How Did TVs Get Cheaper and Better at the Same Time?

There is a line in the note on my front page, written before I had answered any of it, listing the things I keep wondering about. One of them is how TVs got cheaper and better at the same time.

I have been carrying that question around for long enough that it went on the website as an example of a question I hadn’t answered. Then last week I bought a television, and the receipt made the question impossible to keep ignoring.

I paid ₹1,65,581 for a 65-inch OLED.

In 2018, a 55-inch OLED — ten inches smaller, eight years older, meaningfully worse in every measurable way — was listed at ₹2,04,990.

Bigger, newer, better, and about forty thousand rupees cheaper. Not adjusted for anything. Just the two numbers, side by side.

Nothing else I own has done this. Not the laptop, not the phone, not the fridge, and certainly not anything I eat. So what exactly happened to televisions?

First, the size of it

The like-for-like comparison is the one worth doing, so: a 55-inch LG OLED in 2018 against a 55-inch LG OLED in 2026. Same company, same size, same panel technology, eight years apart.

₹2,04,990 then. ₹1,40,490 now. That is a 31% fall in plain rupees, at a time when nearly every other price in India went the other way.

I have to flag that 2018 number, because everything here rests on it and I could not confirm it at source. The evidence pulls in both directions, so here is all of it.

Upward: LG’s launch coverage from July 2018 put its Indian 2018 OLED range at “about ₹2.75 lakhs onwards”, and the B8 was the cheapest model in that range.

Downward, and more awkwardly for me: Indian AV forums from late 2018 have people quoting the 55-inch B8 at ₹1.3 to ₹1.5 lakh. Indian television prices carry a wide gap between the printed MRP and what a shop will actually take, and those threads are buyers comparing dealer quotes.

That exposes a genuine flaw in my comparison, and it is mine rather than the sources’. The ₹1,40,490 I am using for 2026 is a street price, the cheapest listing I could find — and my own C6 came in ₹29,409 under its list. Setting a 2018 list price against a 2026 street price tilts the result before any arithmetic happens.

So treat the headline as the shape of the thing and not its exact size. Like for like, the fall in plain rupees is smaller than 31%, and if those forum quotes are representative it may be close to nothing. What survives is the part that does the real work: a television that merely held its nominal price across eight years of 41% inflation has still become about a third cheaper in real terms — and this one did better than hold.

Which is why the inflation adjustment matters more than it usually does. India’s consumer price index — the general, combined, all-India series — sat at 140 for 2018-19 and reached 198.0 in December 2025. That is 41% cumulative inflation.

So the ₹2,04,990 that a 55-inch OLED cost in 2018 is about ₹2,90,000 in late-2025 money.

OLED prices, in late-2025 rupees

2018 · 55-inch ₹2,90,000

2026 · 55-inch ₹1,40,490

2026 · 65-inch ₹1,65,581

Half price for the same thing. Rather less than half, for something considerably better.

The obvious answer is the wrong one

Everybody reaches for the same explanation, and I did too: Moore’s law. Chips get cheaper every year, televisions contain chips, therefore televisions get cheaper. Case closed.

It isn’t, and understanding why not is genuinely the whole answer.

Moore’s law is an observation about shrinking. Roughly every two years, you can pack about twice as many transistors into the same piece of silicon. The economics follow from that one fact: the thing you are selling gets physically smaller for the same capability, so you fit more of them on a wafer, so each one costs less.

Now try to apply that to a display.

A 65-inch panel has to be 65 inches. That is not an implementation detail you can optimise away — it is the entire product. Nobody has ever wanted a smaller television for the same money. The one lever that makes semiconductors cheaper is the one lever a display manufacturer is forbidden from pulling.

If anything, the display industry has spent thirty years running Moore’s law in reverse. It has been trying to make the product bigger every year, which under any normal manufacturing logic should have made it steadily more expensive.

It got cheaper anyway. So something else was going on.

What actually happened: they made the glass bigger

Panels are not manufactured one at a time. They are cut, like dress patterns, from an enormous sheet of glass called the mother substrate — and the entire economics of the industry lives in that word cut.

A fab handles one sheet at a time. Coating it, patterning it, and moving it through the line costs roughly the same whether you slice it into many small panels or a few large ones. So the question that decides your cost per television is brutally simple: how many panels can you get off one sheet without wasting glass at the edges?

That is why display fabs are described by “generation” — the generation is the sheet size. And the jump that produced the television you can now afford is Gen 10.5, whose sheets measure 2940 × 3370 mm. Very nearly ten square metres of glass, moved through the factory as a single object.

That size was not chosen for elegance. It was chosen because of what divides into it cleanly: eight 65-inch panels, or six 75-inch panels, with very little glass left over.

Work it through and the fit is almost uncomfortably neat. A 65-inch 16:9 panel measures 1439 × 809 mm. Two of them side by side come to 2878 mm across a sheet that is 2940 mm wide. Four of them stacked come to 3238 mm down a sheet that is 3370 mm tall. Eight panels, 94% of the glass used, and about six centimetres of margin in one direction.

The 75-inch cut is the same trick performed sideways: those panels are turned through ninety degrees, three across and two down, and land on 94% as well.

