How We Reduce the Impact of Our Products on the Environment
The NAND flash memory and solid state drive (SSD) products that Kioxia Group supplies are incorporated into a wide range of applications and used in various situations throughout the world. We are making efforts to reduce environmental impacts at every stage of the product lifecycle. This includes reducing the CO2 emissions intensity per unit of product capacity during manufacturing and designing products for improved energy efficiency per unit of product capacity during use.
Given the rapid spread of AI, the increased number of data centers, and the acceleration of technological innovation in autonomous driving, the need for larger capacity and faster NAND flash memory and SSD products has continued to grow. Kioxia Group’s ongoing development of integration technology and the resulting increase in memory capacity are allowing us to reduce the level of CO2 emissions per product capacity during manufacturing. Specifically, this is being achieved by reducing the amount of electricity and materials used in manufacturing per unit of product capacity, and by improving the energy efficiency of our products for processing 1 GB of data when they are in use.
In the field of AI in particular, there is concern that growing demand will lead to an increase in electricity consumption. For this reason, we expect demand for highly energy-efficient, large-capacity flash memory and high-throughput SSD products to continue expanding across a wide range of fields. Furthermore, in order to understand the environmental impact of its products throughout their lifecycles, Kioxia Group has been evaluating CO2 emissions using life cycle assessment (LCA)*. Since FY2023, we have continued to obtain third-party verification from DNV. We plan to continue expanding the number of products evaluated for CO2 emissions using LCA in the future.
To contribute to the realization of a low-carbon society, the Group will continue to prioritize improving energy efficiency and increasing product capacity through technological development.
- A method for quantitatively assessing the environmental impact of a product or service throughout its lifecycle (raw material extraction, manufacture, transportation, usage, and waste disposal or recycle) or at a specific stage in its lifecycle.
Kioxia Group Products Helping to Protect the Environment
Please visit the website below for the latest product information.
Targets for Improving the Energy Efficiency of Our Products
Kioxia Group set a target of reducing the energy consumption of its memory and SSD products per 1 GB of data processed by 50% (an average annual reduction of 6.3%) from FY2017 levels by FY2025 and has worked to increase the energy efficiency of its products. Although the product plan was revised after the goals were set, resulting in a reduction rate of 44% for fiscal year 2025 (an annual average of 5.5%), our measures to reduce energy consumption proceeded as planned.
Improving the Energy Efficiency of Our Products: FY2030 Target
As a new target from FY2026, we aim to achieve a reduction of 30% in the energy consumption required by memory and SSD products to process 1 GB of data compared to FY2025 levels by FY2030 (average 6.0% per year), which is equivalent to the previous target. We will promote product development and increased sales of high-efficiency products as we continue to increase the energy efficiency of our products.
Case Studies: Reducing Energy Consumption during Product Use
UFS Ver. 4.1 Embedded Flash Memory Devices for Next-Generation Mobile Applications
Kioxia Corporation has begun mass production and shipments of new Universal Flash Storage (UFS)1 Ver. 4.1 embedded flash memory devices (UFS 4.1 products) for next-generation mobile applications, including advanced smartphones with on-device AI. Compared with the previous generation of products,2 the new devices deliver improved read/write performance and greater power efficiency, enabling faster data downloads and smoother app performance while consuming less power, thereby contributing to a better user experience.
The products integrate Kioxia’s BiCS FLASH™ 3D flash memory and a controller in a JEDEC-standard miniature BGA package, and incorporate Kioxia’s 8th generation BiCS FLASH™ 3D flash memory3 featuring CMOS directly Bonded to Array (CBA) technology. By directly bonding CMOS circuitry manufactured on a separate wafer to the memory array, CBA technology enables major gains in flash memory power efficiency, performance, and bit density.
Key Features of the Product:
- Available in capacities of 256 gigabytes (GB), 512 GB and 1 terabyte (TB)
- Performance improvement over previous generation2:
- Random writes: 512 GB / 1 TB approx. +30%
- Random reads: 512 GB approx. +45%, 1 TB approx. +35%
- Power efficiency improvement over previous generation2:
- Reads: 512 GB / 1 TB approx. +15% improvement
- Writes: 512 GB / 1 TB approx. +20% improvement
- Reduced package height for the 1 TB model compared to the previous generation4
- Uses Kioxia’s 8th generation BiCS FLASH™ 3D flash memory3
- A product category for a class of embedded memory products built to the JEDEC UFS standard specification. Due to its serial interface, UFS supports full duplexing, which enables both concurrent reading and writing between the host processor and UFS device.
