Samsung Electronics has developed UFS 5.0, the industry’s first mobile storage solution optimized for “on-device AI.” Sequential read speed reaches 10.8 gigabytes (GB) per second — more than double the previous generation (UFS 4.1) — and power efficiency improves by over 40%. Designed to read and write data faster for an era when AI runs directly on the phone, it enters mass production in the fourth quarter of this year.

TL;DR

  • Samsung has developed UFS 5.0, the industry’s first mobile storage optimized for on-device AI. Mass production is set for Q4.
  • Sequential read of 10.8 GB/s and write of 9.5 GB/s — more than twice as fast as the prior UFS 4.1 — with power efficiency up over 40%. Capacities go up to 1TB.
  • Built on 9th-gen V-NAND (V9), it targets not just flagship smartphones but next-gen devices such as XR headsets and AI wearables.

What happened?

Samsung Electronics has become the first in the industry to roll out UFS 5.0, a next-generation mobile storage solution. UFS (Universal Flash Storage) is the standard storage that goes into mobile devices such as smartphones — the channel through which photos, apps and the operating system are saved and loaded. This product widens and speeds up that channel, reflecting the UFS 5.0 standard that the international standards body JEDEC finalized in October 2025.

The biggest change is speed. Sequential read reaches 10.8 GB/s and sequential write 9.5 GB/s — each more than twice as fast as the prior generation, UFS 4.1. Power efficiency improves by over 40%, while the package shrinks to 7.5 mm × 13 mm × 0.9 mm, 16.7% smaller than before. It comes in capacities up to 1TB and is built on Samsung’s 9th-generation V-NAND (V9).

Why does it matter for on-device AI?

The crux is the shift toward running AI directly on the device. Until recently, generative AI mostly computed on servers (the cloud), but on-device AI — running large language models (LLMs) directly inside phones and wearables — is growing fast. This approach has no network latency and keeps personal data off the network, but it makes the ability to read and write large volumes of data instantly all the more important.

If storage is slow, even an excellent AI model hits a bottleneck at the data-loading stage. Faster read speeds, as in UFS 5.0, reduce response lag and speed up replies when running an LLM on the device. That’s exactly why Samsung is positioning this product as “optimized for on-device AI.”

Where will it be used?

The stage isn’t limited to flagship smartphones. Samsung said it would expand UFS 5.0 supply in step with emerging device markets — next-gen smartphones, of course, but also XR (extended reality) headsets and AI wearables. Because such devices must handle heavy AI computation inside a small body, storage that is both fast and power-thrifty is especially valuable.

That said, the point at which it actually ships in consumer devices comes after mass production. Once Q4 production begins, UFS 5.0 could appear as early as next year’s flagship lineups.

The bottom line

This development illustrates the trend of “solving the AI-era bottleneck at storage.” If the AI race had been concentrated on processors and high-bandwidth memory (HBM) that handle computation, the front line now widens to mobile storage that feeds data quickly. Three things are worth watching: whether Q4 mass production stays on schedule, how much real-world performance and battery efficiency improve on actual devices, and how rivals catch up in UFS 5.0 mass production. The more on-device AI spreads, the more the once-overlooked speed of storage emerges as a variable that shapes the user experience.

※ This article is for informational purposes only and is not investment advice.

Sources