The artificial intelligence (AI) boom is creating an unlikely casualty, your next smartphone, laptop, or even home appliance. As AI drives down the world’s supply of RAM, rising memory costs mean everything from laptops to dishwashers is about to get a lot more expensive.
As tech giants like Microsoft, Google, and ByteDance race to secure memory chips for data centers, a global shortage is quietly taking shape, one that is already pushing up prices across consumer electronics.
Memory has always been a cyclical business, swinging between brutal busts and euphoric booms. What’s different this time is that AI is pulling the cycle forward, draining supply from everyday devices to feed data centers.
In February, according to Reuters, companies like Intel and AMD notified Chinese customers of supply shortages for server central processing units. Intel warned of delivery lead times of up to six months. Global demand for smartphones, personal computers and gaming consoles is expected to shrink this year as companies from Britain’s Raspberry Pi to HP raise sticker prices to offset surging memory chip costs.
In November, Samsung Electronics raised prices of certain memory chips by as much as 60% compared to September, owing to the short supply due to the global race to build AI data centers.
Read more: AI & Job Loss in Emerging Markets: Fewer Entry Jobs, Higher Skills, Bigger Stakes
Experts are saying an unexpected side effect of the AI boom this year will be felt by smartphone or PC consumers since prices are predicted to go up because of a global scramble for memory chips. According to Reuters, Microsoft, Google and ByteDance are racing to lock in supply from the three largest memory chip makers SK Hynix, Samsung and Micron.
At the center of this crunch is high-bandwidth memory (HBM), a critical component for AI workloads that is far more profitable to produce than conventional memory used in everyday devices. As manufacturers like Samsung, SK Hynix, and Micron shift production toward AI demand, the ripple effects are being felt across global supply chains, turning what was once a cyclical industry into a structurally constrained one.
The AI frenzy is driving a memory chip supply crisis. An acute global shortage of memory chips is forcing AI and consumer-electronics companies to fight for dwindling supplies, as prices soar for the unglamorous but essential components that allow devices to store data.
This includes humble flash chips used in USB drives and smartphones to high-bandwidth memory (HBM), a type of super-fast computer memory that uses a unique 3D-stacking design to process huge amounts of data quickly and efficiently.
Producing HBM for AI servers is more profitable than making conventional memory for phones, laptops and TVs. As a result, manufacturers are diverting capacity toward AI at a time when they still can’t produce enough high-end chips to keep up with demand. To get an idea of the scale, OpenAI, signed supply agreements with Samsung and SK Hynix last October for its Stargate project, which could require up to 900,000 wafers per month by 2029, roughly double today’s global HBM output.
There’s also a longer-term risk baked into this scramble. The industry hasn’t seen meaningful unit growth in years, yet it’s now being asked to absorb unprecedented capital spending to serve AI workloads. New fabs cost tens of billions of dollars and take years to come online, which means today’s shortage could easily become tomorrow’s oversupply if AI demand slows or shifts.
The crunch is already being felt. In Tokyo’s electronics district of Akihabara, retailers have begun restricting memory purchases to curb hoarding. Samsung has warned that the shortage affects everything from phones to TVs and home appliances. Chinese smartphone makers Xiaomi and Realme said in November itself they may have to raise prices or focus on selling more premium models. IDC and Counterpoint now expect the global smartphone market to shrink next year as higher memory costs filter through.
Memory makers are expanding, but relief won’t come quickly. SK Hynix has said its 2026 production is sold out, while Samsung says customers have already lined up for next year’s HBM. New factories for conventional memory won’t meaningfully come online until 2027 or 2028.
Read more: Nvidia’s H200 Chip Standoff Shows How AI Is Becoming a Geopolitical Currency
Memory has always been a cyclical business, swinging between brutal busts and euphoric booms. What’s different this time is that AI is pulling the cycle forward, draining supply from everyday devices to feed data centers.
There’s also a longer-term risk baked into this scramble. The industry hasn’t seen meaningful unit growth in years, yet it’s now being asked to absorb unprecedented capital spending to serve AI workloads. New fabs cost tens of billions of dollars and take years to come online, which means today’s shortage could easily become tomorrow’s oversupply if AI demand slows or shifts.
For now, the only certainty is that the AI gold rush is on, and the price of admission might just be hidden in the cost of your next gadget.
