HPE与NVIDIA的AI工厂面向智能体时代扩展

<div id="bsf_rt_marker"></div><p><span style="font-weight: 400;">企业正将智能体AI从概念验证推向生产——下一代AI工厂专为智能体时代而构建。</span></p> <p><span style="font-weight: 400;">在持续至6月18日（星期四）的HPE Discover拉斯维加斯大会上，NVIDIA与HPE正在扩展<a target="_blank" href="https://www.hpe.com/us/en/newsroom/press-release/2026/06/hpe-brings-agentic-ai-into-production-with-nvidia-delivering-security-governance-scale-and-sovereignty.html">HPE AI Factory with NVIDIA</a>，包括</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/vera-cpu/"><span style="font-weight: 400;">NVIDIA Vera CPU</span></a><span style="font-weight: 400;">和NVIDIA Agent Toolkit for HPE Private Cloud AI。</span></p> <p><span style="font-weight: 400;">此外，</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/solutions/confidential-computing/"><span style="font-weight: 400;">NVIDIA机密计算</span></a><span style="font-weight: 400;">已扩展至HPE AI Factory，增强的全栈NVIDIA集成——包括NVIDIA加速计算、NVIDIA AI软件和NVIDIA网络——现已覆盖整个产品组合。</span></p> <h2><b>NVIDIA Vera CPU现已支持HPE Private Cloud AI</b></h2> <p><span style="font-weight: 400;">搭载NVIDIA Vera CPU的HPE ProLiant Compute DL394 Gen12将于2027年随HPE Private Cloud AI（与NVIDIA联合设计的交钥匙AI工厂）上市。Vera是首款为智能体打造的CPU——专为智能体循环中所需的工具调用、编排和实时数据处理而设计——为HPE Private Cloud AI带来了确定性、低延迟的性能。</span></p> <p><span style="font-weight: 400;">纽约证券交易所（</span><a target="_blank" href="https://nvidianews.nvidia.com/news/nvidia-unveils-vera-the-cpu-for-agents"><span style="font-weight: 400;">New York Stock Exchange</span></a><span style="font-weight: 400;">）与Redpanda和HPE合作，成为探索搭载Vera CPU的HPE ProLiant Compute DL394 Gen12服务器的早期企业客户。</span></p> <p><span style="font-weight: 400;">Vera CPU是</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/technologies/rubin/"><span style="font-weight: 400;">NVIDIA Vera Rubin</span></a><span style="font-weight: 400;">平台的一部分，该平台正随着HPE提供的</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/"><span style="font-weight: 400;">NVIDIA Vera Rubin NVL72</span></a><span style="font-weight: 400;">机架级系统全面投产。Vera Rubin专为超过1万亿参数的前沿规模模型设计，并将随附全栈NVIDIA机密计算，覆盖每个芯片。</span></p> <p><span style="font-weight: 400;">HPE还将基于</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/hgx/"><span style="font-weight: 400;">NVIDIA HGX Rubin NVL8</span></a><span style="font-weight: 400;">构建的HPE Compute XD700引入HPE AI Factory，每个机架最多支持128个Rubin GPU。</span></p> <h2><b>NVIDIA Agent Toolkit现已支持HPE Private Cloud AI</b></h2> <p><span style="font-weight: 400;">NVIDIA Agent Toolkit——包括</span><a target="_blank" href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/"><span style="font-weight: 400;">NVIDIA Nemotron</span></a><span style="font-weight: 400;">开放模型、</span><a target="_blank" href="https://build.nvidia.com/openshell"><span style="font-weight: 400;">NVIDIA OpenShell</span></a><span style="font-weight: 400;">安全运行时和</span><a target="_blank" href="https://www.nvidia.com/en-us/ai/nemoclaw/?ncid=pa-srch-goog-984177&amp;_bt=804567865336&amp;_bk=nvidia%20openshell&amp;_bm=p&amp;_bn=g&amp;_bg=197993095849&amp;gad_source=1&amp;gad_campaignid=23744621431&amp;gbraid=0AAAAAD4XAoHIFX8bQyKrlsJhCkKKKHStz&amp;gclid=Cj0KCQjwornRBhCrARIsAON5exFPtIQ0D5rmHmh1khvp2JFlEp8HEABcgrFXZc4JYwm4XPXrAdJ4vgcaAtgoEALw_wcB"><span style="font-weight: 400;">NVIDIA NemoClaw</span></a><span style="font-weight: 400;">蓝图——将随HPE Private Cloud AI提供。这些工具共同为企业提供了一个智能体AI操作系统，用于监控智能体行为、执行治理策略，并安全地构建和运行自主、长期运行的多智能体系统。</span></p> <p><span style="font-weight: 400;">HPE Private Cloud AI新增了安全的本地智能体注册功能，允许客户在运行之前根据集中治理和安全策略审批AI模型、技能和工具。新的HPE Zerto软件功能可检测恶意智能体行为，并使用持续数据保护回滚到干净状态。</span></p> <p><span style="font-weight: 400;">在数据方面，达到</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/products/certified-storage/"><span style="font-weight: 400;">NVIDIA认证存储</span></a><span style="font-weight: 400;">基础级别的HPE Alletra Storage MP X10000自动应用元数据和治理策略，为非结构化数据准备AI管道，从而提升令牌吞吐量。</span></p> <h2><b>NVIDIA机密计算覆盖所有HPE AI Factory解决方案</b></h2> <p><span style="font-weight: 400;">通过HPE Services，NVIDIA机密计算现已覆盖整个HPE AI Factory——包括HPE AI Factory at Scale、HPE Sovereign AI Factory和</span><a target="_blank" href="https://www.hpe.com/us/en/newsroom/press-release/2026/03/hpe-accelerates-secure-scalable-production-ready-ai-through-new-innovations-with-nvidia.html"><span style="font-weight: 400;">HPE Private Cloud AI</span></a><span style="font-weight: 400;">。</span></p> <p><span style="font-weight: 400;">AI应用程序访问和使用需要保护和安全的私密及敏感数据。此外，使用专有数据或技术训练的模型需要防止泄露。机密计算对于这些现代AI工作负载至关重要，因为它能保护本地和主权部署中执行期间的模型和私有数据，通过加密证明和每阶段的加密建立信任链。</span></p> <p><span style="font-weight: 400;">此外，HPE ProLiant Compute DL380a获得了</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/products/certified-systems/"><span style="font-weight: 400;">NVIDIA认证系统</span></a><span style="font-weight: 400;"> for NVIDIA Confidential Computing项目的认证，该认证验证了机密计算下的稳健应用性能。这些系统为AI工作负载和敏感数据资产提供基于硬件的保护，同时保持最佳的NVIDIA加速性能。</span></p> <p><span style="font-weight: 400;">在HPE AI Factory解决方案中，</span><a target="_blank" href="https://www.nvidia.com/en-us/networking/products/data-processing-unit/"><span style="font-weight: 400;">NVIDIA BlueField DPU</span></a><span style="font-weight: 400;">和</span><a target="_blank" href="https://www.nvidia.com/en-us/networking/products/software/doca/"><span style="font-weight: 400;">NVIDIA DOCA</span></a><span style="font-weight: 400;">提供芯片内的零信任策略执行、运行时威胁检测和网络加密——在不牺牲性能的前提下保护AI工作负载、智能体和数据。</span></p> <h2><b>全产品组合增强型全栈NVIDIA集成</b></h2> <p><span style="font-weight: 400;">HPE AI Factory at Scale、HPE Sovereign AI Factory和HPE Private Cloud AI现已支持</span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/rtx-pro-6000-blackwell-server-edition/"><span style="font-weigh