When you order a book on Amazon and it arrives in two days, you probably assume it was sitting in a warehouse somewhere. But what if it was printed that morning, on demand, just for you? Daniel's been thinking about just-in-time logistics, and he wants to know what it actually means in the real world. He points to print-on-demand as an example, and asks whether the benefit of avoiding unnecessary inventory comes at the cost of losing economies of scale. Is that a fair trade-off? And what does a supply chain look like before and after making that switch?
So let's start by pinning down what JIT actually means, because it's not just about having less stuff in warehouses. Just-in-time is the practice of producing or delivering goods exactly when they're needed, not before. The contrast is "just-in-case" — the traditional model where you build up inventory buffers because you don't trust your supply chain or your demand forecasts. Under JIT, a component arrives hours before it hits the assembly line. A book gets printed when the order comes in, not when a publisher guesses it might sell.
And the gut reaction is always the same — "isn't that terrifying? What if something goes wrong?" Which is a fair question, but it misses what JIT is actually optimizing for. It's not a cost-cutting gimmick. It's a discipline.
Right. The core insight is that inventory hides problems. If you've got a mountain of buffer stock, you don't notice that your supplier is unreliable, or that your quality control is spotty, or that your demand forecasting is fiction. You just dip into the pile and keep moving. JIT strips away those buffers and forces you to fix the underlying issues. Toyota understood this in the nineteen fifties and sixties when they built the original system. Taiichi Ohno, the engineer behind it, wasn't trying to save money on warehouse space. He was trying to expose waste — what they called "muda" — everywhere in the production process.
So the inventory reduction was almost a side effect of a deeper philosophy.
And the numbers were dramatic. By the nineteen eighties, Toyota had reduced inventory holding costs by forty to sixty percent compared to American automakers. Their Georgetown, Kentucky plant operated with about four hours of inventory on site. Comparable GM plants at the time held two to four weeks of stock. Four hours versus four weeks — that's not a marginal tweak, that's a fundamentally different way of thinking about production.
Four hours of inventory. So if a single truck gets stuck in traffic, the line stops.
And that's the point. The fragility is the feature, not the bug. Toyota built a system called the andon cord — any worker on the line could pull it and stop the entire production line if they spotted a defect. In a traditional factory, stopping the line is a catastrophe. At Toyota, it was a quality forcing function. If you've got four weeks of buffer inventory, a defective batch of door handles just gets swapped out from the pile and nobody investigates why it happened. When you've got four hours of inventory, a bad batch stops production immediately, and that gets everyone's attention. The problem gets fixed at the source.
So JIT is really a quality management system disguised as an inventory strategy.
That's the part most people miss. It's not about being cheap. It's about being incapable of tolerating defects. And this connects directly to what Daniel's asking about print-on-demand. POD is a pure JIT model applied to publishing. No inventory of finished books exists. A customer clicks "buy," and a copy gets printed, bound, and shipped from the nearest facility. IngramSpark, which is one of the big POD networks, operates more than twenty printing facilities globally, each serving a regional market. No warehousing, no returns, no remaindered books getting pulped.
And that last part is bigger than people realize. Traditional publishers typically write off twenty to thirty percent of printed inventory as unsold. They print five thousand copies hoping to sell four thousand, and the rest get returned and destroyed. POD eliminates that waste entirely. But the per-unit cost is higher. For offset printing, a paperback might cost a dollar fifty to two dollars per unit at scale. POD runs four to six dollars for the same book. So you're paying two to three times more per copy.
Which brings us to the breakeven analysis Daniel's hinting at. The crossover point is roughly three hundred units. Below three hundred copies, print-on-demand is cheaper overall because you're not paying for warehousing, you're not eating the cost of unsold inventory, and you're not tying up capital in boxes of books sitting in a fulfillment center. Above three hundred, offset printing's economies of scale kick in and the per-unit savings outweigh the waste. But that crossover shifts depending on your sell-through rate. If a traditional publisher prints five thousand copies at two dollars each — that's ten thousand dollars — but only sells sixty percent, the effective cost per sold book isn't two dollars. It's more like three thirty-three. At that sell-through rate, POD at five dollars a copy is more expensive per unit but you have zero waste and zero upfront capital. The math gets interesting fast.
