Aivres at OCP Global Summit 2026 Showcases Total AI Data Center Infrastructure to Accelerate Time to First Token

SAN JOSE, Calif.—October 12, 2026— Aivres, a data center servers and storage solutions provider, today announced at OCP Global Summit 2026 it is showcasing its end-to-end AI infrastructure solutions for building, deploying and scaling AI factories. The showcase centers on Aivres’ rack-scale and server platforms based on the latest NVIDIA accelerated computing architectures. Through October 15, Aivres will demonstrate its technologies in Booth B3 of the Main Expo Hall at the San Jose Convention Center under the theme “Total AI Infrastructure to Power AI Factories and Accelerate Time to First Token”.

Aivres AI data center (AIDC) infrastructure solutions and capabilities include:

Rack-Scale Agentic AI Supercomputer based on the NVIDIA DGX Vera Rubin NVL72 platform—Aivres KRS8000V4 is purpose-engineered for giga-scale data centers running trillion-parameter training and long-context inference workloads. It offers extreme compute density, integrating 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs in a single, fully liquid-cooled rack and delivering architectural peak performance (up to 2.5 EFLOPS for NVFP4 training and up to 3.6 EFLOPS for NVFP4 inference). KRS8000V4 provides up to 75TB of ultra-fast rack-local memory capacity, to accelerate data staging and checkpointing and support KV-cache expansion for long-context inference. The rack-scale solution offers full-stack networking and acceleration via NVIDIA NVLink 6 scale-up fabric (up to 3.6TB/s per GPU) combined with NVIDIA ConnectX-9 SuperNICS with RDMA for scale-out (up to 1.6Tb/s per GPU). NVIDIA BlueField-4 data processing unit (DPU) offloads storage and security functions to free CPU/GPU cycles for AI compute. Scalability is seamless from a single rack to multi-rack pods, and on to a full AI Factory. Utilizing NVIDIA Quantum-X800 InfiniBand and NVIDIA Spectrum-X Ethernet enables consistent peak performance and operational simplicity across deployments.

Fully Liquid-Cooled Next-Generation AI Compute Tray powered by NVIDIA Vera Rubin—To meet diverse deployment needs, Aivres presents the optimized KL2288 2U AI compute tray supporting the NVIDIA HGX™ Rubin NVL8 turnkey infrastructure solution, providing a blueprint for success in the age of agentic AI. Powered by eight NVIDIA Rubin GPUs, the platform delivers 400 PFLOPS of NVFP4 inference performance and 176TB/s of high-bandwidth memory (HBM) bandwidth, ensuring data is transported through infrastructure as quickly as possible. HGX Rubin NVL8 leverages sixth-generation NVIDIA NVLink to ensure seamless peer-to-peer communication for massive model parallelism. Aivres KL2288 is a 2U all-liquid-cooled, high-density system designed for large-scale deployments in modern, liquid-cooled data centers, pursuing ultimate PUE benefits. The system supports Intel and AMD CPU platforms, meeting enterprise compatibility and multi-task processing requirements.

Enterprise AI Server for Agentic AI to Digital Twins—For professional workloads at the edge of AI and advanced visual computing, Aivres will also feature the KR6268 6U PCIe system with powerful NVIDIA RTX PRO™ 6000 Blackwell Server Edition and NVIDIA RTX PRO 4500 Blackwell Server Edition GPU support, merging unmatched AI compute with cutting-edge visual performance and enabling broad application reach. KR6268 redefines possibilities across agentic AI, physical AI, scientific computing, photorealistic rendering, 3D graphics and real-time video processing.

Beyond its AI server and rack-scale platforms, Aivres delivers end-to-end capabilities to deploy AIDCs, helping enterprises and hyperscalers rapidly and reliably build AI factories from the ground up – from designing total infrastructure integrating NVIDIA AI platform solutions to planning and delivering fully liquid-cooled AI data center facilities. Engagements begin with full-lifecycle strategic planning spanning design, testing and operations, and cover the complete AIDC architecture, from system configuration and cluster topology to server, storage and liquid cooling infrastructure design. Through agile, lean manufacturing of optimized, pre-integrated solutions and modular deployment of prefabricated units, Aivres enables customers to bring AI data centers online faster than traditional builds, accelerating time to first token.

The full showcase of Aivres AIDC infrastructure platforms and solutions at OCP Global Summit 2026 are now available for deployment. Contact Aivres to learn more.

OCP Global Summit is the Open Compute Project Foundation’s flagship annual event, bringing together technology leaders, engineers and end users from across the global data center ecosystem to collaborate on open hardware and infrastructure innovation for the AI era. Click here for more information on the 2026 event.