Computer prices have fallen almost every year since the early 1980s. That four-decade streak just broke, and the reason has nothing to do with tariffs, chip design costs, or inflation in the usual sense. According to Oxford Economics, the artificial intelligence boom has triggered the first sustained increase in personal computer prices in more than 40 years, and the mechanism behind it is almost mundane: there simply is not enough memory to go around anymore.

The Trend Line That Broke After Four Decades

Bernard Yaros, lead economist at Oxford Economics, described the shift to CBS News in terms that underline how unusual this moment actually is. Products that have historically seen prices hold flat or decline are now moving in the opposite direction, and it has been happening consistently for months, not as a single bad quarter. The culprit is a shortage of DRAM and NAND flash memory, the same components that power both the laptop on your desk and the AI data centers reshaping global computing demand.

Gartner analyst Ranjit Atwal put numbers on the damage back in February: global PC shipments are projected to fall 10.4 percent in 2026, smartphone shipments 8.4 percent, while PC prices climb roughly 17 percent and smartphone prices roughly 13 percent compared with 2025. Atwal's more unsettling point was about duration rather than size. He said this shortage differs from prior memory price spikes specifically because of how long it is expected to last, with no meaningful pricing normalization likely before the end of 2027.

Why Your Next Phone Might Come With Less RAM

For device makers, memory is not a minor line item. IDC's Francisco Jeronimo notes that memory can represent 15 to 20 percent of the total bill of materials for a mid-range smartphone, and 10 to 15 percent even for a high-end flagship. When that single component's cost spikes sharply, manufacturers have exactly two options: raise the sticker price, or quietly strip out capability while charging the same amount.

Both are already happening. Consumer Reports has documented manufacturers downgrading internal specifications while holding prices steady or raising them anyway, a pattern TrendForce vice president Avril Wu says will likely occur simultaneously rather than as alternatives. The clearest example is at the low end of the smartphone market, where entry-level Android devices are expected to slide back down to 4 gigabytes of RAM in 2026, reversing years of specification creep that had made even budget phones reasonably capable. Gartner goes further, projecting that the entire sub-500-dollar laptop segment will disappear completely by 2028. The cheapest tier of computing is not just getting more expensive. It may stop existing.

The Cooldown Nobody Should Mistake for Relief

There is a recent data point that could easily be misread as good news. TrendForce's latest memory pricing survey, reported by Tom's Hardware, projects DRAM contract prices to rise 13 to 18 percent quarter over quarter in the third quarter of 2026, with NAND flash climbing 10 to 15 percent. Those are real increases, but they represent a sharp slowdown from the roughly 60 percent jumps recorded just one quarter earlier.

The report is explicit about why the pace is cooling, and the reason is not reassuring. It is not driven by improving supply. It is driven by consumer electronics manufacturers hitting the ceiling of what they are willing and able to absorb after months of relentless increases. Buyers, in other words, are tapped out before the shortage itself has eased. TrendForce's own analysis expects the memory market to keep splitting further into two separate tracks, with server and AI infrastructure demand staying strong straight through 2027 even as the consumer side of the business weakens under price fatigue. Slower growth in the number on your receipt does not mean the underlying scarcity is resolving. It means shoppers are running out of room to keep paying more.

What's Actually Driving This: One Chip, Three Tradeoffs

The root cause traces back to a single technical decision playing out across the entire memory industry. High-bandwidth memory, or HBM, is the specialized, tightly stacked memory that AI accelerator chips from Nvidia and others require, and it is dramatically more difficult to manufacture than the conventional DRAM used in ordinary laptops and phones. According to Micron executive Sumit Sadana, producing one bit of HBM requires giving up roughly three bits of conventional memory production capacity, a direct tradeoff built into the manufacturing process itself.

With only three companies, Samsung, SK Hynix, and Micron, controlling nearly the entire global memory supply, and all three racing to capture premium AI pricing, consumer devices are quite literally losing out on capacity that used to be theirs by default. Micron confirmed as early as January that it was already sold out of memory production for all of 2026. Samsung and SK Hynix have both separately warned that AI-driven shortages could persist into 2027 and beyond. This is not a temporary supply hiccup working its way through the system. IDC's own analysis calls it a potentially permanent, strategic reallocation of the world's silicon wafer manufacturing capacity, away from the everyday devices people buy and toward the infrastructure powering AI.

How Companies Are Passing the Cost Along

Apple has already moved first among major device makers, raising prices on MacBooks and iPads and describing the underlying memory shortage in its own statement as an "unprecedented challenge." Best Buy's incoming chief executive, Jason Bonfig, told reporters the company already saw staggered price increases in its first quarter of 2026 and expects average selling prices to climb further while unit sales soften from an elasticity standpoint, meaning some shoppers will simply buy less or wait longer as prices rise. IDC research manager Jitesh Ubrani projects prices across PCs, tablets, and smartphones climbing somewhere between 10 and 20 percent before the end of 2026, a wide range that itself reflects how much uncertainty remains in exactly how hard this will hit different product categories.

When Does This End, and What Happens Meanwhile

New manufacturing capacity is the only real fix, and it is years away. Micron's new fabrication plants in Boise, Idaho are not expected to begin production until 2027 and 2028, and a separate planned facility in Clay, New York is not projected to come online until 2030. Until that capacity actually exists, every additional gigabyte of memory funneled toward an AI data center is a gigabyte that does not make it into a consumer laptop or phone, and no amount of retail pricing strategy changes that basic constraint. If you have been putting off a new laptop or phone purchase, the honest read of this data is that waiting for prices to drop back to where they were is not really an option on any near-term timeline. The better question is whether to buy now, before the next round of increases lands, or extend the life of what you already own until the supply picture genuinely improves sometime after 2027.