The new MiTAC server with NVIDIA RTX PRO Blackwell GPUs and SuperNIC connectivity redefines performance in dense AI and LLM inference environments.
The Taiwanese company MiTAC Computing Technology Corporation has unveiled its new G4527G6 server at COMPUTEX 2025, a platform designed to tackle the increasing challenges of AI-accelerated computing in data centers and enterprise environments. Built on the NVIDIA MGX modular architecture, the G4527G6 supports configurations with up to 8 NVIDIA RTX PRO 6000 Blackwell Server Edition or NVIDIA H200 NVL GPUs, in addition to a high-performance connectivity network based on 4 SuperNICs ConnectX-8 capable of reaching up to 800 Gb/s.
This new server is specifically designed for generative AI, large-scale inference, and HPC applications, with a particular focus on thermal efficiency, extreme performance, and scalability. Its design is part of NVIDIA’s Enterprise AI Factory initiative, offering a validated solution ready for large-scale enterprise deployments.
G4527G6: An Infrastructure Ready for Modern AI Challenges
The MiTAC G4527G6 is based on Intel Xeon 6 processors, integrates up to 8 TB of DDR5-6400 memory, and includes 16 E1.S hot-swap drive bays. In terms of connectivity, its architecture features a NVIDIA BlueField-3 DPU, offering efficient north-south connectivity, along with the aforementioned ConnectX-8 SuperNICs with next-generation InfiniBand and Ethernet technology.
This creates an ideal platform for tasks requiring extreme bandwidth and low latency, such as LLM inference with trillions of parameters, autonomous agents, and real-time data analysis. Additionally, the use of In-Network Computing technologies such as NVIDIA SHARP allows for the acceleration of complex processes directly within the network.
NVIDIA-Validated Enterprise AI: Performance and Modularity
The server positions itself as one of the most advanced solutions for building “AI factories” in on-premise environments, providing over 100 modular configurations thanks to the flexible design of NVIDIA MGX. The NVIDIA RTX PRO 6000 Blackwell Server Edition and H200 NVL GPUs offer:
- Up to 1.8 times faster LLM inference compared to previous generations
- Up to 1.3 times more HPC performance
- Scalability for physical AI workflows, autonomous decision-making, and enterprise agents
Thus, MiTAC joins the ranks of companies betting on AI as critical infrastructure for the business future.
MiTAC’s Commitment: Quality, Scalability, and Global Support
With decades of experience, MiTAC Computing stands out for its comprehensive approach covering from design to manufacturing and global support, with tailored solutions for edge computing, AI, HPC, and cloud. Its commitment to technologies such as liquid cooling, integration of Intel DSG and TYAN products, and quality at the system and rack level positions MiTAC as a key player in the evolution of smart data centers.
The G4527G6 is not just a technical advancement but also a strategic statement that reinforces the collaboration between MiTAC and NVIDIA to democratize access to powerful, efficient, and sustainable AI infrastructures.
The G4527G6 redefines what it means to be ready for AI in 2025: power, modularity, extreme connectivity, and the assurance of an infrastructure designed for the future.