- Compared to the previous generation UFS 4.0 embedded flash memory 512 GB device “THGJFMT2E46BATV” and 1 TB device “THGJFMT3E86BATZ,” respectively (512 GB / 1 TB models only)
- 512 GB / 1 TB models only
- The previous generation 1 TB UFS 4.0 embedded flash memory 1 TB device “THGJFMT3E86BATZ”
- Notes:
- 1. In every mention of a Kioxia product: Product density is identified based on the density of memory chip(s) within the product, not the amount of memory capacity available for data storage by the end user. Consumer-usable capacity will be less due to overhead data areas, formatting, bad blocks, and other constraints, and may also vary based on the host device and application. For details, please refer to applicable product specifications. (The definition of 1 GB = 2^30 bits = 1,073,741,824 bits.)
- 2. Read and write speeds are the best values obtained in a specific test environment at Kioxia Corporation, and Kioxia Corporation warrants neither read nor write speeds in individual devices. Read and write speed may vary depending on device used and file size read or written.
- 3. Company names, product names and service names may be trademarks of third-party companies.
KIOXIA LC9 Series Enterprise NVMe™ SSDs for generative AI
Kioxia Corporation has expanded its high-capacity KIOXIA LC9 Series enterprise NVMe™ SSD lineup for generative AI. The product is compatible with the PCIe® 5.0 interface and is available in 2.5-inch, E3.S, and the new E3.L form factor. It is purpose-built for the performance and efficiency demands of generative AI environments.
Generative AI places unique demands on storage, including the need to store vast datasets for training large language models (LLMs), and to create embeddings and vector databases that support inference through retrieval augmented generation (RAG). These workloads require storage solutions with exceptional capacity, speed, and efficiency.
The 245.76 terabyte (TB)1 product, the highest-capacity model in the KIOXIA LC9 Series, was achieved through the development of an 8 TB compact 154 BGA package. Leveraging KIOXIA’s high-precision wafer processing, material design, and wire bonding technologies, the package features a 32-die stack of 2 terabit (Tb)2 BiCS FLASH™ 8th generation QLC flash memory chips.
KIOXIA LC9 Series SSDs deliver high-volume data ingestion and rapid processing while providing high-density storage in a compact footprint with greater capacity per watt. This enables effective utilization of costly GPUs while reducing overall server power consumption, the number of drive slots required in host systems, and cooling costs. These features help drive total cost of ownership (TCO) improvements for data centers across power efficiency, storage density, and thermal management. For example, from a storage perspective, replacing eight 30 TB HDDs with a single 245.76 TB KIOXIA LC9 Series SSD can reduce energy consumption by approximately 67%* while also reducing the space required for drive installation.
- The illustrative example above is based on the performance specifications of KIOXIA LC9 Series NVMe™ SSDs and the published data transfer performance specifications of 30 TB-class high-capacity HDDs from major HDD vendors, based on Kioxia’s research as of July 22, 2025. Energy consumption during 2,200 MB/s sequential writes has been estimated for a system with a total capacity of 2,400 TB.
Features of the KIOXIA LC9 Series:
- Up to 245.76 TB SSD in E3.L form factor
- Up to 122.88 TB available in 2.5-inch and E3.S form factors
- PCIe® 5.0 (max. 128 GT/s Gen5 single x 4, dual x 2)
- NVMe™ 2.0 specifications
- Notes:
- 1. Definition of SSD capacity: Kioxia Corporation defines a gigabyte (GB) as 1,000,000,000 (10^9) bytes and a terabyte (TB) as 1,000,000,000,000 (10^12) bytes. A computer operating system, however, reports storage capacity using powers of 2 for the definition of 1GB = 2^30 bytes = 1,073,741,824 bytes and therefore shows less storage capacity. Available storage capacity (including examples of various media files) will vary based on file size, formatting, settings, software and operating system, and/or pre-installed software applications, or media content. Actual formatted capacity may vary.
- 2. The flash memory capacity is calculated as 1 terabit (1 Tb) = 1,099,511,627,776 (2^40) bits, and 1 terabyte (1 TB) = 1,099,511,627,776 (2^40) bytes.
- 2.5-inch indicates the form factor of the SSD and not its physical size.
- Read and write speed may vary depending on various factors such as host devices, software (drivers, OS etc.), and read/write conditions.
- Performance is preliminary and subject to change without notice.
- NVMe is a registered or unregistered trademark of NVM Express, Inc. in the United States and other countries.
- PCIe is a registered trademark of PCI-SIG.
- Other company names, product names and service names may be trademarks of third-party companies.