YY Group Holding announced on Tuesday a strategic investment in deploying high-performance, on-premises computing infrastructure equipped with NVIDIA GeForce RTX 5090 GPUs, which are based on the NVIDIA Blackwell architecture. This dedicated computing facility will be used for the company's internal development, fine-tuning, and accelerated deployment of its proprietary vertical workforce AI models.

YY Group Holding Limited (ASX: YYGH)

Chief Executive Officer Mike Fu of YY Group commented, "True workforce AI requires more than generic, automated API wrappers; it demands vertical models trained on high-fidelity operational data and supported by dedicated hardware. By bringing NVIDIA's specialized Tensor Core compute power directly into our internal development flow, our engineering teams gain the processing speed required for local fine-tuning and testing of enterprise-grade AI. Leveraging the NVIDIA CUDA ecosystem, we are accelerating the build of a smarter, more scalable, and cost-effective global workforce management ecosystem."

The investment focuses on three clear technical objectives. First, it aims to utilize the NVIDIA CUDA platform for local large language model customization on open-source weight models ranging from 7B to 140 billion parameters, employing advanced optimization techniques like QLoRA to capture regional workforce nuances. Second, it plans to leverage NVIDIA TensorRT to accelerate candidate matching and re-ranking, providing enterprise clients with ultra-low latency, context-aware candidate shortlists. Third, it involves deploying high-throughput internal API endpoints by tightly integrating open-source inference engines like vLLM with NVIDIA hardware acceleration, for seamless integration into YYGH's existing product suite.

Headquartered in Singapore, YY Group's YY Circle platform operates across 12 countries and regions in Asia and globally. The company is transitioning from a workforce solutions provider to a global, AI-native infrastructure platform for the hospitality and service industries. It recently launched its first AI product, Yolara, and is advancing the deployment of its agent AI, OpenClaw. The company has reaffirmed its revenue guidance for the 2026 fiscal year, projecting between $103 million and $110 million, with an expectation to achieve non-IFRS profitability.