---
格式版本: 2
标题: "NVIDIA to Showcase Blackwell and Networking Solutions at OCP Summit | NVIDIA Blog"
原文链接: "https://blogs.nvidia.com/blog/blackwell-networking-ocp-summit/"
发布日期: "2026-07-14"
发布时间校准状态: "found"
发布时间来源: "llm:original_extracted_text"
发布时间证据: "第 1 行：original_dom_semantic: time class=related-news-date nvidia-article-date datetime=2026-07-14T08:00:20-07:00: Jul 14, 2026"
发布时间校准原因: "候选日期位于标题下方/元数据区域，且为最新日期，符合文章发布时间特征。"
发布时间校准置信度: "1"
发布时间候选数量: 36
发布时间严格候选数量: 12
发布时间原页读取状态: "原页面来自已抓取 HTML"
发布时间未找到原因: "候选日期无效或 LLM 未确认"
发布时间校准时间: "2026-07-20T10:37:08+08:00"
发现时间: "2026-07-20T09:24:15+08:00"
入库时间: "2026-07-20T02:51:29.614Z"
来源平台: "NVIDIA Blog 搜索"
搜索渠道: "source_template"
搜索词: "https://blogs.nvidia.com/?s=Cold%20Plate"
匹配关键词:
  - "Cold Plate"
  - "GPU"
  - "Liquid Cooling"
  - "Marvell"
相关厂家:
  - "NVIDIA"
  - "AMD"
  - "Meta"
  - "Microsoft"
  - "Google"
  - "QCT"
  - "Wiwynn"
  - "Ingrasys"
相关专家:
  []
内容类型: "网页"
抓取工具: "AgentKey Scrape"
清洗工具: "AgentKey Markdown + LLM 正文裁剪"
原始附件:
  []
AI优质: "否"
AI打分: 62
AI分档: "召回候选"
AI质检状态: "不通过"
AI打分理由: "NVIDIA官方发布，主题涉及OCP峰会及AI机柜、液冷等，但正文主要为会议预告和议程列表，缺乏具体架构参数、量产部署或技术细节展开，信息较泛。"
AI质检模型: "qwen3.6-plus"
AI质检时间: "2026-07-20T10:51:29+08:00"
AI主题相关性: 15
AI来源权威性: 15
AI新颖性: 12
AI技术细节: 8
AI商业部署信号: 7
AI完整性: 5
图片摘要:
  - "★ ./assets/img-83d2f103.jpg | diagram | 展示大规模液冷服务器集群与互连管线，体现AI工厂的高密度部署，呼应正文的能效与扩展性方案。"
  - "✗ ./assets/img-1f6575f1.png | infographic | 图片主题为Vera CPU及单线程性能，正文主要介绍Blackwell、Grace CPU及OCP展示，主题不符。"
  - "✗ ./assets/img-3511966c.jpg | photo | 仅为NVIDIA品牌Logo及建筑实拍，属品牌宣传图，无具体技术信息。"
  - "✗ ./assets/img-e47e99dc.png | diagram | 图片主题为推理软件栈与Token成本，正文主要聚焦硬件展示（Blackwell、网络、液冷），主题不符。"
采集批次: "2026年7月20日9点23分34秒"
采集批次ID: "20260720-092334-062"
去重键: "https://blogs.nvidia.com/blog/blackwell-networking-ocp-summit"
---

A variety of open compute, networking, storage, automated provisioning, energy efficiency and security solutions will take center stage at the Open Compute Project (OCP) Global Summit, taking place Oct. 15-17 in San Jose, Calif.

Attendees, speakers and sponsors will present and discuss the impact of these technologies when run at cloud scale, with a special focus on AI workloads, liquid cooling, enhanced networking, accelerator management, memory fabrics and sustainability.

NVIDIA has long contributed to OCP designs and data center innovations. At the show, the company will showcase a wide range of OCP-related solutions, including AI factories, scalable GPU compute clusters, Arm-based CPUs and DPUs, storage and networking.

NVIDIA partners and customers will display and demo the latest NVIDIA accelerated computing platforms, including [Blackwell AI servers](https://www.nvidia.com/en-us/data-center/technologies/blackwell-architecture/), AI compute racks, liquid-cooled hardware, DPUs and SuperNICs, network switches and network operating systems.

