Advantech and Spingence Join Forces to Empower NTT ARC’s Next-Gen Autonomous Driving Systems
12/22/2025
Location: Japan
A New Benchmark for Cross-Border Tech Collaboration
A New Benchmark for Cross-Border Tech Collaboration: Advantech and Spingence Join Forces to Empower NTT ARC’s Next-Gen Autonomous Driving Systems
As autonomous vehicles transition from lab environments to real-world deployment, safety has become a central concern. To strengthen the compliance and reliability of autonomous systems, the NTT DATA Automobiligence Research Center (ARC), a subsidiary of the NTT DATA Group, has partnered with three key technology providers: Advantech, Spingence, and their Japan-based integration partner Forcesteed Robotics (FSR). Together, they are leveraging Advantech’s SKY-602E3 GPU server and Spingence’s Edgestar on-premises AI solution to develop a new generation of generative AI-driven autonomous driving systems.
Driving Safer Autonomous Systems with Generative AI
Founded in 1973, the NTT DATA Automobiligence Research Center has nearly five decades of experience in autonomous driving technologies, focusing on advanced research, system development, and consulting services. In recent years, the center has actively explored how AI—especially generative AI—can enhance perception, reasoning, and decision-making to deliver safer autonomous driving outcomes.
A key factor enabling this collaboration was the involvement of Forcesteed Robotics (FSR).
As Spingence’s long-term local technology partner in Japan, FSR has deep domain knowledge of the country’s robotics and automation ecosystem. Understanding both industry adoption processes and enterprise needs, FSR strongly recommended the Edgestar solution to NTT ARC. Their facilitation played a pivotal role in aligning technical requirements across all stakeholders and accelerating the formation of this three-party collaboration.
Project Manager Masanori Yokota emphasized that safety remains the most critical priority for autonomous driving. With generative AI, models such as large language models (LLMs) and vision-language models (VLMs) enable systems to make human-like decisions in complex traffic environments, choosing actions that are both rational and compliant with traffic regulations.
On-Premise AI Deployment: The Optimal Balance of Cost and Performance
To support an LLM/VLM-centric development framework, NTT ARC chose an edge-computing architecture over cloud-based services. The team deployed Spingence’s Edgestar solution alongside the compact Advantech SKY-602E3 tower GPU server to provide the required compute power for model training, inference, and fine-tuning.
“We opted for local deployment primarily due to long-term cost efficiency,” Yokota explained.
ARC found that on-premise infrastructure with comparable performance could operate reliably for five years at a total cost equivalent to only one year of cloud services. Considering autonomous driving is a long-term R&D initiative with ongoing generative AI exploration, cloud-heavy architectures would dramatically increase operational costs.
Spingence’s Edgestar platform integrates PHISON’s aiDAPTIV+ technology, which expands effective DRAM capacity using SSDs—addressing memory bottlenecks common in large-model training. Combined with the SKY-602E3 GPU server, the solution enables full-parameter fine-tuning even under resource constraints, delivering both stability and high performance for long-term research workloads.
Why Advantech SKY-602E3 is Ideal for Autonomous AI R&D
Advantech’s SKY-602E3 tower GPU server stood out to ARC for three core strengths:
compact form factor, high expansion flexibility, and long-term supply assurance.
Despite server-grade hardware, its small footprint allows deployment in standard office environments without a dedicated server room. The system supports flexible GPU options—from cost-efficient NVIDIA® RTX™ 4000 Ada Generation to high-end PRO™ 6000 Blackwell Series GPUs—and scalable storage through both U.2 and M.2 interfaces.
Advantech’s 5-year warranty and 5–7-year supply guarantee ensure smooth long-term operation, reducing downtime risks and preventing unnecessary system replacement—a crucial factor for multi-year AI research programs.
Why Advantech SKY-602E3 is Ideal for Autonomous AI R&D
Beyond hardware capabilities, Advantech’s brand reputation and responsive technical support played a key role in the success of this project.
According to Spingence CEO Jesse Chen, Advantech’s strong market trust and consistent product quality significantly enhance the persuasiveness of the Edgestar solution in enterprise deployments. Additionally, Advantech’s rapid response to integration and customization requests helped ensure a smooth installation and testing process for NTT ARC.
The collaboration between NTT DATA ARC, Advantech, Spingence, and FSR represents a successful example of cross-border technological synergy. By combining Advantech’s SKY-602E3 GPU server, Spingence’s Edgestar on-premise AI platform, and FSR’s local integration expertise, the project enables NTT ARC to explore generative AI applications in autonomous driving with greater cost efficiency, scalability, and operational stability—paving the way toward safer and more intelligent mobility systems.
System Diagram
-25/SKY-602E320240201102210.png)