Now do it for a size that doesn’t fit. A 55-inch panel is 1218 × 685 mm. Turn those sideways and four sit across the sheet, two deep — eight panels again, but covering only 67% of the glass. A third of the sheet becomes offcut.

Same factory, same sheet, same process, and a third of the raw material turns into nothing, purely because of the number on the box.

That is the whole mechanism, and it cuts both ways: it is also why nobody makes 55-inch panels on a Gen 10.5 line. They are cut on smaller, older sheets that 55 inches divides into neatly. Every fab generation has its own handful of sizes that fit, and the sizes that fit the newest and largest sheets are the ones that get cheap fastest. Over the eight years this post is about, those were 65 and 75 inches — which is exactly why the 65-inch television stopped being an extravagance while the 55-inch one merely got somewhat less expensive.

Read that again, because it is the least intuitive fact in this post: the reason 65-inch televisions became affordable is that 65 inches happens to divide neatly into a sheet of glass somebody specified in a factory plan. Not because of a breakthrough in pixels. Because of arithmetic on a rectangle.

It also explains something you have probably noticed without thinking about it — why television sizes come in strange clusters, why 65 and 75 are everywhere and the sizes between them are oddly expensive. Those are the cuts. Everything else wastes glass.

And these factories are not small bets. A single Gen 10.5 fab costs somewhere between three and seven billion dollars, which is why there are so few of them, and why the ones that exist are overwhelmingly in China, run by BOE and TCL CSOT. That concentration is the second half of the price story: enormous fixed costs, enormous output, and a fierce need to keep the line running — which pushes panel prices down and keeps them there.

The law that actually applies

If Moore’s law is the wrong model, there is a right one, and it is less famous than it deserves to be.

Wright’s law says that the cost of making something falls by a roughly constant percentage every time cumulative production doubles. Not per year — per doubling. It is a statement about accumulated practice: yields improve, defects fall, handling gets faster, the thousandth sheet goes through better than the first.

Yield is the part worth dwelling on, because for OLED it was brutal. A defect in the wrong place doesn’t damage a panel, it destroys it — and early on, a significant share of what came off the line was unsellable. Every percentage point of yield recovered afterwards is almost pure cost reduction, because you already paid to make the thing.

That is the honest shape of the answer. Televisions didn’t get cheaper because of one invention. They got cheaper because an industry made the same object tens of millions of times and got progressively less bad at it, on ever larger sheets of glass, in factories too expensive to ever be allowed to idle.

And then there is the other half

Everything above explains how a television got cheaper to make. It does not fully explain how it got cheaper to buy, because those are different questions, and the gap between them is where this stops being a manufacturing story.

At some point the television stopped being the product.

The clearest public evidence comes from Vizio, a US manufacturer, because it is one of the few that had to publish the numbers. In the third quarter of 2021, its hardware business — actual televisions, sold to actual people — produced $25.6 million in gross profit. Its Platform Plus division, which is advertising and viewer data, produced $57.3 million.

More than double the profit, from the software, on a fraction of the revenue.

Gross profit, Q3 2021 · Vizio

Selling televisions $25.6M

Advertising and viewer data $57.3M

The mechanism is automatic content recognition. The television samples what is on its own screen and matches it against a fingerprint database — so it knows what you are watching whether it arrived by app, cable box, console or HDMI. That is not a side feature; Vizio bought a company specifically for it in 2015, and the ad business was built on top.

So part of the reason a modern television is cheap is that it is no longer only a television. It is a screen with a business model attached, and some of the price you didn’t pay at the shop is being collected afterwards, in a currency that doesn’t appear on the receipt.

I should be careful here, because Vizio is an American company and my television is an LG bought in India, and I have not found an equivalent public breakdown for LG. What I can tell you is what I can see: I use the set’s own home screen and its own apps, and that home screen has advertising on it. The direction of travel is not subtle.

Which brings it back to Moore’s law

Here is the part I find genuinely satisfying, and it took me the whole post to get to.

Moore’s law does apply to a modern television. Just not to the picture.

It applies to the computer inside it — the system-on-chip running the interface, decoding the video, driving the apps. That part is silicon, and silicon behaves exactly as Moore described: it gets faster and denser and more capable with each generation, and the software written for it grows to fill whatever it is given.

So a television is really two products in one cabinet, obeying two different laws. The panel follows the economics of glass — it gets cheaper by getting bigger, and once made, it simply keeps working. The computer follows the economics of silicon — it gets better every year, and it ages.

I have a 2018 OLED in my living room that demonstrates the split precisely. Its picture is, as far as I can see, perfect. Its motherboard has been replaced once already.

That is a different post. But it is the same fact.

So: how did televisions get cheaper and better at the same time? Because the expensive part stopped being a technology problem and became a logistics one — how neatly a rectangle divides — and because the part that is a technology problem got quietly repurposed into a business that doesn’t need you to pay for it at the till.

The picture on your wall got cheaper because a factory in China got very good at cutting glass. The computer behind it got cheaper because it found somebody else to bill.

What I checked, and how sure I am

Every claim in this post that could be wrong, with where it came from and how much weight it will take. The arithmetic at the top is the part I would most like someone to check, because it rests on a single price I could not confirm at source.