So the trade-off isn't "cheap versus expensive." It's "predictable cost with zero waste versus lower per-unit cost with significant upfront risk." And that's the same calculus Toyota was making, just with different numbers. Hold inventory and you might save on production costs, but you're paying for storage, insurance, obsolescence, and the capital itself has an opportunity cost. Eliminate inventory and your per-unit cost rises, but you're not bleeding money on all the hidden carrying costs.
And there's a knock-on effect here that's even more interesting. JIT shifts risk from the buyer to the supplier. In Toyota's system, suppliers had to be geographically close and operationally synchronized. They'd deliver components multiple times per day, sometimes in the exact sequence they'd be installed on the line. That meant the supplier, not Toyota, was holding the buffer stock or maintaining the flexible production capacity to respond instantly. The risk got pushed upstream to smaller, less diversified firms.
Which is the opposite of what resilience theory would recommend.
You're concentrating risk in the part of the chain least able to absorb it. And this became catastrophically visible in twenty eleven when the Tōhoku earthquake hit Japan. Toyota's production lines shut down for weeks — not because their factories were damaged, but because a single supplier of a two-dollar sensor was offline. That sensor was used across multiple vehicle platforms, and there was no buffer stock anywhere in the system. The entire global production network stopped for want of a component that costs less than a sandwich.
A two-dollar sensor. That's the JIT fragility in one sentence. And yet Toyota went right back to JIT after the earthquake, because the system had delivered decades of competitive advantage. They weren't about to abandon it for a once-in-a-generation disruption.
They did adapt eventually, but we'll get to that. Let's talk about the second big knock-on effect, which is that JIT enables mass customization. The classic example is Dell in the nineteen nineties and early two thousands. Dell's build-to-order model let customers configure their own PCs — processor, memory, storage, graphics card — and the machine would be assembled only after the order was placed. Components arrived at the Austin facility within ninety minutes of being needed on the line. That's not just inventory management, that's a completely different business model. Under a traditional just-in-case model, Dell would have had to pre-build every possible configuration and hope customers bought them. With JIT, they could offer thousands of combinations and only build what was actually ordered.
And that model crushed the competition for a solid decade. Compaq and HP were stuck with warehouses full of pre-configured machines that they'd have to discount when the next chip generation launched. Dell had no such problem because they held almost no finished goods inventory.
The third knock-on effect is that JIT creates data-rich supply chains. When every unit is produced in response to a specific order, your demand signals are real-time and granular. You know exactly what's selling, where, and at what margin. That enables predictive analytics, dynamic pricing, and rapid response to market shifts. But it also exposes you to demand volatility in a way that buffered systems aren't. If a TikTok influencer suddenly makes your product go viral, a JIT system can't handle the spike. There's no inventory to absorb the surge.
Which is why we've seen so many companies stumble on viral demand in the last few years. The supply chain is optimized for predictability, not virality.
And that brings us to the post-twenty-twenty hybrid model. The semiconductor shortage from twenty twenty to twenty twenty-three was a wake-up call. Toyota, the company that invented JIT, announced in their twenty twenty-three supply chain resilience report that they were now holding two to four weeks of critical semiconductors — up from two to three days before the crisis. They didn't abandon JIT. They created what analysts are calling "just-in-case plus" — strategic buffers for critical, hard-to-substitute components, while maintaining JIT for everything else.
So even the high priests of JIT are now hedging. That's telling.
It's a recognition that volatility is the norm, not the exception. The twenty twenties taught everyone that lesson. And it's not just semiconductors. We're seeing the same pattern in pharmaceuticals, in food supply chains, in construction materials. Companies are mapping their supply chains to identify single points of failure — the equivalent of that two-dollar sensor — and building targeted buffers around them. It's a more nuanced approach than the binary "JIT versus JIC" debate that dominated business school curricula for decades.
Let's bring this back to print-on-demand for a moment, because it illustrates the hybrid model in a different way. POD is pure JIT for the finished product — the book doesn't exist until it's ordered. But the POD providers themselves hold inventory of raw materials. Paper, ink, binding supplies. They're not ordering a truckload of paper every time someone clicks "buy" on Amazon. So there's a buffer at the component level even in a system that appears bufferless at the finished goods level.
