Cheap computers are disappearing

The new semester is approaching. Multiple computer brand stores in Shenzhen have confirmed that HP, Lenovo, ASUS and other brands will collectively raise product prices starting from September 1, with a maximum increase of 15%. Some HP products have already seen a cumulative price hike of over 20% previously, with another price adjustment made on August 28, and a "major surge" expected in September; Lenovo has also notified that sales will be conducted at the new prices; ASUS estimates that the price will rise by at least 500 yuan.

While Huawei has not adjusted its prices, its low-cost entry-level models have been out of stock, and Xiaomi stated that it does not rule out raising prices at any time.

Entry-level models are hit the hardest: a low-end HP laptop originally priced at just over 3,000 yuan is now approaching 6,000 yuan, and the price tags of entry-level models at multiple brand stores are generally no less than 6,000 yuan.

Some parents in Shenzhen complained that their original budget of 10,000 yuan for a high-spec laptop is now far from enough, and the total cost of adding a new mobile phone is approaching 20,000 yuan, so they have to give up buying a tablet to afford the computer.

Dealers in Hangzhou also said that the price of hard drives has risen from more than 400 yuan to 1,000 yuan, and students who plan to assemble their own computers have been discouraged by the high prices.

Some parents reluctantly "bite the bullet" and make purchases before the new semester starts, while others choose thousands of yuan worth of lightweight laptops, and ask their children to cover the overspent part with their lucky money.

The problem lies on the supply side, where some parties have actively cut off the supply of goods

The price of memory chips has indeed risen.

Senior executives of HP publicly stated that the proportion of memory cost in the total material cost of a complete machine has risen from 15% to 18% in the previous quarter to 35% now.

Indeed, the price of storage chips has more than doubled over the past year or more.

But there is a strange point here: if the memory cost rises from 500 yuan to 1,000 yuan, the material cost of a laptop increases by 500 yuan, why does the retail price rise from 3,000 yuan to 6,000 yuan?

Who on earth has taken the extra more than 2,000 yuan in between?

It should be noted that in the global DRAM market, the market for chips used in computer memory sticks, Samsung, SK Hynix, and Micron together account for more than 90% of the market share.

The situation of NAND flash particles is similar, where the market is dominated by a few companies.

This market structure determines that when these manufacturers make the same choice at the same time, downstream computer brands have almost no room for bargaining.

Over the past two years, the artificial intelligence industry has exploded, and the demand for storage chips has surged sharply.

AI servers require a large amount of high-bandwidth memory, namely HBM, whose profit margin is far higher than that of DDR memory used in ordinary computers.

As a result, the three major storage manufacturers made a very rational decision to prioritize production lines for HBM and reduce the production capacity of consumer-grade memory. Public data shows that about 70% of the new production capacity flows to enterprise-level storage and HBM.

This means that the supply of consumer-grade memory is not failing to keep up with the growth of demand, but is actively being cut.

This logic is critical.

The reason for the price increase is not that more people are buying, but that fewer people are selling.

Storage manufacturers have essentially diverted the production capacity originally used for ordinary computers to produce more profitable chips for AI. The remaining supply in the consumer market is naturally pushed up in price due to scarcity.

Moreover, storage manufacturers have no incentive to reverse this situation, because AI orders have high gross profit margins, stable customers, and huge growth space. In contrast, the consumer market has large cyclical fluctuations, high price sensitivity, and thin profits.

After rational calculation, everyone will choose the AI track.

But the problem is that computer brands have played an amplifying role in this process.

Normally, if the cost rises by 500 yuan, the selling price rises by 500 to 800 yuan, and the profit margin remains unchanged, which is normal cost transmission. But what actually happened is that the price of entry-level products has nearly doubled, and the reason is hidden in the profit structure of the PC industry.

Low-end models have little profit margin at all. For a 3,000-yuan computer, the manufacturer may only make a profit of 100 to 200 yuan. When the storage cost suddenly doubles, the profit of low-end models is completely wiped out and even turns into a loss.

