From June 2–5, COMPUTEX 2026 brought together the global technology industry in Taipei under the theme "AI Together." As one of the world's leading technology exhibitions, the event showcased the latest advances in AI computing, robotics, and next-generation technologies.

At the exhibition, Gooxi presented its latest portfolio of AI servers designed for large-model training, inference, high-performance computing, and high-density storage. With AI adoption accelerating across industries, Gooxi continues to deliver the infrastructure needed to support increasingly demanding workloads and large-scale AI deployment.

G6860T5: Optimized for AI Inference

Designed for large-model inference and high-concurrency AI services, Gooxi G6860T5 supports dual 4th/5th Gen Intel® Xeon® Scalable processors and up to eight 600W GPUs. Compatible with over 90% of mainstream triple-fan accelerator cards, the platform combines powerful performance with deployment flexibility.

Its innovative cooling architecture ensures efficient thermal management for high-density GPU configurations, while flexible CPU-GPU interconnect options help improve multi-GPU communication efficiency. Equipped with 32 DDR5-5600 memory slots, 12 front-accessible drive bays, and 13 PCIe expansion slots, the G6860T5 delivers the scalability required for modern AI inference workloads.

G6460A5: High-Density Compute for AI Training

Built for AI training environments, the Gooxi G6460A5 supports dual AMD EPYC™ 9004/9005 series processors with up to 500W TDP per CPU. The platform offers 24 DDR5 memory slots, 14 PCIe expansion slots, and support for up to eight dual-width GPUs, providing the performance and flexibility needed for large-scale model training.

With support for high-speed InfiniBand networking, the G6460A5 enables efficient multi-node scaling for demanding AI clusters and training environments.

G6A80A5: Open and Flexible AI Computing

Gooxi G6A80A5 OAM Training & Inference Server is built around an open OAM architecture, enabling broad compatibility across today's diverse AI accelerator ecosystem.

Supporting OAM 2.0 modules such as MuXi C550, Hygon DCU 3, Kunlun P800, and HGX-based architectures, the platform simplifies deployment while reducing migration costs. Its forward-compatible design also prepares organizations for future accelerator upgrades.

R6480T5: High-Density Storage for AI Workloads

To address growing AI data requirements, Gooxi showcased the R6480T5, a 4U 60-bay high-density storage server engineered for data-intensive environments.

Powered by dual 4th/5th Gen Intel® Xeon® Scalable processors and 32 DDR5 memory slots, the system supports AI training, data preprocessing, and distributed storage workloads. With 60 front-accessible 3.5-inch drive bays, 10 front 2.5-inch drive bays, and four M.2 SSD slots, it delivers petabyte-scale storage capacity in a compact footprint.

Advancing AI Infrastructure

As AI continues to transform industries, efficient and scalable infrastructure has become more critical than ever. Through continuous innovation in compute, storage, and accelerator compatibility, Gooxi is helping organizations build the foundation for next-generation AI applications.

By delivering solutions that span training, inference, and data infrastructure, Gooxi remains committed to enabling faster AI deployment and unlocking greater value from intelligent computing.