Micron Ships Fastest And Thinnest Mobile DRAM, Calls It A Game Changer

In the endless race for smaller and thinner devices, every micrometer counts. That's why device manufacturers are likely excited for Micron's latest LPDDR5X memory, which is 14% shorter than the last generation. For our part, though, we're also pretty keen on the upgraded transfer rate, which is a speedy 10.7 Gbps per pin, up from 7.5 Gbps of the previous generation 1β (1-beta) parts.

Micron's latest is its 1γ (1-gamma) memory, which we first told you about back in February. The fresh news today is that Micron is now ramping its LPDDR5X memory on 1-gamma to target volume production for the mobile segment, which is a welcome development, as it means we might start seeing it in devices sooner rather than later. Why should you care? Well, in addition to the new RAM's svelte dimensions and higher speeds, it also uses 20% less power.


In combination, that means smaller devices, better performance, and better battery life too. Of the three, the performance is actually the largest change. 10.7 Gbps translates to 10,700 MT/s, or "LPDDR5X-10700", a significant upgrade from even the Samsung LPDDR5X-8533 you see in some systems (e.g. Lunar Lake laptops). It's a full 33% faster than the 8-Gbps memory supported on AMD's Ryzen AI MAX processors as well; this upgrade would give those machines a full 342 GB/second of memory bandwidth, putting them on the same footing as the just launced Radeon RX 9060 XT.

These benefits were achieved through a move to advanced EUV lithography—the same process that is typically used for advanced microprocessors. While it benefits from the manufacturing advancements, DRAM doesn't necessarily need all of the features of the latest process technology in the same way that logic (processors) does, thus the lag in implementing EUV fabrication tech until this generation.


Of course, the word of the day in tech is still "AI", so that's what Micron's press release is all about. Micron specifically notes that when running the Llama 2 large language model on a phone, the new RAM improved query response time by 30% and translation results delivery time by 50%, with a 25% upgrade in performance on a recommendation engine. Highly specific metrics, yes, but AI developers are probably excited to hear those numbers.

Micron says that the chips it's currently sampling have a 16GB capacity, but that it will offer 8GB and 32GB capacities in the future. The memory maker specifically notes "2026 flagship smartphones" as a primary use case, so it may be a bit longer before these parts show up in laptops, tablets, and mini-PCs.