Manufacturers only have two choices: either shut down this product line, or raise the price to a level that can cover the cost. As a result, the "entry-level model" priced at 5,999 yuan appeared. Products in the 3,000-yuan price range have essentially been eliminated from the market.

The price increase of mid-to-high-end models is moderate, and the reason is also very simple.

Users who buy high-end machines are not so sensitive to prices. If manufacturers raise prices too much, these users will switch to other brands or simply choose Mac.

Therefore, the strategy of manufacturers has become to raise prices sharply on low-end products and slowly on high-end products. On the surface, it is a cost problem, but in essence, it is layered harvesting based on price elasticity: the cheaper the product, the more sharply its price rises, because people who buy cheap goods have no way out.

The resources absorbed by AI are ultimately paid for by ordinary consumers

The shift in production capacity of storage chips is just a microcosm of the reallocation of resources across the entire industrial chain. The same thing is happening simultaneously in all links including chip design, wafer manufacturing, packaging and testing. NVIDIA has merged its gaming graphics card business into the edge computing segment, and the cycle of graphics card iteration has been extended from a major upgrade every two years in the past to nearly five years.

CPU giants prioritize supplying the most advanced process technology to data center product lines, and consumer-grade products can only wait in line for the next round of upgrades.

TSMC's production capacity allocation logic is very clear: data center customers have large order amounts, high profits, and cutting-edge technical requirements, so they naturally get priority.

The consequence of this is that the more prosperous AI technology is, the less cost-effective the computers ordinary people buy will be. Over the past two decades, consumers have been accustomed to the rhythm of Moore's Law, that they can buy products with significantly improved performance by spending the same amount of money every one or two years. This experience is built on the premise that the dividends of technological progress will be distributed to the consumer market.

This premise is now starting to falter. The most advanced technologies, the most scarce production capacity, and the best talents in the chip industry are all converging toward AI infrastructure.

What is left for consumer-grade PCs is the previous generation of process technology, compressed production capacity, and rising costs.

The situation of the PC industry itself makes this problem more difficult to solve. After global PC shipments peaked at 365 million units in 2011, they never returned to that level. This is a stock market with no growth, or even a slowly shrinking market. The cycle for consumers to replace their computers has extended from three or four years to five or six years, and mobile phones and tablets have also diverted part of the demand.

In a market lacking growth, manufacturers have no confidence to invest in new technologies or subsidize low-end product lines. Their only strategy is to hold on to profits and try to earn more from each existing customer.

As a result, an uncomfortable combination has emerged: reduced technology investment and rising costs are occurring simultaneously in the same market. In the past, the performance of computers consumers bought for 3,000 yuan was improving year by year; now, the performance improvement of computers consumers buy for 6,000 yuan is far less than before.

Products are more expensive, but not better. This combination is unacceptable to anyone, but its emergence is probably because upstream resources have been siphoned off by the AI industry. PC manufacturers have no bargaining power, and ultimately all cost pressure and resource shortages are transmitted down the industrial chain and end up on end consumers.

Another detail is that the price elasticity of demand in the PC market is very high, which means that price increases will directly discourage some consumers.

The Hangzhou computer dealer mentioned in the report of China Business News put it very bluntly: in previous school seasons, students who assemble their own computers would come to upgrade their devices and expand memory, but now they back off as soon as they hear the price. The price of hard drives has risen from more than 400 yuan to more than 1,000 yuan, and student groups are the most price-sensitive, so they are the first to leave the market.

What is the problem behind this phenomenon? It is conceivable that if the AI industry siphons off all the resources of the consumer market, will the consumer market shrink to a point where it cannot support the long-term development of AI?

The answer may only be seen in a longer cycle.

But at least from the current situation, the squeeze of the AI industry on the consumer electronics market is real, and there is no sign of easing. Senior executives of HP expect that the proportion of storage costs will continue to rise within 2026.

This means that the pressure for computer prices to continue to rise will remain for some time in the future.

It's not just computers that are unaffordable

For a long time in the past, a laptop priced at around 3,000 yuan was an "affordable digital ticket" for many families. It could process documents, take online courses, watch videos, use social software, and meet the basic learning and living needs of college students. For many ordinary families, this 3,000 yuan was a planned expense that they could afford.

