The Middle Class’s New Digital “Gold” Is Surging Like Crazy

End-of-year is usually when tech companies pack the calendar with new product launches. But this year feels different.

In the past, launch events were basically price wars—who could push the cost lower, who could offer more for less.

This year, brands are busy doing something else entirely: preparing everyone for higher prices.

On December 25, Xiaomi officially unveiled the Xiaomi 17 Ultra. During the livestream, Xiaomi president Lu Weibing openly admitted that this generation’s price increase was “a bit much.” The top-end 1TB model is reportedly 700 yuan more expensive than its predecessor.

And this isn’t Xiaomi’s first “cost shock” moment. When the Redmi K90 launched earlier, its starting price also quietly edged upward.

It’s not just phones anymore. Tablets, laptops, complete PCs—everything is catching the same wave.

Behind all those price tags, one common force is doing the pushing:

Memory.

And calling this round of memory price hikes “crazy” is not exaggeration—it’s accurate.

Take a standard 8GB DDR4-3200 stick. Before the surge, the average price dropped to around 70 yuan.

By mid-December 2025, that same spec was averaging around 350 yuan—an increase of over 280%.

The first people to feel the chill were PC builders.

Memory is non-negotiable in a build. You can compromise on peripherals—skip the fancy keyboard, go cheap on a mouse—and still survive. But upstream memory supply is controlled by a handful of giants, and there’s no real budget alternative.

If GPUs get expensive, you can stop chasing AAA titles. If CPUs go up, you can tolerate a few extra seconds of waiting. But memory is the one component you can’t “just live without.” Modern operating systems and browsers—especially Chrome—are memory-hungry by design. Open a few tabs on 8GB and you’re already flirting with crashes.

Normally, memory isn’t the flashy part of a PC build. It’s rarely the performance bottleneck, and it usually doesn’t dominate the budget. The expensive pieces are typically the CPU and GPU.

But this time, the price jump is so extreme that it’s rewriting how people calculate a build.

Using the earlier example: 8GB is entry-level, and most people need two sticks for 16GB just to feel comfortable in daily use.

A Kingston 8GB×2 kit on JD is now listed at 899 yuan. Earlier this year, it was around 219 yuan.

In just a few months, the total cost of building a PC has effectively jumped by 600–700 yuan.

Gold rising 70% in a year got called “insane.” Compared to memory, gold looks tame.

At the start of the year, 2,000+ yuan could get you a basic setup. Now, if you don’t bring 3,000 yuan to the table, people almost feel embarrassed to even draft you a parts list.

Two 16GB sticks can now cost what an entire budget PC used to cost. Some people who bought memory in early 2024 can sell used sticks today and still come out ahead.

Yes—your “electric meter” is basically running backward.

People used to buy memory to play games.

Now they buy memory to watch the return.

If we’re being honest, since Moutai hasn’t really been skyrocketing this year, memory is starting to look like the new Moutai.

It doesn’t just rise like Moutai—it also disappears like Moutai.

Not long ago, several Japanese PC makers—including TSUKUMO, Sycom, and Mouse Computer—announced they were halting orders, citing their inability to maintain costs and inventory.

Mouse Computer added that orders may resume next year—but pricing won’t be what it used to be.

For PC builders, it’s collective pain.

But while people who want to build today are frustrated, those who stocked up earlier are grinning.

One person bought a high-end 32GB kit to run large models, never really used it, and then discovered later that it had appreciated enough to net over 1,000 yuan in value.

Still, for DIY hobbyists, that’s pocket change.

The real winners are the upstream memory giants.

Micron recently reported financial results showing its share price up roughly threefold this year, with margins surpassing even TSMC. Samsung and SK hynix are also thriving—and the production lines that used to “mysteriously” suffer accidents during downturns suddenly look a lot more stable when money is flowing.

When profits are this good, everyone finds motivation.

So how did memory become “electronic Moutai”?

Memory price cycles aren’t new. Veteran PC builders have seen it all before.

This industry is famously cyclical. Historically, if you were patient, prices would come back down eventually. The memory market is also highly standardized—there’s limited differentiation—and it’s dominated by just a few major players like Samsung, SK hynix, and Micron.

At the same time, memory manufacturing requires massive investment and long lead times. That lag causes producers to react late.

When demand is strong and prices rise, everyone expands capacity. But by the time the new supply actually hits the market, demand often cools—and prices crash.

And because the market has so few players, price “management,” open or subtle, has always existed.