That's a crucial distinction. Every JIT system has buffers somewhere. The question is where you place them and who pays for them. In Toyota's original system, the buffers were at the suppliers. In POD, they're at the raw material stage. In Dell's model, they were at the component suppliers who held inventory near Dell's assembly plants. The buffers don't disappear — they get relocated.
Which means the real question Daniel's asking — "is the trade-off worth it?" — depends entirely on where you sit in the chain. If you're the one holding the buffer, JIT is a cost imposed on you by your customer. If you're the one demanding JIT delivery, it's a cost reduction and a quality improvement. The same system looks very different depending on your bargaining power.
And bargaining power determines whether you can push that buffer upstream. Walmart is famous for this. They've essentially forced their suppliers to hold inventory on Walmart's behalf, while Walmart's own distribution centers run on extremely lean JIT principles. The supplier pays for the warehouse space, the insurance, the working capital. Walmart gets the benefits of JIT without bearing the costs.
So let's make this concrete for someone who's actually trying to decide whether to use POD or traditional printing. Daniel mentioned economies of scale, and that's exactly the right framework. You need to map your specific numbers. What's your per-unit cost at various print runs? What are your holding costs — storage, insurance, the time value of money tied up in inventory? What's your expected sell-through rate, and over what time horizon? And critically, how confident are you in that forecast?
The last point is the one that trips people up. If you're a publisher with a known author and reliable sales history, offset printing probably makes sense above a few thousand copies. Your demand is predictable, your sell-through is high, and the per-unit savings are real. But if you're launching a new title with no track record, POD is essentially an insurance policy. You're paying a higher per-unit cost to eliminate the risk of printing thousands of books that nobody buys.
And the same logic applies far beyond publishing. Any business with physical products faces this calculation. A small apparel brand deciding between made-to-order and bulk manufacturing. A hardware startup choosing between on-demand PCB fabrication and a minimum order quantity from a factory in Shenzhen. The math is the same — map your holding costs against the per-unit premium for flexibility, find the crossover point, and be honest about your demand forecast.
There's a framework I find useful here. JIT works best when two conditions hold. First, demand is reasonably predictable. Not perfectly — no demand forecast is perfect — but within a range you can plan around. Second, supply is reliable. Your suppliers deliver on time, with consistent quality, and you've got alternatives if one fails. If either condition breaks down, you need buffers. And the twenty twenties demonstrated that both conditions can break down simultaneously and globally.
That's the actionable insight, really. The JIT decision isn't a philosophy. It's a breakeven analysis done SKU by SKU. Some of your products should be JIT, some should be buffered, and the mix changes over time as your suppliers change, your demand patterns shift, and the world gets more or less reliable.
And for small-scale creators — which is a big chunk of our audience — POD and on-demand manufacturing are enablers, not compromises. They let you test markets without financial risk. The economies-of-scale argument is only relevant if you can actually sell the volume that makes those economies kick in. If you're selling two hundred copies of a book, the fact that offset printing would be cheaper at five thousand copies is irrelevant. You're not selling five thousand copies. POD is the economically rational choice at your actual scale.
I think that's the point that gets lost in a lot of supply chain discourse. People talk about JIT versus JIC as if they're competing ideologies, when really they're just different points on a cost curve. The right answer depends on your numbers, not your principles.
And the numbers keep shifting. One thing I'm watching closely is how AI-driven demand forecasting changes this calculus. Better predictions reduce the uncertainty that makes buffers necessary. If you can forecast demand with high accuracy, you can run leaner without increasing risk. But there's a paradox here — the better your forecasting, the more tightly you couple your supply chain, and the more catastrophic the failure when the forecast is wrong.
Tighter coupling, higher systemic risk. That's the lesson of the two-dollar sensor. The more optimized the system, the more fragile it becomes to the specific failure modes you didn't anticipate.
And the next frontier makes this even more acute. We're moving toward what you might call "just-in-time everything." Three-D printing of spare parts means you don't need to hold inventory of replacement components — you print them when you need them. On-demand manufacturing platforms let you produce custom goods in batch sizes of one. The tradeoffs we've been discussing — per-unit cost versus flexibility, efficiency versus resilience — are going to become relevant to more and more products.
The airline catering system I've mentioned before is already operating under constraints that make Toyota look relaxed. They're producing thousands of meals that have to arrive at specific aircraft at specific times, with no second chances. A delayed meal means a delayed flight. And the shelf life is measured in hours, not days or weeks. That's JIT with a hard deadline and a perishable product, which is about as unforgiving as supply chains get.
