 percent. But two percent does not arrive spread evenly across every terminal and every hour. It lands on particular vessel strings, particular terminal gates and particular days, and on those days the queue is not two percent longer, it is the whole day longer.

Where do you sit inside a number that big? Assume you move forty containers a month through the two ports. Across a full year that is 480 boxes, against the roughly 2,039,829 TEU the two of them moved in September alone. You are not a rounding error to your own business, but you are invisible to the port, and that is the point. Nobody at the terminal is going to design an appointment system around your forty boxes. What you can do is stop behaving like a customer who expects the system to bend, and start behaving like one who books against it. That means appointments made at the time of booking rather than at arrival, chassis confirmed with the drayage order rather than at the gate, and free time tracked box by box rather than month by month.

Let me price what that means on a normal import book, because the daily charges are not where the money is. Assume you import forty 40-foot containers a month through Los Angeles. Assume four free days are included at the terminal. Assume one container in three busts free time in a month like September, so about thirteen boxes. Assume those run two days over at 150 dollars a day in combined demurrage and detention. That is 13 times 2 times 150, so 3,900 dollars a month, about 46,800 dollars a year.

Annoying but survivable. Now assume the surge costs you one missed appointment that pushes a whole order past a retail delivery window, and the chargeback is 3 percent of invoice. On a twenty-container order averaging 60,000 dollars, that is 1.2 million dollars of goods and a 36,000 dollar chargeback. One missed window costs roughly nine times the entire year of per diem. That is the real shape of it. The daily charges are noise; the one order that misses is the loss.

Pre-pulling looks like the obvious answer and usually is not. Assume moving a box to an off-dock yard and back costs 285 dollars: dray it out, store it, redeliver it. Assume the per diem you avoid by doing that is three days at 150 dollars, which is 450 dollars. The break-even is 285 divided by 450, about 63 percent. So pre-pull only when you are more than roughly two-thirds sure the box will bust free time. Pre-pulling everything as insurance costs 40 times 285, which is 11,400 dollars a month, against an expected exposure of 3,900 dollars. That is a bad trade, and in a record month it is the one I watch people make most often. Pre-pull the thirteen boxes you know about, not the forty you are worried about.

Detention and demurrage invoices are where most of the friction surfaces, and much of it is avoidable. Assume you are paying 150 dollars a day and you pick up an invoice for four days on twenty boxes in a month, which is 12,000 dollars. Before you pay it, check two things. Check the start clock against the actual gate-out or discharge timestamp, because invoices come out of the carrier's system and that system does not know the terminal was closed, or that the box was never made available for release. Then check who was holding the box. If the delay sat on the rail ramp or inside the terminal, that is not your detention to pay, and it is worth a written dispute with the timestamps attached. I have recovered a decent share of these invoices for clients just by asking for the underlying event log.

There is a planning mistake buried in the quarter figures too, and it is worth being blunt about. Nearly 3 million TEU in the third quarter works out at about 1 million a month. Close to 4 million across June to September works out at about 1 million a month as well. The four-month number is not a spike sitting on top of a normal quarter. It is the same run rate held for four months, which means it is a plateau, not a peak. Assume your own fourth-quarter volumes land 10 to 15 percent above the third quarter, which is normal for importers building into year-end. Assume you booked your Q4 drayage and warehousing capacity off a Q3 baseline. Then you are 10 to 15 percent short at exactly the moment the port is running about 2 percent above the highest level in its history. The port's two percent and your fifteen percent are different problems, and only one of them is yours to fix.

Your own dock is where the arithmetic gets surprising. Assume you run forty containers a month through Los Angeles and import growth of 19 percent holds for another year, so forty-eight a month. Assume each container takes three hours of door time to unload, so that is 144 door-hours a month. Assume twenty-two working days and two dock doors at eight hours each, which is 352 door-hours of capacity. Utilization is 41 percent, which reads as comfortable. Now assume the surge means 60 percent of the month's containers arrive inside a ten-day window: that is twenty-nine containers in ten days, or 87 door-hours against the 160 available in those ten days. Still 54 percent. So your doors are not the constraint. Your labour plan is, because twenty-nine containers in ten days is 2.9 a day against a normal 2.2, and casual labour does not get booked that way. Get the extra crew in writing, by name and shift, before you need them.

