UfiSpace Debuted Liquid-Cooled 1.6T Networking Platform
The hardware aims to address the thermal constraints of high-density AI clusters planned for 2027 and 2028 deployment.
Updated on Oct. 13, 2026 in Data Centers
Live Poll
Do you trust that new AI infrastructure technology will successfully reduce long-term data center energy consumption?
At the 2026 OCP Global Summit, UfiSpace demonstrated a liquid-cooled 1.6T networking platform, the S9331-64HO3L. This system is designed for next-generation AI data center architectures.
Why it matters
As AI workloads push server power densities higher, thermal management has become the primary bottleneck for infrastructure scaling. This hardware targets that constraint specifically to enable higher-performance inter-cluster communication.
The S9331-64HO3L platform features 64 ports of 1.6T OSFP interfaces, supported by a dedicated Cooling Distribution Unit. This hardware complements existing air-cooled platforms, such as the 102.4Tbps 3RU S9331-64HO and the deep-buffered 51.2Tbps S9630-32HO router.
The players
UfiSpace
A provider of open network white-box switching solutions that builds hardware for AI and data center connectivity.
GeNopsys Technologies
An optical networking firm focused on integrated switching solutions that minimize signal conversion latency.
The details
The platform utilizes a Cooling Distribution Unit—a component that manages the flow of coolant to regulate heat at the rack level—to validate system-level thermal feasibility in high-density environments. UfiSpace also collaborated with GeNopsys Technologies on an integrated Ethernet-optical solution, which uses dynamic optical switching to bypass O-E-O conversion—the process where signals are converted between optical and electrical formats, which often adds latency and power overhead.
Timeline
October 12-15, 2026: OCP Global Summit held in San Jose.
2027/2028: Expected deployment window for AI clusters using liquid-cooled architectures.
The Tech Race
The OCP Global Summit serves as the primary stage for vendors to standardize the high-density interconnects required for AI infrastructure. UfiSpace is currently competing to define the physical cooling and throughput standards that will underpin massive-scale GPU clusters.
These platforms are specialized enterprise hardware intended for data center operators building 2027 and 2028 AI clusters. The infrastructure changes focus on eliminating packet loss in AI workloads through deep buffering and cooling systems, rather than direct consumer-facing hardware.
The takeaway
The industry is shifting toward liquid cooling as a mandatory requirement for high-speed AI interconnects. Watch for the 2027 rollout of these platforms to see how standardizing 1.6T OSFP ports impacts real-world latency in massive GPU clusters.
Further reading
Explore the ongoing shift toward specialized hardware in Data Centers to support the increasing thermal and bandwidth requirements of modern AI systems.
More information
View the technical specifications for these new platforms on the UfiSpace AI networking solutions portal.
Source note: This article includes information reported by Macau Business.
Live Poll
Do you trust that new AI infrastructure technology will successfully reduce long-term data center energy consumption?
