DGX-Spark Deployment Support Renewed: Athena Technologies Standardizes Standalone Environment, Enables Deployment in as Little as Two Weeks

Athena Technologies Co., Ltd. (Headquarters: Bunkyo-ku, Tokyo; CEO: Takeshi Abe; hereinafter "Athena") announces the renewal of its "DGX-Spark Deployment Support" service. This update redefines the provision in a "standalone environment" physically isolated from external networks as a standard service. This simplifies the network configuration process previously required for each organization's environment, aiming for secure AI environment construction in as little as two weeks from order. The service scope has also been clarified, enabling customers to select options tailored to their individual circumstances.

What is DGX-Spark Deployment Support?

This service leverages NVIDIA's AI supercomputer "DGX Spark" to build a high-security, high-performance local LLM environment on-premises.

*In addition to the "DGX Spark," construction using various AI workstations (such as Lenovo's "ThinkStation") is also possible, flexibly adapting to customer needs.

【Main Features and Benefits】

High Safety through Physical Isolation: Operates on a local LLM cut off from external networks, allowing safe business use of highly confidential information without transmitting it externally.

No-Code Application Development: Tools like "Dify" enable the construction of AI agents incorporating in-house materials without the need for programming.

For detailed service content, please also refer to our previous press release.

DGX Spark Deployment Support Service | Matsuo Lab-originated Startup Athena Begins Support for Building "Fully On-Premises LLM" Protecting Confidential Data. Utilizing the latest NVIDIA "DGX Spark," deployment in as little as two weeks.

Aiming to Balance "Rapid AI Implementation" and "Flexible Response to Individual Requirements"

Service Scope

We have summarized the details of the service scope for your reference. This renewal provides end-to-end support, including a two-unit DGX Spark cluster configuration, setting up the Dify and Open WebUI environments, and providing LLM inference services.

Furthermore, delivery in a "standalone environment" that operates independently without connecting to the internal network is now standard. This ensures security in handling confidential information while enabling rapid start of operations.

Please see below for details.

【Functional Requirements】

Requirement Name Requirement Details

Two-Unit Cluster Configuration Cluster two DGX Spark units to achieve distributed inference

Dify Environment Install Dify and enable basic operations (start/stop/restart)

Open WebUI Environment Install Open WebUI and enable basic operations (start/stop/restart)

LLM Provision Provide inference service for LLM (e.g., gpt-oss-120b)

Embedding Provision Provide embedding API (e.g., Qwen3-Embedding-0.6B)

Reranker Provision Provide reranking API (e.g., Qwen3-Reranker-0.6B)

【Non-Functional Requirements】

Requirement Name Requirement Details

Security Enable standalone use of Dify, Open WebUI, LLM, Embedding, and Reranker

Security Minimize privileges and separate operational users

Performance Design GPU memory usage limits per model

Performance Ensure headroom when LLM/Embedding/Reranker coexist

Time Synchronization Configure NTP to maintain log time consistency

Flexible Individual Design via Options

For advanced network and security requirements specific to an organization that fall outside the standard service scope, we offer individual design options. Main options include:

Network Fixation: Assigning fixed IP addresses to each node

Security: Supporting HTTPS communication, controlling SSH access via key authentication, etc.

In addition to the above, we can provide individual quotes for configurations tailored to the customer's specific circumstances.

Case Studies and Detailed Service Materials

We have published materials summarizing the latest service specifications after the renewal and specific case studies, including one from Mie Prefectural Government. Please use them as hints to concretize the image of utilization within your own company.

⇒ Download materials here

Representative Comment

Takeshi Abe, CEO and Representative Director

"With this renewal, by making the 'standalone' environment the standard, we aim for speed-driven deployment led by the field. At the same time, we will reliably respond to complex requirements through options.

Going forward, Athena will continue to advance steadily towards realizing a society where everyone can safely and quickly utilize cutting-edge AI technology."

Company Overview

【About Athena Technologies Co., Ltd.】

As a "Matsuo Lab-originated Startup (Note 1)," we combine cutting-edge academic technology with social implementation capabilities, offering secure AI development and governance support.

Company Name: Athena Technologies Co., Ltd.

Established: December 2023

Representative Director: Takeshi Abe

Location: HONGOEGG, 6-25-14 Hongo, Bunkyo-ku, Tokyo (Tokyo Office); Hyakumanben Building 3F, Tanaka Monzen-cho, Sakyo-ku, Kyoto City, Kyoto Prefecture (Kyoto Office)

URL: https://athenatech.jp

(Note 1) "Matsuo Lab-originated Startup" refers to a select group of startups founded by alumni of the Matsuo Laboratory or established with support from the laboratory, recognized for their growth potential in both technology and business, and who share the laboratory's philosophy and work together to foster the next generation. (Registered trademarks No. 6667237, No. 6710069)

Inquiries regarding this matter:

Athena Technologies Co., Ltd. Public Relations

URL: https://athenatech.jp/contact