Smartphones are facing a memory bottleneck as they attempt to run larger AI models locally.
While mobile processors can handle these large language models (LLMs), current LPDDR memory struggles to feed data fast enough.
>>> Redmi Turbo 5 vs OnePlus Nord 6: Which Mid-Range Phone Wins in 2026?
To address this, a new memory format called LLW (Low Latency Wide DRAM) is reportedly under development.
Inspired by HBM (High Bandwidth Memory) used in AI datacenters, LLW aims to provide significantly higher bandwidth—10 to 15 times faster than the best LPDDR memory—without the complex packaging and cooling that HBM requires.
>>> Lenovo Launches Affordable Lecoo Bellator GK101 Mechanical Keyboard with Tri-Mode Connectivity
LLW achieves this by using a wide interface to connect memory layers close to the processor, similar to HBM, but in a form factor suitable for smartphones.
It promises lower latency and higher bandwidth than LPDDR while avoiding the thermal and space constraints of stacked memory designs.
According to tipster Fixed Focus Digital, LLW could reduce power consumption by about 50% and improve performance by roughly 1.5 times.
>>> Ugreen Launches 30W Magnetic Phone Cooler with -25°C Cooling and RGB Lights
However, these claims come from industry bloggers and not official sources, so caution is warranted.
The same leak suggests LLW is still years away from widespread adoption, with larger-scale deployment not expected until the second half of 2027.
>>> Samsung unveils 40,000-nit RGB OLEDoS microdisplay for XR devices
Xiaomi and Huawei are mentioned as potential early adopters, but neither company has publicly commented on the technology.