Now the price tags of entry-level models at multiple brand stores are generally no less than 6,000 yuan, and models with slightly better configurations directly cost nearly 10,000 yuan. For families in first-tier cities, this may only be a matter of a slight increase in budget.

But for the broader group of ordinary families, the budget increase from 3,000 yuan to 6,000 yuan means that this sum of money has to be squeezed out from other expenditures, or they simply give up buying new computers and continue to make do with old devices.

The gap in equipment performance will be further widened in the AI era.

For a long time, the gap in computer performance was mostly reflected in running speed and storage space, which had a relatively limited impact on study and work.

What about in the years of AI development? We find that more and more applications are beginning to rely on local computing power, such as running AI assistants, doing data analysis, editing videos, and processing programming tasks.

The gap between an old computer and a new computer is no longer a matter of "slower speed", but a matter of "whether it can be used".

When AI becomes the infrastructure for study and work, the gap in equipment performance will directly translate into a gap in efficiency, a gap in skill acquisition, and even a gap in employment opportunities.

Some parents say that if there is no professional demand, a lightweight laptop worth a few thousand yuan is enough.

That's true, but it assumes a premise that non-professional students do not need high-performance devices.

This premise is being shaken by the popularization of AI applications.

Even a liberal arts student may need to run a language model locally to assist in writing, or process some simple data visualization tasks. When these demands become common, the threshold for the minimum usable configuration will be raised.

The cost of raising this threshold is borne by families whose income growth has not kept pace.

The shrinking of the DIY market is actually one of the important signals.

The Hangzhou computer dealer mentioned that now the price of hard drives has doubled, and students who assemble their own computers no longer come once they hear the price. On the surface, the DIY market is just a circle of people who assemble computers themselves, but its significance goes far beyond that. This market is the first stop for ordinary people to get in touch with hardware technology, and it is the starting point for many people to cultivate interest in technology, learn systematic knowledge, and enter the technology industry.

If the threshold of this market is raised significantly, it means that a group of young people with potential in the future will be blocked from the technology world for economic reasons. Storage manufacturers have made profits from AI servers, but they may inadvertently stifle the soil for the next generation of technical personnel to grow.

Faced with this situation, simple waiting will not solve the problem automatically. The storage industry is cyclical, and prices will fall after rising, which is a fact.

This round of price increase is different from the past. The core driving force behind the price increase is not the cyclical fluctuation of demand, but the structural shift of production capacity brought about by the AI industry. As long as the demand for AI remains strong, storage manufacturers have no reason to shift production capacity back to the consumer market. There may be a cyclical decline, but it is unrealistic to return to the original price level in the short term.

I think there are two viable solutions. One is the real rise of domestic storage. Yangtze Memory and ChangXin Memory have already had a certain production capacity and are catching up in technology, but their penetration rate in the consumer market and production capacity scale are not enough to shake the pricing power of the three giants.

This direction requires continuous investment and the cooperation of the entire industrial chain. It cannot be rushed, but it cannot be stopped.

The other is that the regulatory authorities pay attention to the competition order of the storage market.

In history, there have been many cases in the storage industry where production cuts were used to maintain high prices. This kind of oligopolistic tacit agreement has new manifestations in the AI era, such as compressing consumer supply under the pretext of shifting production capacity.

If regulatory authorities intervene earlier, they can at least prevent prices from being further manipulated.

Both paths have the same goal, which is to give ordinary consumers the opportunity to afford computers again.

The disappearance of 3,000-yuan computers is a price issue on the surface, but in essence it is a problem of how technical resources are allocated.

When the fruits of technological progress are monopolized by a few industries and a few companies, and the digital threshold for ordinary people is constantly raised, the narrative that technology changes destiny will become a slogan with less and less convincing power.

Every price increase reminds us that technology itself has no position, but the way technology is distributed does.

This article is from the WeChat official account "Dongzhen Strategy", the author is Dongzhen Strategy, and 36Kr publishes it with authorization.