TAIPEI, June 4, 2026 /PRNewswire/ — Cooler Master, a global leader in thermal solutions, and Spingence, a premier Taiwanese AI architecture developer, jointly announced their global AI manufacturing transformation achievements at NVIDIA GTC Taipei 2026. The partners have successfully deployed the NVIDIA three-computers for physical AI across four major production bases in Taiwan, China, Vietnam, and the United States. By deeply integrating AI visual inspection agents, thermal physics simulation, digital twins, and enterprise knowledge systems, they have established a closed-loop AI manufacturing ecosystem linking R&D, production, and simulation. This collaboration serves as a landmark case for Physical AI in manufacturing, demonstrating how AI can evolve from isolated applications into a core competitive advantage for cross-border collaboration.
Cooler Master CIO Jason Wu (left) and Spingence CEO Jesse Chen (right) served as speakers at the 2026 NVIDIA GTC Taipei Session Panel under the theme, “From Virtual to Reality: The Rise of Global AI Factories.” Photo courtesy of Spingence.
A Tripartite Architecture: Solving the Challenge of Global Process Consistency
As Generative AI and Physical AI accelerate their penetration into the industrial sector, multinational corporations face critical hurdles in maintaining “global quality consistency” and “technical knowledge transfer.” Jason Wu, CIO of Cooler Master, stated: “For more than 30 years, Cooler Master has accumulated deep expertise in thermal engineering and manufacturing operations. Our vision is to transform this experience from individual know-how into scalable enterprise intelligence.
Through AI agents, digital twins, and global data integration, we are building a manufacturing platform that continuously learns, evolves, and enables faster global execution.”
To achieve this, Spingence built an automated, continuously-learning feedback mechanism powered by Three Computers for physical AI:
- Training Computer: AI models are trained using decades of thermal engineering expertise, process knowledge, production data, and defect samples accumulated across global operations.
- Inference Computer: AI agents and inference systems are deployed directly on manufacturing lines worldwide, enabling real-time visual inspection, anomaly detection, operational assistance, and decision support.
- Digital Twin Computer: Using digital twins and virtual factory simulation, production line changes and factory optimization scenarios can be validated virtually before physical deployment, creating a continuous feedback loop between simulation and real-world manufacturing. When real training is limited, synthetic data generation is critical for accelerating model training time and accuracy.
Together, these three layers establish a closed-loop AI ecosystem capable of continuously improving production quality, operational efficiency, and global process standardization.
Integrating Software and Hardware to Realize the Value of Physical AI
Spingence integrated its AINavi AI defect detection platform, Edgestar enterprise on-premise LLM cluster management platform, and SpinZone digital twin platform to help Cooler Master build a robust AI infrastructure. Using the NVIDIA Factory Operations Blueprint (FOX), Spingence is building a centralized factory manager agent that connects, fine-tunes, and operates a diverse set of agents and applications to enable faster global execution.
Based on production data from Cooler Master factories, the transformation has yielded significant results:
- Automated Quality Management with AINavi: Implementing NVIDIA Defect Image Generation skill, NVIDIA Cosmos, and NVIDIA TAO for synthetic data generation and model fine-tuning achieved a 99.6% defect detection rate, increased inspection capacity by 3x, and reduced the escape rate by 78%.
- 100x R&D Efficiency: Utilizing NVIDIA PhysicsNeMo surrogate models, thermal simulation verification speeds increased a hundredfold, shortening sample development cycles by 60% and unifying inspection standards globally.
- Digital Twin-Optimized Reconfiguration with SpinZone: Through NVIDIA Omniverse libraries and NVIDIA Isaac, an open robotics platform, production line reconfiguration—which previously took weeks—was reduced to days, lowering trial-and-error costs by 50% and significantly reducing downtime losses.
These workloads can run on NVIDIA RTX PRO 6000 Blackwell Server Edition or NVIDIA FGX. a purpose-built industrial computing platform designed to support a multi-agent system for robotic factory operations.
From Manufacturing to Decision-Making: Building a 24/7 Knowledge System
Beyond shop floor optimization, Cooler Master is simultaneously driving a transformation in knowledge management. By integrating NVIDIA NeMo Curator, NVIDIA NeMo Guardrails , and the Edgestar platform, the time required to generate process reports has been slashed from hours to minutes. This establishes a 24-hour AI decision-support system, ensuring corporate intelligence can respond to market changes in real time.
Chen Ching-Wei, Founder and CEO of Spingence, noted: “The key to successful AI transformation lies in whether the technology truly fits existing factory workflows. Our partnership with Cooler Master empowers AI to grow alongside the enterprise, creating a feedback loop that makes the AI system smarter over time.”
Looking ahead, Spingence and Cooler Master will continue to deepen their collaboration, extending AI applications to intelligent supply chain decision-making and dynamic cross-plant resource scheduling, leading the global manufacturing industry toward becoming AI-native enterprises.
Media Contact
Veronica Wang / Manager of Brand Engagement, Spingence
[email protected] | 02-7723-8823 #31
