A newly bought laptop costs a little more than last year, and the prices of several flagship phones have crept up as well. Most people's first instinct is to argue at the counter, but the difference is not written on the phone's price tag. It shows up much earlier, and far away, in a batch of machine rooms that are competing with artificial intelligence for production capacity.
According to an article published by the Economic Daily on November 1, "When AI 'Swallows' Memory", the main force pushing end prices upward is a scramble for memory chips driven by artificial intelligence. The article offers an analogy: the price changes you and I feel are only the small cost that lands at the very end of that scramble.
One thing should be set straight first. Every number mentioned below is an industry purchase price, a contract price or a share of cost, not the retail sticker on any particular phone. The two are not the same account, and they cannot be used to claim that a given phone must have risen by so much.
The Extra Charge at the Counter
The article opens in a very everyday way: a newly bought laptop has become more expensive, and phone makers' flagship models have also gone up. It does not say by how many yuan any specific model rose; it simply sets out the phenomenon and then follows it upstream to look for a cause.
Following it upstream brings you to a group of memory makers and a group of data centers being built out. The article's point is that on one side production is bustling, while on the other the consumer end cannot get enough chips; supply is clearly expanding, yet the gap has not been filled.
For an ordinary reader, what matters most in this passage is not who gained or who lost, but a direction: the bill travels down the supply chain layer by layer and finally comes to rest at the checkout counter. Where exactly it stops depends on how each maker chooses to absorb it.
Nor should the account be rushed into a single device. The article is about a cost change faced by consumer electronics as a whole; when it comes down to a particular brand or model, configuration, stocking and pricing each add their own differences.
The Share of the Bill Has Changed
The first set of figures the article gives is the share of memory chips in the total material cost of a smartphone: from roughly 10 to 15 percent in the past to above 30 percent. In other words, to build the same phone, the part that simply remembers data now takes a heavier slice.
The report also notes that for some flagship models with high storage configurations, this share is higher still. This article does not copy down finer numbers, and keeps only a qualitative reminder: the more a model piles on high-capacity storage, the more sensitive it is to this round of cost changes.
A share is a proportion, not an amount of money. It can grow because the memory part got more expensive, or because something else got cheaper; a lone percentage cannot say how much more was paid for one phone. With numbers like these, ask first what the denominator is, then what the numerator is.
Putting the proportion back into the whole phone makes it easier to grasp. In a phone, the screen, battery, camera and casing each have their own cost, and memory is only one box. That box moving from a little over a tenth to more than three tenths means that, with the other boxes unchanged, it has quietly worked its way to the front row.
The phrase material cost refers to the money for the parts needed to build a phone, leaving out research, marketing and distribution. What the article discusses is the memory share within that one box, not how much memory counts for in the retail price. The two statements sound alike but do not add up to the same account.
Two Price Tags for Two Chips
More concrete are the purchase prices. According to the article, the purchase price of a single 12GB phone memory chip rose from about 200 yuan to nearly 600 yuan, while a 1TB flash chip went from just over 200 yuan to about 600 yuan. Both figures carry a currency, and both are approximate, not counter quotes.
Read those two numbers aloud to someone at home and it comes out roughly like this: the part of a phone that keeps short-term notes, and the part that stores photos for the long term, both cost noticeably more to buy in. The article uses them to show how cost presses down from upstream to the terminal, not to work out the total bill for anyone's single device.
The caveat bears repeating. These are purchase prices between chip makers and device makers, with stocking, contracts, configuration and pricing in between. Seeing 200 and 600, do not casually take them as the difference you would pay in a shop; the distance between the two is far greater than the figures suggest.
Nor should the two price tags be added together. One concerns short-term memory, the other long-term storage; the capacity units differ and the uses differ, and their ending up at similar amounts is coincidence. Collapsing them into a line about memory becoming twice as dear sounds tidy, but muddles two things into one.
Contract Prices Moved First
The article gives another set of figures: in the first half of this year alone, contract prices for general-purpose memory nearly doubled, and flash memory rose by more than half. Note the words contract prices. They refer to the prices at which large customers take delivery under agreements, not the stickers on a retail counter.
This is exactly the place most easily misread. Set the rise in contract prices beside the rise in the purchase prices above and it is easy to blurt out that memory went up by more than double; yet the two sets differ in timing, in who they cover and in how they are calculated, and placing them together only muddles the account.
The original Economic Daily article and a November 1 repost by China News Service use the same measure for these figures. This article follows the original in marking sources and approximate values, and does not convert any set into what you should pay extra, because the original does not make that calculation either.
Contract prices move first and retail prices later, with a stocking cycle in between. Large customers negotiate prices by the quarter or on longer cycles, while the sticker on a counter depends on when a maker adjusts it. The two ends run on different rhythms, which is why one round of increases looks different in different places.
Who Ends Up Holding the Bill
The article also uses rather heavy wording to describe how tight things are just now, and this piece will not pile on. An exaggerated metaphor makes a report readable, but it should not be taken as something a company said standing at a press conference, still less as a judgment about what comes next.
What is really worth taking away is the shape of that chain: data centers first compete away the capacity, cost passes to device makers, and the device makers then decide whether to raise prices or adjust configurations. What consumers see at the far end is only the last short segment.
So the bill is written in the machine room but handed over at the counter. Reading it, first tell apart purchase price, contract price and retail price, then look at who said it and when. The same word rising means different things when it lands in different boxes.
As for whether prices will keep rising, or whether now is the time to replace a phone, the article gives no answer and this piece will not write one either. All that can be certain is the chain that carried the cost over, and the few approximate price tags attached to it. Reading those clearly is more useful than rushing to a conclusion.
Disclaimer: This article is based on publicly available information and is for reference only. It does not constitute investment advice.