Aivres at GTC 2025 to Demonstrate “Comprehensive AI Infrastructure to Empower Transformation at Every Scale”
From AI-Optimized Servers to Liquid-Cooled Racks to Data-Center Management, Aivres Delivers Solutions Leveraging NVIDIA Technology to Unleash AI
SAN JOSE, Calif.—March 18, 2025— Aivres, a data center and cloud computing solutions provider, announced that at GTC 2025, they will showcase “Comprehensive AI Infrastructure to Empower Transformation at Every Scale.” Aivres is a Diamond Sponsor of the March 17-21 event at the San Jose Convention Center and will demonstrate its technologies in Booth 522.
Aivres is working with NVIDIA to drive innovations in total AI infrastructure, from AI-optimized servers to liquid-cooled racks to data-center management, to support every scale of AI workload. Through integration and innovation with leading-edge NVIDIA AI technology, Aivres server platforms deliver breakthrough performance, efficiency and flexibility to tackle the most challenging AI workloads in the industry today—large models, machine learning, generative AI and beyond.
Aivres’ showcase at NVIDIA GTC 2025 will include three complete, liquid-cooled rack solutions for AI data centers along with an in-row CDU (coolant distribution unit):
- KRS8000 NVIDIA GB200 NVL72 Rack-Scale Solution—Aivres KRS8000 is an L11 AI rack based on the NVIDIA GB200 NVL72 platform, integrating 36 NVIDIA (™) CPUs and 72 NVIDIA Blackwell GPUs interconnected via NVLink in a rack-scale, liquid-cooled architecture. The KRS8000 achieves breakthrough performance in real-time, trillion-parameter large language model (LLM) inference and training.
- AI rack based on KR5288 server supporting NVIDIA HGX B200 8-GPU—Aivres KR5288 is a 5U, high-performance, heterogeneous accelerated server, featuring a fully modular design and NVIDIA NVLink(™) The entire system can support two CPUs with a TDP of 500W each and eight GPUs with a TDP of 1000W each. Direct-to-chip liquid cooling on CPU, GPU and NVSwitch enables a high-efficiency, high-density AI infrastructure in a space-saving form factor.
- AI rack based on KR6288 server supporting NVIDIA HGX H200 8-GPUs—Aivres KR6288 is an advanced, modular 6U AI system made for hyperscale data centers, delivering high performance with NVIDIA HGX H200 8-GPUs. This server delivers industry-leading 32 PFlops of AI performance and lightning-fast CPU-to-GPU interconnect bandwidth, with the H200 Transformer Engine supercharging training speeds for GPT LLMs. Its optimized power efficiency and modular design with flexible configuration makes it ideal for the most demanding AI tasks in various scenarios like hyperscale data centers, AI model training and metaverse workloads.
In addition, Aivres will display an AI-optimized networking switch – the X400 which is compatible with the KRS8000 NVIDIA GB200 NVL72 rack. Based on the NVIDIA Spectrum-X™ networking platform, X400 delivers generative-AI network performance accelerated by 1.6 times compared to traditional Ethernet fabrics. It offers peak AI workload efficiency with up to 400Gb/s RoCE network connectivity between GPU servers, ultra-high port density and scalability to connect hundreds of thousands of GPUs, as well as chip-, system- and link-level monitoring and fault isolation to ensure undisrupted operation.
Rounding out the Aivres showcase is a wide portfolio of AI server solutions for the data center.
To support large scale AI model training and inference of up to 400 billion parameters, Aivres’ most powerful AI systems are based on the advanced NVIDIA HGX 8-GPU platform: the KR9288 10U server and KR5288 5U server feature NVIDIA HGX B200 while the KR6288 6U features NVIDIA HGX H200.
For LLM inference and fine-tuning, model training up to 175 billion parameters, and advanced real-time graphics, Aivres PCIe GPU systems include: the KR6268 6U AI platform supporting 8 NVIDIA H200 NVL GPUs; the KR4268 4U and KR2280 2U servers supporting up to 8 NVIDIA L40S or NVIDIA H100 NVL GPUs.
NVIDIA GTC 2025 offers both in-person and virtual options for attendees. For more information on the event, please visit https://www.nvidia.com/gtc/.
About Aivres
Aivres is a data-center servers and storage solutions provider committed to delivering innovative technologies that propel the world’s leading industries to new frontiers. Aivres widely delivers and deploys cutting-edge hardware products and designs to major data centers, supporting critical modern applications that include cloud, AI, big data, mass storage and edge. The company’s unique JDM model—“joint design manufacturing”—enables open collaboration with partners and customers to develop purpose-built, performance-optimized solutions which tackle different workloads and overcome real-world challenges. For more information on Aivres, please visit https://aivres.com/.