The next time the price of a phone goes up, the reason may not be the camera, the screen or even the processor. It could be the tiny memory chips inside it.
The semiconductor industry is facing another supply squeeze, but this one looks very different from the chip shortage that caused problems during the pandemic. Back then, factories were disrupted, shipping was delayed, and demand for electronics suddenly jumped. This time, something else is happening.
The factories are operating, but a growing share of their capacity is being diverted to artificial intelligence. AI data centres need huge amounts of memory, especially a type called High Bandwidth Memory, or HBM. It works alongside powerful AI processors, enabling them to move enormous amounts of data quickly. As companies such as Google, Microsoft, Amazon and Meta continue spending billions on AI infrastructure, demand for these specialised memory chips has exploded.
For memory manufacturers, that demand is difficult to ignore.
Companies such as Samsung, SK hynix and Micron make several kinds of memory products. The same broad manufacturing ecosystem supplies components used in phones, computers, cars and data centres. But manufacturers do not have unlimited factory space. When demand for one type of chip becomes much more profitable, production naturally starts moving in that direction.
That is where smartphone companies are beginning to feel the pressure. A modern smartphone needs memory for almost everything it does. RAM helps the phone keep apps running, while NAND storage holds photographs, videos, applications and other files. Increase the price of those components and the cost of making a phone rises.
The difficult part for manufacturers is deciding who gets priority when supply becomes tight. An AI data centre might place an enormous order and be willing to pay a premium for specialised memory. A phone company, meanwhile, may be ordering conventional DRAM and NAND in huge volumes but at lower margins.
From the chip manufacturer's point of view, the choice can be fairly obvious. More production goes towards AI. That does not mean phone memory has disappeared. It means smartphone manufacturers have to compete harder for the supply that remains.
This is why the current shortage has picked up names such as “RAMmageddon” and “RAMpocalypse.” The names are dramatic, but they describe a problem that could eventually reach ordinary consumers.
Industry forecasts have pointed towards memory prices rising during 2026 while smartphone shipments decline sharply. Some estimates have suggested a fall of around 13 to 14 per cent in global smartphone shipments, which would put considerable pressure on manufacturers already operating in a competitive market.
The impact will probably not be the same for every brand.
Apple and Samsung have an advantage because they are among the industry's largest buyers. They have stronger relationships with semiconductor suppliers and greater purchasing power. They can negotiate large contracts and plan component requirements further ahead.
Smaller manufacturers have less room to manoeuvre. Budget smartphone brands such as Transsion and Xiaomi are particularly exposed because their customers are highly sensitive to price. If the cost of memory rises, a company cannot simply add the entire increase to the price of a phone and assume customers will accept it.
It has to make a choice. The company can accept a smaller profit. It can raise the price. It can reduce the amount of memory in some models. Or it can try to find cheaper components elsewhere.
For consumers, this could mean that the phone market starts changing in less obvious ways. A manufacturer may keep a phone at roughly the same price but offer less RAM or storage. Another company might increase the price while keeping the specifications unchanged. Some brands could also delay launches or concentrate their memory supply on more expensive models.
So when someone says, “Phones are getting expensive because memory is expensive,” that does not necessarily mean the phone's entire cost has suddenly doubled. It means one important component has become harder and more expensive to secure.
And smartphones are not the only products affected. Cars are becoming computers on wheels. Modern vehicles contain memory chips in digital dashboards, navigation systems, infotainment units, driver assistance technology and various electronic control systems. Electric vehicles add even more electronic hardware because batteries and motors have to be monitored and controlled by sophisticated systems.
That creates another potential customer for the same semiconductor industry. A car manufacturer may not be directly competing with a smartphone company for the same chip, but both industries depend on the same global memory manufacturing network. If manufacturers continue prioritising the most profitable AI products, other electronics industries can feel the effects.
PC manufacturers are already watching the situation closely. The interesting thing about this shortage is that consumers may not immediately recognise what is causing it.
A phone shop does not display a sign saying, “AI data centres bought your RAM.” Instead, the changes appear slowly.
A phone costs a little more.
A laptop configuration changes.
A car manufacturer adjusts its production plans.
A particular model launches with a different memory configuration.
Behind those small changes is a much larger battle taking place inside semiconductor factories.
The AI industry has created an enormous appetite for computing hardware. Training advanced AI systems requires powerful processors, but those processors are only useful when they can access huge amounts of data quickly. That is one reason HBM has become such an important part of the AI hardware market.
The demand is unlikely to disappear soon.
Tech companies are still building data centres and investing heavily in AI. Governments are also encouraging domestic semiconductor manufacturing because chips have become a strategic industry rather than simply another consumer product.
But increasing supply takes time.
A semiconductor factory is not something that can be built in a few months. It requires billions of dollars, specialised machinery, highly controlled manufacturing environments and skilled workers. Even after construction is complete, bringing a new facility to large-scale production takes additional time.
That is why nobody can give consumers a simple answer about when memory prices will return to normal. Some industry estimates point towards improvement around 2027. Others suggest the pressure could continue much longer if AI demand keeps growing at its current pace.
There is another possibility too. The market could eventually correct itself. If AI companies stop ordering hardware at today's extraordinary pace, or if memory manufacturers successfully expand production, the balance between supply and demand could change. Prices could then stabilise or fall.
But for now, the pressure is moving in the opposite direction. AI is attracting investment, factory capacity and semiconductor production because the companies building AI infrastructure are prepared to spend heavily.
That leaves traditional electronics manufacturers trying to secure the components they need without destroying their margins. The strange part is that this entire story began with something that most people would never connect to their phone bill.
Artificial intelligence.
People usually think of AI as software. Chatbots, image generators, coding tools and automated systems are the visible side of it. Behind them is a massive physical industry filled with processors, memory chips, servers, cooling systems and data centres.
And that physical infrastructure is now competing with the hardware we use every day. So the next time a smartphone manufacturer announces a price increase, the explanation may be much bigger than the phone itself. Your phone could be competing with an AI data centre for the same piece of semiconductor capacity. And right now, AI is proving to be the customer manufacturers most want to keep happy.
References:
- IEA — Energy and AI https://www.iea.org
- Samsung Semiconductor https://semiconductor.samsung.com
- SK hynix https://www.skhynix.com/?utm_source=chatgpt.com
- Micron Technology https://www.micron.com
- TrendForce https://www.trendforce.com