The same DRAM and NAND chips powering your favorite chatbot are now in a bidding war with your laptop, and the laptop is losing.
Consumer electronics prices are climbing fast because AI companies are hoovering up the memory chips that phones, consoles, and laptops need too. The squeeze pushed Microsoft’s new Surface Pro to $1,599, a $600 jump over the last generation, with more increases expected to roll through 2026.
Picture yourself in the electronics aisle, eyeing that shiny new console you finally promised yourself. You check the price tag twice. It is fifty bucks more than it was a month ago, and the company has literally apologized for it. Welcome to the memory-chip squeeze, where the bill for the AI boom is quietly landing in your shopping cart.
What Happened
The price hikes are arriving all at once, and they are not shy. Microsoft’s latest Surface Pro now starts at $1,599, a full $600 above the previous generation. Nintendo bumped the Switch 2 to $499 after a $50 increase in May, complete with a corporate apology for the impact on customers. Sony’s PS5 Pro jumped to $900 from $750 back in April. And the iPhone 18 Pro could land at an eye-popping $1,299 when it ships in September.
Apple is not pretending otherwise. The company confirmed plans to raise prices because of climbing memory costs. CEO Tim Cook said Apple had been trying to shield buyers from the increases, but that it was no longer possible. “There’s less supply at a time when consumers want devices and the memory guys are passing along huge price increases,” Cook said.
The Backstory
The culprit is a class of components called DRAM and NAND flash memory, the chips that move and store data inside your devices. They are also exactly what artificial-intelligence companies crave to train and run their models. As AI adoption exploded, the memory industry hit a brutal capacity crunch.
Here is the kicker: this whole market is controlled by just three companies. South Korea’s SK Hynix and Samsung Electronics, plus Boise, Idaho’s Micron Technology. That concentration has been spectacular for them. The market values of SK Hynix and Micron both roughly quintupled to more than $1 trillion each in the past six months. Consumers, meanwhile, are footing the bill.
The demand has also been shifting under everyone’s feet. When AI chatbots first took off, the hunger was for high-bandwidth memory, or HBM, made by stacking DRAM chips. But as the models matured, companies like OpenAI and Anthropic leaned harder on NAND storage to do things like remember a user’s query history. “It’s been HBM for the last few years, but it’s a relatively new thing that companies are moving capacity to NAND,” said Indong Kim, a memory sales and marketing VP at Samsung. “We’re starting to observe that demand for NAND is approaching the level of DRAM.”
The Plan
You would think the fix is simple: build more factories. It is not. Memory makers are racing to add manufacturing lines, but semiconductor fabs cost billions of dollars and take two to three years to build, with even more time needed to ramp to full capacity. Analysts expect the shortages and price hikes to stick around for a few more years.
The pain is already rippling outward. In April, Intel, which makes PC processors but not memory chips, warned that rising component costs would shrink the total PC market by a low double-digit percentage. “We’re prudently planning for PC demand to weaken in the second half of the year,” said Intel CFO David Zinsner. The numbers behind the scenes are wild: over the past year, Micron’s contract price for one of the most common data-center memory chips climbed from $350 to $1,300, according to Massachusetts-based reseller Circular Technology. “These companies are trading at high valuations, and because they have to keep their margins high, the consumer is the one who loses,” said Circular’s head of global research, Brad Gastwirth. “The laptops coming at the end of the year are not going to be cheap.”
The Business Model Angle
This is a clinic in what happens when you own the bottleneck. Three firms control a scarce input that two enormous, desperate buyer groups both need at once: AI labs with effectively unlimited budgets, and consumer-device giants with thinner margins and price-sensitive customers. The memory makers get to set the price. Everyone downstream becomes a price-taker, and the consumer absorbs the final markup.
That is the same lesson sitting underneath Nvidia’s business model: the most durable profits in a gold rush rarely go to the prospectors. They go to whoever sells the scarce, hard-to-replicate shovel. For founders, the takeaway is uncomfortable but clear. If your company sits one layer downstream of a constrained input you do not control, your margins are not really yours. They belong to whoever owns the chokepoint, and they will claw them back the moment demand spikes.
The Risk
Here is the honest counterpoint, and it is a big one: memory is the most famously cyclical business in tech. Boom always invites bust. The same oligopoly minting trillion-dollar valuations today has been flattened before by oversupply, when everyone adds capacity at the top of a cycle and then watches prices collapse together.
That two-to-three-year fab timeline cuts both ways. If a wave of new factories all light up around the same moment, or if AI demand cools even slightly, the shortage can flip to a glut with brutal speed. Those quintupled valuations are pricing in permanence that the industry’s own history does not support. And consumers have a vote too. Intel is already bracing for weaker demand, because the simplest response to a $600 price jump is to just not buy this year.
Quick Questions
Why are phone and laptop prices going up in 2026?
Because AI companies are buying up huge volumes of the memory and storage chips that consumer devices also depend on. Less supply plus surging demand means higher prices, and device makers are passing those costs straight to you.
What are DRAM and NAND chips, exactly?
They are the memory and storage components inside your gadgets. DRAM handles fast data transfer while your device runs, and NAND flash stores your files and apps. Both are now in heavy demand from AI data centers.
Which companies make most of the memory chips?
Just three: SK Hynix and Samsung in South Korea, and Micron in the United States. That tight concentration is a big reason prices can rise so sharply with so little pushback.
Will memory chip prices come back down?
Not soon. New factories take two to three years to build and longer to hit full output, so analysts expect the squeeze to last a few more years. The wildcard is the memory cycle, which has a long history of flipping from shortage to glut fast.
The Business Model Analyst Take
When a scarce input gets caught between two buyers who both must have it, the company that controls supply wins and everyone downstream pays. The memory makers are running that playbook to perfection right now. For operators, the lesson is to map your dependencies before a crunch exposes them: know exactly which input could double in price overnight, and whether you have any leverage at all when it does. Pricing power lives at the chokepoint, and if you are not standing on it, you are standing under it.
Source: Robbie Whelan, The Wall Street Journal.
