Micron's CEO says the memory chip shortage squeezing your next laptop won't end until at least 2028

On the company's fiscal fourth-quarter earnings call, CEO Sanjay Mehrotra said memory demand will keep outpacing supply through calendar 2027 and into 2028. "In calendar 2027 as well as 2028, we see demand exceeding supply," he said, adding that the industry will see "greater tightness" in those years than even in 2026, which was already a tight year.

Why it matters: Mehrotra didn't hedge this as a maybe. He said Micron has "no line of sight to when supply and demand will return to balance," which is about as direct as a CEO gets on a call investors are listening to closely.

Mehrotra said customers will pay "much higher prices" for memory than they did this year. He didn't attach an exact number to that, but the direction was unambiguous.

In real life it's like being told your rent is going up again next year, and also the year after that, and nobody managing the building can tell you when it'll stop.

Background: the shortage traces back to how AI data centers are built. Companies like Nvidia need huge amounts of high-bandwidth memory (HBM) to feed their AI accelerator chips, and HBM is manufactured on much of the same production lines as the regular DRAM that goes into laptops and desktops.

Why it matters: HBM carries far higher profit margins than consumer RAM. Memory makers including Micron, Samsung, and SK Hynix have all shifted a large share of their production capacity toward HBM, since that's where the money is, leaving less manufacturing capacity for the memory chips in everyday devices.

Industry trackers following the memory market through 2026 have reported steep increases in consumer DRAM pricing, with some DDR5 kit sizes reportedly up well over 100% compared to earlier in the year.

PC maker Dell has already raised prices on notebooks and desktops configured with larger amounts of memory. Laptop manufacturers have reported that memory now accounts for a much bigger share of their total device costs than it did just a few months earlier.

Why it matters: this isn't an abstract supply-chain story. It shows up directly on the price tag the next time someone buys a laptop, builds a PC, or upgrades a desktop's memory.

Micron said it plans roughly $25 billion in capital spending in the first half of its new fiscal year, aimed at expanding manufacturing capacity.

Why it matters: new factories take years to plan, build, and ramp up. Even with that spending committed now, Mehrotra's own 2027-2028 shortage timeline suggests any capacity it adds won't meaningfully ease the crunch until the back half of this decade at the earliest.

By the numbers: Micron posted fiscal fourth-quarter revenue of $54.23 billion and full-year revenue of $133.2 billion, up 256% year over year, with gross margins above 80%. That's a record for the company, driven largely by surging AI-related demand for its chips.

Why it matters: the same supply crunch squeezing consumers and PC makers has been extraordinarily profitable for the company at the center of supplying it, a dynamic that's common in commodity shortages but still worth naming plainly.

Background: yes, repeatedly. The industry has gone through several boom-and-bust "supercycles" over the decades, including one tied to the early Windows PC era in the 1990s and another in 2017-2018, when cloud data center buildout and a shift toward 3D NAND production squeezed conventional DRAM supply. DDR4 retail prices roughly doubled over that 2017-2018 stretch.

Why it matters: those earlier cycles eventually ended when supply caught up or demand cooled, and prices crashed hard afterward. In 2019, DRAM prices fell by roughly 60% over four quarters once hyperscale cloud companies stopped over-ordering. That history is part of why Mehrotra's 2028 timeline stands out: he's describing a shortage with a longer runway than several past cycles combined, because AI demand for memory shows no comparable sign of suddenly stopping.

By the numbers: the DRAM industry has consolidated sharply over the decades, from more than 20 meaningful suppliers in the mid-1990s to effectively three dominant players today: Micron, Samsung, and SK Hynix. Each downturn historically eliminated weaker manufacturers who couldn't survive the price crashes that followed previous boom cycles.

Why it matters: with just three companies controlling the overwhelming majority of global memory production, and all three facing the same incentive to prioritize high-margin AI memory over consumer RAM, there's little competitive pressure pushing any of them to prioritize everyday device supply over the more profitable AI business.

Background: graphics cards have been hit especially hard, since video memory (VRAM) can account for a huge share of a GPU's total cost. On a high-end card like Nvidia's RTX 5090, industry estimates put memory at over 80% of the bill of materials, roughly $820 worth of memory chips alone.

Why it matters: many GPU makers had locked in fixed-price memory supply agreements that kept costs steady through the end of 2025. Those contracts have largely expired now, and 2026 renewals are landing at dramatically higher prices, which is part of why graphics card prices have climbed well beyond their original list prices this year.

• If you're planning a PC build or laptop purchase, memory prices aren't likely to come down soon. Mehrotra's own timeline points to continued tightness through at least 2028.
• Buying more memory now rather than later could save money, if you know you'll need an upgrade within the next year or two, given the trajectory described on the earnings call.
• This is an industry-wide dynamic, not a Micron-specific problem. Samsung and SK Hynix face the same incentive to prioritize AI memory production over consumer chips.
• Expect the effect to show up across device categories, not just desktop RAM, since the same memory chips go into laptops, phones, and game consoles.