英伟达与Rebellions、SK电讯合作韩国主权超级智能（SI）：R200神经处理器接入NVLink Fusion，系统预计2028年可部署
英伟达11月8日宣布与韩国超级智能（SI）芯片企业 Rebellions 及电信运营商 SK 电讯（SK Telecom）合作，推进韩国主权超级智能（SI）基础设施。按合作安排，Rebellions 将把其下一代 Rebel R200 神经处理单元（NPU）通过英伟达 NVLink Fusion 接入英伟达平台，并在英伟达 MGX 机架架构下开发机架级系统；SK 电讯计划把相关系统部署到其超级智能（SI）数据中心，并与自有基础模型及面向公众的服务衔接。
英伟达称，NVLink Fusion 把英伟达平台扩展到定制 XPU，使专业芯片厂商可以使用英伟达的高带宽、低延迟纵向扩展网络与基础设施生态。Rebellions 表示，R200 面向前沿模型的低延迟、高能效推理设计，覆盖多万亿参数的混合专家模型；该公司现有旗舰产品为采用芯粒架构与开源软件栈的推理芯片 Rebel100。按三方时间表，R200 系统预计在2028年准备好部署，本次公告未披露部署规模、投资金额、订单数量与性能实测数据。
韩国科学技术信息通信部对该合作提供支持。公告称，该部支持英伟达与韩国 NPU 企业之间的合作，首批即 Rebellions 与 SK 电讯；SK 电讯方面表示，将把韩国超级智能（SI）半导体与全球基础设施技术结合，以加强其主权超级智能（SI）竞争力。此前在6月7日，SK 电讯已公布使用英伟达 DSX 架构在韩国建设吉瓦级超级智能（SI）云的计划，本次处理器合作是该布局的后续一步。
NVIDIA Partners With Rebellions and SK Telecom on Korean Sovereign AI: Rebel R200 NPUs to Connect via NVLink Fusion, Systems Expected Ready for Deployment in 20
NVIDIA announced on November 8 a collaboration with Korean AI chip company Rebellions and telecommunications operator SK Telecom to advance South Korea's sovereign AI infrastructure. Under the arrangement, Rebellions will connect its next-generation Rebel R200 neural processing units (NPUs) to NVIDIA's platform through NVIDIA NVLink Fusion and develop rack-scale systems around the NVIDIA MGX rack architecture, while SK Telecom plans to deploy the resulting systems in its AI data center and connect them with its proprietary foundation models and public-facing services.
NVIDIA said NVLink Fusion extends its platform to custom XPUs, allowing specialized chipmakers to use its high-bandwidth, low-latency scale-up networking and infrastructure ecosystem. Rebellions is developing the R200 for low-latency, energy-efficient inference with frontier models, including multitrillion-parameter mixture-of-experts models; its current flagship is the Rebel100 inference chip, built on a chiplet architecture with an open-source software stack. The companies expect R200 systems to be ready for deployment in 2028. The announcement disclosed no deployment size, investment amount, order quantity or measured performance results.
Korea's Ministry of Science and ICT supports the collaboration, which it described as support for cooperation between NVIDIA and Korean NPU companies beginning with Rebellions and SK Telecom. SK Telecom said combining Korean AI semiconductors with global infrastructure technology would strengthen its sovereign AI competitiveness. The processor collaboration follows a separate June 7 announcement of SK Telecom's plan for a gigawatt-scale AI Cloud in Korea using NVIDIA's DSX architecture.
—-
来源：英伟达、Rebellions（公司
发布：GSM·美国（发布链接）
图片来源：数据中心芯片（资料图）。（Wikimedia Commons）|Data-center chips (file photo). (Wikimedia Commons)
GSM Global · 本篇英文版 English edition of this story
