Taiwan-based INFINITIX has established its South Korea subsidiary to strengthen local capabilities and expand its footprint in the country.
Leveraging the software company’s expertise in AI infrastructure management and GPU orchestration, the move aims to address issues surrounding fragmented GPU resources, low utilisation, and complex environments.
AI-Stack
INFINITIX’s flagship platform, AI-Stack, addresses such challenges through heterogeneous orchestration and its proprietary CTAs (Core Type Aware Scheduler) technology.
It combines GPU partitioning, aggregation, and cross-node scheduling to enable centralised control across NVIDIA and AMD GPUs and NPUs, increasing utilisation from ~30% to over 90%.
CTA’s ability to map workloads to specific compute cores (e.g., CUDA and Tensor cores) also optimises performance and enables concurrent execution on a single GPU, delivering a “zero-idle, zero-waste” compute model.
AI Cloud Service Platform
INFINITIX also introduces ixCSP (AI Cloud Service Platform), which aims to transform AI Stack orchestration into a fully commercialisable cloud architecture, enabling rapid deployment of AI-driven business models.
The solution enables new revenue streams, including GaaS, MaaS, and TaaS, supported by multi-tenant management, precision billing, and standardised APIs, not only optimising infrastructure but also monetising it.
Strategic partnerships
INFINITIX has partnered with local distributors such as Miruware, Tera Tec, DS&G, and NUMP to accelerate AI-Stack and ixCSP adoption in the country.