That's a great example. And it highlights something we haven't talked about — JIT in services and perishable goods operates under fundamentally different constraints than JIT in durable goods manufacturing. A book can sit on a shelf for years. A printed circuit board has a shelf life measured in months because of moisture sensitivity. A salad for an airline meal has hours. The cost of getting it wrong scales dramatically as the product becomes more perishable.
Which is why the hybrid model makes so much sense. Buffer the things that are hard to replace or quick to spoil. Run JIT on the things that are predictable and reliable. And constantly re-evaluate because the world changes.
Let's put some concrete numbers around this for the publishing case, because I think it crystallizes the decision framework. Imagine you're printing a paperback. Offset printing at a run of two thousand copies might cost two dollars per unit — four thousand dollars total. POD at five dollars per unit means the same two thousand copies would cost ten thousand dollars. So offset looks like the obvious winner. But now add the carrying costs. Warehousing two thousand books for a year might cost five hundred dollars. Insurance, another hundred. If you only sell twelve hundred copies, you've written off eight hundred books — that's sixteen hundred dollars in wasted production cost. Your effective cost per sold book under offset is four thousand plus six hundred in holding costs, divided by twelve hundred sold — about three eighty-three per book. POD at five dollars per book with zero waste: five dollars flat. Offset still wins on per-unit cost, but the gap is a lot narrower than the sticker price suggested.
And that's assuming you can forecast the twelve hundred sales accurately. If you print two thousand and only sell six hundred, POD is dramatically cheaper. The breakeven sell-through rate for that scenario is around sixty percent. If you're confident you'll sell more than sixty percent of your print run, offset wins. Below that, POD wins. How many first-time authors are confident they'll sell sixty percent of two thousand copies?
Very few, which is why POD has transformed independent publishing. But here's the thing — major publishers use POD too, just not for frontlist bestsellers. They use it for backlist titles, academic monographs, niche non-fiction. Books that sell steadily but slowly, where holding inventory for years makes no economic sense. POD keeps those titles in print indefinitely without warehousing costs. That's a pure win — the book stays available, the publisher makes money on every sale, and nobody's paying to store boxes in a warehouse in Indiana.
So the same technology that enables a self-published author to sell fifty copies also enables Penguin Random House to keep a thirty-year-old backlist title alive. Same economics, completely different scale.
And that's the thread that runs through all of this. JIT isn't one thing. It's a set of tools and tradeoffs that apply differently depending on your position in the supply chain, your demand patterns, your supplier relationships, and your tolerance for risk. The companies that do it well aren't the ones that treat it as a religion. They're the ones that treat it as a spreadsheet.
So for someone listening who's trying to think about this for their own business or creative work, the framework is: map your holding costs, map your per-unit premium for flexibility, find your crossover point, and be brutally honest about your demand forecast. That's it. Everything else is commentary.
And revisit it regularly. The crossover point moves. Supplier reliability changes. Your demand patterns evolve. What made sense two years ago might not make sense today. Toyota learned that the hard way with semiconductors. The good news is you probably don't need to weather a global chip shortage to learn the lesson.
Here's what I'll predict. Within three years, the hybrid model — strategic buffers for critical components, JIT for everything else — will be standard practice in every major manufacturing industry. The pure JIT evangelists and the pure JIC traditionalists will both look outdated. The conversation will shift from "which philosophy is right" to "how do we model the optimal buffer for each SKU."
I'll go a step further. By twenty twenty-eight, AI-driven demand forecasting will be good enough that some companies will start reducing those strategic buffers again, and we'll see a new wave of supply chain fragility emerge. The cycle will repeat, just with better software. The underlying tension between efficiency and resilience doesn't go away — it just gets more precisely measured.
If this episode made you think differently about the book on your nightstand or the parts in your car, share it with someone who works in logistics. They will have opinions, and they will not be shy about them.
Thanks to our producer Hilbert Flumingtop for keeping this operation running, which I'm now realizing probably involves its own just-in-time challenges I don't want to think about.
This has been My Weird Prompts. Find us at my weird prompts dot com, or email the show at show at my weird prompts dot com.
We'll be back soon. Try not to hold excess inventory in the meantime.
