From AI Servers to AI Factories: How 10 COMPUTEX Exhibitors Are Shaping Next-Generation AI Infrastructure

---

The NVIDIA AI Factory MGX Ecosystem showcase at COMPUTEX 2026. AI servers are evolving beyond standalone systems toward rack-scale architectures and complete AI infrastructure.

As demand for AI computing continues to grow, the scope of AI server development is expanding beyond individual systems to rack-scale architectures, high-speed networking, storage, and liquid cooling. At COMPUTEX 2026, exhibitors showcased a new generation of AI servers and infrastructure solutions.

Bringing together 10 exhibitors recently featured in the COMPUTEX AI Server series, this article highlights the products and technologies they showcased and explores how next-generation AI infrastructure is evolving—from computing platforms and system integration to liquid cooling.

 The NVIDIA Vera Rubin Generation Takes Shape
Next-generation AI computing platforms are becoming a major focus for server manufacturers. From large-model computing and high-density AI servers to manufacturing and system integration, ASUS, Compal, and Foxconn are positioning themselves across different parts of the NVIDIA Vera Rubin ecosystem.

ASUS: Bringing Trillion-Parameter Models to AI PODs
ASUS showcased the ASUS XA VR721-E3 for ASUS AI POD, featuring the NVIDIA Vera Rubin Platform and purpose-built for trillion-parameter models. It helps customers reduce deployment risks and accelerate their journey from planning to real-world implementation.

Compal: HGX Rubin NVL8 Powers High-Density AI Servers
Compal showcased high-density AI server solutions based on NVIDIA HGX Rubin NVL8, aligning with the NVIDIA Vera Rubin architecture and demonstrating its engineering readiness for next-generation AI supercomputing infrastructure.

Foxconn: A Broad AI Server Portfolio Highlights System Integration Capabilities
Foxconn showcased its manufacturing and system integration capabilities for the NVIDIA Vera Rubin NVL72, along with the latest AI server and high-speed interconnect technologies, including NVIDIA Groq 3 LPX, the NVIDIA Vera CPU platform, NVIDIA HGX Rubin NVL8, and NVIDIA MGX 2U and 4U modular systems.

AI Servers Shift Toward Rack-Scale Systems and Integration
As AI systems continue to scale, server manufacturers are extending their focus from individual platforms to rack-scale architectures and system integration. PEGATRON, Inventec, and Wiwynn all showcased NVIDIA Vera Rubin NVL72 solutions, highlighting different approaches to next-generation AI infrastructure through high-speed networking, rack-scale systems, and multi-platform integration.

PEGATRON: Integrating Next-Generation Servers with High-Speed Networking
PEGATRON showcased next-generation server platforms and high-speed networking architectures for AI factory and data center deployments. Its lineup included the NVIDIA Vera Rubin NVL72, NVIDIA HGX Rubin NVL8, and Liquid-Cooled NVIDIA RTX PRO AI GPU Server, supporting the high-performance computing, data transmission, and system reliability requirements of AI factories and data centers.

Inventec: Artemis III Targets Large-Scale Generative AI
Inventec extended its expertise into advanced rack-scale solutions with Artemis III, powered by NVIDIA Vera Rubin NVL72. Specifically engineered for large-scale generative AI applications, the system is now in full production.

Wiwynn: NVIDIA and AMD Platforms Expand Rack-Scale Infrastructure Options
Wiwynn showcased both NVIDIA Vera Rubin NVL72 and AMD Helios rack-scale solutions. Based on the OCP ORv3 open standard, AMD Helios integrates AMD Instinct MI400 Series GPUs with 6th Gen AMD EPYC CPUs, demonstrating Wiwynn’s capabilities from reference design to production.

From Infrastructure Integration to AI Factory Deployment
AI infrastructure is evolving beyond standalone computing systems toward more comprehensive integration and deployment. QCT focuses on the co-design of compute, networking, and storage, while GIGABYTE approaches AI factory deployment through pre-deployment validation and optimization—highlighting different stages in the journey from infrastructure development to deployment readiness.

QCT: Extreme Co-Design Across Compute, Networking, and Storage
QCT showcased extreme co-design across seven chips spanning compute, networking, and storage, along with five purpose-built rack-scale systems for Agentic AI workloads: NVIDIA Vera Rubin NVL72 Built by QCT, NVIDIA Vera CPU Rack, NVIDIA Groq 3 LPX Inference Accelerator Racks, NVIDIA BlueField-4 STX AI-Native Storage Rack, and NVIDIA Spectrum-6 SPX Networking Racks.

GIGABYTE: GAIFA Strengthens Pre-Deployment Validation and Optimization
In addition to showcasing rack-scale AI infrastructure including NVIDIA Vera Rubin NVL72, GIGABYTE presented GAIFA (GIGABYTE AI Factory Accelerator). By integrating next-generation computing platforms, high-speed networking, and the NVIDIA software ecosystem, GAIFA helps customers conduct system validation, performance tuning, and workload optimization before full deployment.

High-Density AI Computing Meets Liquid Cooling
As AI computing density continues to increase, thermal management and data center space are becoming critical considerations in system design. From high-density liquid-cooled racks to coolant distribution units, liquid cooling is becoming increasingly integrated into AI server and rack-scale architectures.

MiTAC Computing: 52U Liquid-Cooled Rack Increases GPU Density
MiTAC Computing debuted its 52U High-Density AI Liquid-Cooled Rack, integrating 96 AMD Instinct MI355X GPUs in a single rack. Compared with standard AI configurations, the solution delivers 50% higher GPU density while reducing data center floor space by 33%. It supports a range of AI workloads, including model training, inference, and retrieval-augmented generation (RAG).

ASRock Rack: Liquid-Cooling CDU Supports Next-Generation AI Factories
ASRock Rack showcased AI server platforms integrated with advanced liquid-cooling solutions from ecosystem partners. Its exhibit highlighted the NVIDIA Vera Rubin NVL72, cooled with a double-rack-width liquid-to-air coolant distribution unit (CDU) to meet the requirements of next-generation AI factory environments.

From next-generation computing platforms and rack-scale systems such as NVIDIA Vera Rubin and AMD Helios to high-speed networking, AI storage, deployment validation, and liquid cooling, the technologies showcased by COMPUTEX 2026 exhibitors illustrate how AI servers are evolving beyond individual systems toward complete AI infrastructure and AI factory deployment.

COMPUTEX 2027 will take place from June 1–4, 2027. Follow the official COMPUTEX social media channels for the latest updates from the show and its exhibitors.

COMPUTEX YouTube https://www.youtube.com/@COMPUTEXTAIPEIshow

COMPUTEX Facebook https://www.facebook.com/ComputexTaipei/

COMPUTEX LinkedIn https://www.linkedin.com/company/computex-taipei

Back