VCI Global Limited, a diversified global holding company with a strategic focus on AI & Robotics, Fintech, Cybersecurity, Capital Market Consultancy, announced the launch of SecureGPU™, a next-generation encrypted AI compute server developed by its cybersecurity and infrastructure subsidiary, V Gallant Sdn Bhd.
SecureGPU™ is the first known server platform integrating GPU compute with real-time PCIe-level encryption, purpose-built for sovereign, defense, and regulated enterprise use. Designed to operate entirely within zero-trust, air-gapped environments, the server enables secure training and inference of AI models without relying on cloud infrastructure.
SecureGPU™ launches at a time of heightened geopolitical risk and regulatory tightening around AI, data governance, and infrastructure sovereignty. Governments and institutions worldwide are reassessing cloud dependency and investing in local, confidential compute environments.
Also Read: SeeTrue Launches Next-Gen AI X-ray Screening for Enhanced Security
According to Investopedia, analysts estimate the sovereign AI infrastructure market could reach approximately US$1.5 trillion globally, with strong demand in Europe and Asia-Pacific. Meanwhile, Mordor Intelligence projects that global AI infrastructure spending will grow to approximately US$197.6 billion by 2030, fueled by on-premises and secure deployment models.
SecureGPU™ aligns with VCI Global’s broader platform strategy by integrating natively with V-Gallant’s Vault orchestration platform and the QuantGold encrypted data exchange. This positions VCI Global to generate recurring revenue from a bundled stack of sovereign compute, encrypted storage, and AI lifecycle services.
Key Technical Features of SecureGPU™
- Real-time PCIe Hardware Encryption
A proprietary PCIe card encrypts data during training and inference, minimizing risk of model or data exposure.
- Zero-Trust Compute Architecture
Designed for isolated, on-premise deployment with no reliance on cloud infrastructure.
- Support for LLM and Regulated Data Use Cases
Enables secure training and deployment of large language models on sensitive or classified datasets.
- Modular Rack-Based Design
Scalable configuration supports both pilot and full-scale data center deployments.
- Platform Integration
Seamlessly connects with V Gallant’s Vault and QuantGold for secure AI orchestration and data exchange.
- Hardware Root-of-Trust
Includes secure boot, encrypted firmware, and tamper alerts for physical and firmware-level protection.
Built on NVIDIA®-class GPU architecture, SecureGPU™ delivers enterprise-grade AI performance with full data locality, auditability, and compliance control.
SecureGPU™ is currently available for pilot deployment with government and institutional partners in Malaysia, Singapore, and select international markets. The solution is positioned as a cornerstone of VCIG’s sovereign AI stack and is expected to contribute to the company’s long-term platform monetization and partner expansion strategy.
“With SecureGPU™, we are delivering a breakthrough in sovereign AI infrastructure. In an era of cross-border data threats and AI model leakage, this product gives our partners the ability to deploy AI workloads securely, at scale, and with full hardware-level control,” said Dato’ Victor Hoo, Group Executive Chairman and Chief Executive Officer of VCI Global.
Source: Globenewswire