Additionally showcased will be a new [NVIDIA Grace CPU](https://www.nvidia.com/en-us/data-center/grace-cpu/) C1 configuration — a single-socket Grace CPU Arm-based server that is ideal for hyperscale cloud, high-performance edge, telco and more — as well as [NVIDIA GH200 Grace Hopper](https://www.nvidia.com/en-us/data-center/grace-hopper-superchip/) and Grace Blackwell Superchips.

Partners showcasing NVIDIA technology and products at the OCP Global Summit include Aivres, Arm, ASRock Rack, ASUS, Dell Technologies, GIGABYTE, Hewlett Packard Enterprise, Hyve Solutions, Ingrasys, Inventec, Lenovo, Mitac, MSI, Pegatron, QCT, Supermicro, Wistron and Wiwynn.

## OCP Summit Keynotes and Sessions

Ian Buck, vice president of hyperscale and HPC at NVIDIA, will deliver a keynote on Tuesday, Oct. 15, at 9:24 a.m. PT. He’ll address the challenges of designing data centers that can meet future AI workload demands and how NVIDIA is collaborating with the OCP community to overcome them.

Attendees can learn more about NVIDIA’s cloud-scale offerings in more than 20 OCP Global Summit and OCP Future Technologies Symposium sessions, including:

- Addressing the future thermal challenges with AI, an executive session with panelists from Lenovo, Meta, NVIDIA and Vertiv (Tuesday, Oct. 15, at 2:30 p.m. PT)
- The Insatiable Demand AI Places on the Rack and Datacenter, by Google and NVIDIA (Wednesday, Oct. 16, at 8:05 a.m. PT)
- Standardizing Hyperscale Requirements for Accelerators, by AMD, Google, Meta, Microsoft and NVIDIA (Wednesday, Oct. 16, at 9 a.m. PT)
- AI/HPC Future Fabrics Panel (Future Technologies Symposium), by Baya Systems, Lawrence Berkeley National Labs, Lipac Inc., NIST, Marvell and NVIDIA (Wednesday, Oct. 16, at 9:40 a.m. PT)
- Fabric Resiliency at Scale, by NVIDIA (Wednesday, Oct. 16, at 10:25 a.m. PT)
- High-Performance Data Center Storage Using DPUs, by Supermicro and NVIDIA (Wednesday, Oct. 16, at 1:10 p.m. PT)
- CTAM — Compliance Tool for Accelerator Manageability, by Meta, Microsoft and NVIDIA (Wednesday, Oct. 16, at 4:05 p.m. PT)
- Protecting AI Workloads From Noisy Neighbors in Cloud Networks With NVIDIA Spectrum-X, by NVIDIA (Wednesday, Oct. 16, at 4:10 p.m. PT)
- OCP Cooling Environments and ASHRAE TC9.9 Roadmap for the Future Collaboration, by AEI, BP, Meta and NVIDIA (Thursday, Oct. 17, at 9:45 a.m. PT)
- NVIDIA MGX: Faster Time to Market for the Accelerated Data Center, by NVIDIA (Thursday, Oct. 17, at 10:50 a.m. PT)
- Study of Fluid Velocity Limits in Cold Plate Liquid-Cooling Loops, by 1547 Datacenter and NVIDIA (Thursday, Oct. 17, at 3:20 p.m. PT)
- Enabling Chiplet Interoperability and Breakthroughs in AI, by Arm and NVIDIA (Thursday, Oct. 17, at 3:25 p.m. PT)

OCP started with one type of efficient server for Facebook (now Meta) and has grown to encompass accelerators, multiple server types, storage, networking — including adapters, switches and network operating systems — and a host of related data center technologies, including AI.

See what NVIDIA and its partners are contributing to the Open Compute Project by visiting the [NVIDIA OCP Global Summit page](https://www.nvidia.com/en-us/events/ocp-summit/).

![Why Performance per Watt Is the Ultimate Metric for AI Infrastructure Efficiency](./assets/img-83d2f103.jpg)