The 2026 AGIC Shenzhen International General Artificial Intelligence Industry Expo & GERX Global Embodied Intelligence Robot Industry Expo opened on August 26 at the Shenzhen World Exhibition & Convention Center, bringing together the latest developments in embodied intelligence, humanoid robots, large models, intelligent devices, and other emerging AI applications.

From humanoid robots and live demonstrations of embodied intelligence to more than 100 new technology launches, AGIC 2026 highlighted a clear trend: AI is moving beyond the digital world and into real-world applications.

As robots become better at perceiving, reasoning, and acting, the infrastructure behind them is facing new demands. Model training, inference, multimodal processing, and real-time decision-making all require scalable computing and high-speed data processing.

Infrastructure Is Becoming Critical to AI Deployment

As embodied intelligence and AI applications expand, computing infrastructure needs to keep pace with growing workloads and increasingly diverse requirements.

Gooxi has been building its server capabilities around these needs. Under its “All-in-AI” strategy, the company combines expertise in motherboard design, system development, manufacturing, and solution delivery, with a portfolio spanning AI computing, general computing, edge computing, and storage.

Open Architecture for Diverse Workloads

AI workloads increasingly require flexible and heterogeneous computing environments. Gooxi works with chipmakers including Intel, AMD, Hygon, Loongson, and MetaX, supporting platforms from x86 to domestic architectures.

The Gooxi G6A80A5 OAM Training & Inference Server is based on the open OAM 2.0 architecture and supports mainstream AI accelerator modules from multiple vendors, providing a flexible foundation for heterogeneous AI infrastructure.

High-Density Computing for AI

Large-scale inference and model training continue to drive demand for high-density GPU infrastructure.

Gooxi G6660T5 supports two 5th Gen Intel® Xeon® Scalable processors, with up to 385W TDP per CPU, and up to eight 600W double-width GPUs. PCIe 5.0 and direct CPU-GPU connectivity enable high-bandwidth, low-latency communication for demanding multi-GPU workloads.

High-Density Storage for AI Data

AI also generates growing volumes of data that need to be stored and processed efficiently.

Gooxi R6480T5 packs 60 × 3.5" drives, 10 × 2.5" front-access drives, and four M.2 SSDs into a 4U platform, delivering petabyte-scale storage capacity for AI training, data preprocessing, distributed storage, and other data-intensive workloads.

Built for Scale

AI infrastructure at scale requires not only performance, but also reliable manufacturing and delivery.

Gooxi's 30,000-square-meter manufacturing facility in Dalingshan, Dongguan has an annual server production capacity of up to 320,000 units, with system-level customization and rapid delivery capabilities. Gooxi currently serves more than 2,000 customers across 30+ countries and regions.

AGIC 2026 reflects the broader shift from AI demonstrations to real-world deployment. As embodied intelligence, large models, and AI applications continue to advance, scalable and reliable computing infrastructure will remain essential.

Gooxi will continue to develop flexible and reliable server solutions to support the evolving demands of AI and digital transformation.