Diversion is the option everybody asks about, and it has a price that is easy to calculate and easy to get wrong. Assume moving a quarter of your volume, ten boxes out of forty, to Oakland, Seattle, Houston or an East Coast routing costs 450 dollars more per box in inland freight. That is 4,500 dollars a month. In exchange you take a quarter of your exposure off San Pedro Bay. Whether that is worth it depends entirely on where your goods actually go. If your distribution centre is in the Inland Empire, 450 dollars a box is a real premium and you should not pay it to solve a problem that is costing you 3,900 dollars. If your freight is going to Chicago or Dallas, the inland arithmetic can run the other way and the diversification is close to free. Run it by delivery geography, not by the congestion headlines.

The buffer stock question follows straight out of that 19 percent import growth figure. Assume you run forty containers a month and hold five weeks of cover, which on a 4.33-week month is about forty-six boxes on hand. Assume your volumes track the port and grow 19 percent over the coming year, so forty-eight a month. Five weeks of cover on the bigger number is about fifty-five boxes. The gap is nine boxes, roughly 1.62 million dollars of inventory at 180,000 dollars a box, and about 27,000 dollars a month in carrying cost at 20 percent a year. Whether you buy it depends on where your demand uncertainty sits, but the figure is far smaller than most people expect, and carrying nine extra boxes is cheaper than a stockout in November.

What you promise your own customers is the last piece, and it is where importers burn the most goodwill. Assume your quoted transit time from origin to your distribution centre is thirty-five days. Assume September conditions add three days at the port end. If your sales team is still quoting thirty-five, every order booked in October lands late against the promise even though nothing went wrong operationally. Put the buffer into the quote instead of absorbing it silently, and give a date range rather than a single day. A customer told forty to forty-four days who gets forty-two is happy. A customer told thirty-five who gets thirty-eight is on the phone.

One more piece of arithmetic is worth doing once and then reusing. Assume your average dwell from discharge to gate-out is four days in a normal month. Assume September pushed it to six. Those two extra days across forty boxes at 150 dollars a day is 12,000 dollars a month. Now assume half of that extra dwell is not the terminal's doing but yours: the appointment was open and the truck did not show, or the paperwork was not ready at release. That is 6,000 dollars a month you can fix without negotiating with anybody. The way to find it is to record five timestamps per container: discharge, availability, customs release, appointment, gate-out. Four gaps, and each gap has an owner. Most importers record two of them and argue about the rest.

The record itself has one more thing to say, and it is about what comes next. 20,000 TEU above the previous record is about 2 percent, and the ports did not break under it, which tells you the physical system had the capacity. It also tells you there is no spare capacity left to absorb the next 2 percent. Assume October and November run at the same level. The gate hours, the chassis fleet and the appointment inventory are all fully committed already. In that state the thing that gives way is not the crane, it is the schedule, and the thing that lands in your P&L is not demurrage, it is the order you could not promise.

The traps in a month like this are consistent enough to list. Free time is quoted in different units by different parties: the terminal counts calendar days from discharge for demurrage, the carrier counts days from gate-out for detention, and the chassis provider is running a third clock. A box sitting on a chassis is racking up two bills at once. Appointment availability tightens first at the terminals where your carrier holds the fewest slots, which is not always the terminal where the port is busiest. And the empty record is double-edged, because the sweep that clears boxes off the street is the same sweep that makes an empty pickup hard to get on short notice.

So here is what I would do with the next ninety days. Inside two weeks, pull your last ninety days of entries at both ports and count how many boxes went past free time; if it is more than 10 percent, price pre-pulling at 285 dollars a box against the 150 dollars a day you actually paid. By the end of October, name the boxes that cannot absorb a two-day slip and book drayage and chassis for those alone. From November, hold a two-day buffer in every export booking and reserve empty pickup the same day you book the vessel. Before the end of November, have your peak labour in writing and have your diversion economics worked out by destination. And in January 2027, compare your own dwell and per diem against the September baseline and reset your buffer for the next peak.

I have been reading these monthly numbers long enough to know that a record month is almost never about the record itself. 1,042,652 TEU at Los Angeles, 997,177 next door at Long Beach, roughly 68,000 boxes a day through the two of them, and 371,206 of the Los Angeles lifts were empty. What that combination tells me is that the system is running at a level it has never run at before, with about a third of the work earning nobody anything, and with roughly 48,000 boxes of slack left in the machine. If you are importing through San Pedro Bay this quarter, the job is to know which of your boxes are the ones that cannot take a two-day slip, and to have booked the capacity for those before everyone else does. That is the whole of it, and it is the same job every peak. I will come back with the October numbers when they land.

↑ Back to the key points

— By Leo