In past years, whenever inventory piled up and prices dropped, factories would suddenly face fires or floods, followed by production cuts and a price rebound.

People complained, but still bought anyway—because they had no choice.

In recent years, Chinese players like CXMT and YMTC began to chip away at that dominance. With more competition, prices came down, and DDR4 even hit “cabbage price” territory. Builders celebrated, convinced the good days had arrived.

But in 2025, this price surge forced everyone to accept a hard reality:

This time is different.

And the reason can be summed up in two letters:

AI.

Some upstream manufacturers reportedly have orders lined up into 2027 and beyond. Not because consumers suddenly bought more PCs—but because AI compute is devouring memory at a terrifying rate.

Beyond raw compute, AI’s other major bottleneck is memory bandwidth.

Phones and PCs mostly use DDR memory.

AI models, however, hunger for HBM (High Bandwidth Memory)—faster, more advanced, and dramatically more expensive. Per gigabyte, HBM can consume multiple times the resources required by DDR.

In the AI arms race, big players are spending with effectively no ceiling. That’s why memory makers are shifting capacity toward higher-margin HBM lines.

And what gets squeezed first?

Lower-margin consumer DDR.

Micron even shut down Crucial’s consumer-facing positioning in some areas to go all-in on enterprise markets. Meanwhile, reports that Samsung and SK hynix may phase out DDR4 production sent DDR4 pricing even higher.

For memory makers, the situation is simple:

There’s money everywhere. The only question is which pile to pick up first.

But for everyday consumers, this is where the pain begins.

Memory is foundational infrastructure across tech. If memory prices stay elevated, next year’s hardware will feel it everywhere.

Phones and tablets will likely take the hit first.

The iPhone Pro’s 12GB memory component costs have reportedly surged sharply. Whether the iPhone 18 pricing will reflect that remains uncertain.

Apple can sometimes absorb cost increases because of its margins and pricing power. But mid-range and budget devices don’t have that luxury.

Redmi K90 has already seen price increases across the lineup, reportedly in the 100–400 yuan range. Xiaomi tablets have also quietly adjusted upward, and more brands are expected to follow.

On the PC side, the outlook isn’t any better.

Dell has announced commercial PC price increases starting mid-December, with reported hikes ranging from 10% to 30%. HP appears ready to play the “less for the same price” strategy—cutting 16GB to 8GB and daring buyers to switch brands.

If you’re planning to buy a phone or a computer, the practical advice is simple:

Buy earlier, not later.

And yes—this doesn’t stop at consumer electronics.

Even new energy vehicles aren’t immune, because infotainment systems and onboard computing also rely heavily on memory.

What we’re seeing is an AI tax, evenly distributed across everyone.

AI doesn’t just “consume memory.” It already made gamers pay the first bill through GPU pricing.

Since NVIDIA declared its “iPhone moment,” its stock has surged, and the company has evolved from a gaming GPU brand into a compute supplier powering the world’s AI ambitions.

For gamers, the result is brutal: you’re no longer just competing with other players. You’re competing with AI workloads for the same expensive hardware.

The market now offers only two options:

Expensive—and even more expensive.

And AI’s appetite doesn’t stop at chips.

Data centers are massive power consumers. A large data center can use as much electricity as a mid-sized city. Estimates suggest AI data centers already account for around 5% of U.S. electricity usage, with more growth ahead.

To keep these compute clusters stable, power companies are investing billions into grid upgrades, substations, and new generation.

Those costs don’t disappear.

They get averaged into everyone’s electricity bill.

China is in a comparatively better position, but data center electricity usage still represents a meaningful share nationally—and the upward trend looks hard to ignore.

The ripple continues further.

To expand power infrastructure, demand for copper—one of the most basic industrial materials—has surged, driving prices up sharply this year. Once copper rises, anything involving electricity, wiring, or heat management feels the shock.

Air conditioner makers are already exploring “aluminum replacing copper” designs—not because it’s better, but because copper has become too expensive to use the old way.

Before AI creates its full promised value for society, it’s already reshaping consumer prices.

You can choose not to use AI products.

But you can’t avoid the AI tax.

GPUs, electricity, phones, computers—its impact leaks into everything.

And as long as AI giants keep buying capacity without blinking, as long as data centers keep running at full throttle, memory prices may never return to that “cabbage price” era people thought was normal.

Of course, AI isn’t only driving prices up.

It’s also driving labor costs down.

This year alone, big tech companies like Amazon and Microsoft reportedly laid off over 50,000 people—and many of those roles are being replaced by AI.