AEWIN Introduces Cutting-Edge AI Servers at OCP 2026 with Advanced Cooling Solutions

AEWIN Technologies Unveils Revolutionary AI Servers at OCP 2026

In an exciting announcement from San Jose, California, AEWIN Technologies, a part of the Qisda Group, is set to reveal its latest flagship modular systems based on the DC-MHS framework during the OCP 2026 Global Summit. This event, taking place from October 12 to 15, will highlight AEWIN's innovative next-generation servers and advanced two-phase direct liquid cooling (2P DLC) solutions tailored for rack-scale architectures. These offerings are a testament to AEWIN’s commitment to enhancing AI infrastructure.

AEWIN’s strategic focus aligns closely with Qisda Group's goals in AI infrastructure, leveraging its extensive expertise in computing, networking, storage, and thermal management. The company aims to deliver a flexible and energy-efficient infrastructure capable of meeting the evolving demands from AI factories.

Next-Gen Server Performance

The flagship servers feature high-performance AMD EPYC™ 9006-series CPUs, combining powerful computing capabilities with PCIe Gen6 connectivity suited for the AI era. Recognizing the thermal hurdles induced by increasing computational power, AEWIN has ingeniously integrated its 2P DLC solutions to manage high-density infrastructures exceeding 200 kW per rack. This innovative approach ensures a reduced Power Usage Effectiveness (PUE) and promotes long-term sustainability.

The BG18-A10710 models are designed for extreme computing density, including support for CPUs with 600 W thermal design power (TDP) and two 600-W GPUs within a compact 1U form factor. Similarly, the 2U AI server model BG28-A10710 hosts four FHFL-PCIe-6.0-x16 slots, catering to the need for robust GPU support. The crown jewel of AEWIN’s showcase is the BG48-A20813, a 4U AI server accommodating up to eight FHFL GPUs, specifically optimized for sophisticated agent-based AI applications and next-gen model training.

For handling data-intensive workloads, the BS28-A10710 offers up to 16 E3.S drive bays, while the BX28-A10710 is capable of accommodating up to 12 U.2 NVMe SSDs or 24 SATA drives. This flexibility allows for customizable storage configurations which are essential for AI data pipelines and intelligent data management solutions.

Reliable Foundation for Future Growth

The innovative systems incorporate OCP DC-SCM2 based system management controlled by the ASPEED AST2700 BMC, along with redundant 80 PLUS Titanium power supplies. This infrastructure foundation is designed to support AI inference, high-performance computing (HPC), cloud-native, and Neo cloud deployments efficiently.

By leveraging the modular, DC-MHS architecture alongside a rack-scale approach, AEWIN underscores its capability to provide powerful, scalable, and reliable servers specifically for AI infrastructure needs. The partnership with Arivor for 2P DLC technology, combined with the holistic technology ecosystem of the Qisda Group, effectively addresses the increasing appetite for AI computational power, high-speed connectivity, storage scalability, and thermal efficiency. This innovative ecosystem empowers organizations in their quest to establish sustainable and high-performance AI infrastructures.

Attendees are encouraged to visit AEWIN at Booth D9 during the OCP Global Summit 2026 from October 12 to 15 to explore these groundbreaking solutions in person. More information about AEWIN Technologies can be found at aewin.com.