Sakura Internet Debuts High Power VRT Cloud for GenAI

Sakura Internet Inc. officially launched the NVIDIA H100-equipped plan for its “High Power VRT” VM series, part of its “High Power” cloud service for generative AI. By expanding its server fleet, Sakura Internet has established a stable supply of GPU resources. Based on feedback from numerous customers, the company has determined that the quality of the service has reached a sufficient level for official launch. While the beta version required prior application procedures through a dedicated application form, the official version allows users to create an environment directly from the “Sakura Cloud” control panel. This allows users to immediately begin using the service. Pricing remains unchanged from the beta version.

The beta version’s Ishikari Zone 2 service zone will be Ishikari Zone 1 in the official version. “High Power VRT” is a VM-based GPU cloud service that uses NVIDIA Hopper GPUs and can be used on a per-GPU, hourly basis. It supports tasks requiring large amounts of computation, such as machine learning and deep learning, as well as applications requiring high processing performance, such as language model inference and image generation, and is capable of high-speed processing even in fields where real-time performance is essential. The specifications of the “High-Power VRT” H100 plan include one NVIDIA H100 SXM 80GB GPU, 24 CPU cores, 240GB memory, and 6.9TiB of temporary storage (NVMe). The fee (tax included) is 990 yen per hour, 23,100 yen per day, or 385,000 yen per month.

Also Read: NTT Docomo Biz Debuts Secure Platform “Sekitei TEE”

Sakura Internet is running a “Generative AI Full-Power Support Campaign – 100 Hours of Free NVIDIA H100” offering a coupon for up to 100 hours of free use of the “High-Power VRT” H100 plan exclusively to corporate customers. The campaign application period ends August 22nd. Additionally, for the “High Power” bare metal series “High Power PHY,” the company is preparing servers equipped with new NVIDIA Blackwell GPUs, with the B200 model scheduled to be available in mid-August, and is running a pre-order campaign for the B200 model. Additionally, online orders for the existing H100 and H200 models have resumed, and the supply of GPU resources for the “High Power PHY” series is expected to be further strengthened.

SOURCE: Yahoo