en.Wedoany.com Reported - On May 26, Taiwan, China-based AI infrastructure software company INFINITIX announced it will showcase its AI-Stack and ixCSP dual platform stack at COMPUTEX 2026, under the theme "From AI Infra to AI Cloud Economy." The company will exhibit at Booth R0113, 4th Floor, Hall 2 of the Taipei Nangang Exhibition Center, presenting solutions for heterogeneous computing power management, model deployment, and AI cloud service commercialization.

The core of this showcase is transforming enterprise-owned GPUs, NPUs, and AI computing resources from "hardware assets" into manageable, schedulable, and billable AI cloud capabilities. With the rapid growth in demand for large model training and inference, enterprises face challenges beyond just "having GPUs or not"—they must also address operational issues such as resource utilization, tenant isolation, scheduling efficiency, model service deployment, billing metering, and cost recovery. INFINITIX noted in its release that enterprise GPU utilization often falls below 30%, and its product direction is precisely aimed at converting underutilized computing power into governed, revenue-generating AI service infrastructure.

AI-Stack is the heterogeneous computing power orchestration platform featured in this showcase. The platform natively supports Kubernetes, covers resources including NVIDIA and AMD GPUs, NPUs, and Phison aiDAPTIV+, and offers capabilities such as GPU partitioning and aggregation, cross-node computing, multi-tenant management, and real-time visual monitoring, while integrating with frameworks like TensorFlow, PyTorch, and Slurm. For data centers, enterprise AI teams, and cloud service providers, the value of such a platform lies in unifying computing power from different sources, architectures, and workload characteristics into a scheduling system, thereby reducing AI cluster resource fragmentation.

CTAs Scheduler is a key scheduling technology highlighted by INFINITIX. This technology can differentiate between CUDA Core and Tensor Core workloads and perform scheduling optimization focused on parallel efficiency and resource utilization. The company will also showcase a collaborative solution with Phison Electronics, combining aiDAPTIV+ intelligent storage with AI-Stack to support large language model training and inference, aiming to lower the total cost of ownership for AI infrastructure without complete reliance on HBM GPU procurement.

ixCSP is designed for AI cloud operations and commercialization. The platform's full name is AI Cloud Operations Solution Platform, integrating AI Gateway, BOSS billing, and AI-Stack resource management, supporting deployment models such as GaaS, MaaS, and TaaS. It helps enterprises, telecom operators, and data centers transform existing GPU assets into billable, governed AI cloud platforms. As Token-based billing gradually becomes a key metric in AI service operations, computing power platforms need to shift from traditional machine-time or instance sales toward refined operations centered on Token traffic management, inference efficiency, and FinOps.

The industrial value of such platforms is concentrated in the AI infrastructure operations layer. In the past, when enterprises built AI clusters, the focus was often on procuring servers, GPUs, networks, and storage; however, when model deployment enters multi-department, multi-business, and multi-tenant scenarios, what truly impacts input-output ratios is whether resources can be continuously scheduled, costs accurately accounted for, model services scaled on demand, and idle computing power converted into external services. INFINITIX's combined showcase of AI-Stack and ixCSP indicates that AI infrastructure competition is shifting from "computing power stacking" to "computing power governance and commercial operations."

INFINITIX stated that it is an NVIDIA certified partner and an AMD GPU ecosystem partner, and is expanding from its R&D core in Taiwan, China, to Japan, South Korea, and Southeast Asia. The company will present a three-layer architecture covering AI infrastructure, AI platforms, and the AI cloud economy during COMPUTEX 2026, aiming to cover the complete chain from underlying computing power management to AI service monetization.

Subsequent project milestones include live demonstrations at COMPUTEX 2026, validation of AI-Stack and ixCSP among telecom operator and data center clients, application feedback on the Phison Electronics solution, and whether GPU resource management, Token billing, and AI cloud service operations can form a replicable deployment. At this stage, it can be confirmed that INFINITIX has disclosed its exhibition plans and dual platform stack capabilities; publicly available information has not disclosed new customer lists, contract amounts, deployed GPU scale, actual Token revenue, or commercial operational results. Therefore, it should not be extrapolated that the related platforms have already generated confirmed revenue.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com