    |
||
 |
 |
 |
|
||||||
|
|
|
|
这建立在我们向CoreWeave交付首批GB200和GB300系统的基础上，强化了基于速度、执行力以及大规模AI交付的共同关注的合作。 这也反映了人工智能基础设施大规模设计、建设和部署方式的根本转变。 从组件到架构 曾经关于GPU的讨论，现在已经变成了关于全机架系统和人工智能工厂的讨论。约束不再只是计算。它指的是计算、数据、网络、电力和冷却如何协同运作。把这些整合起来很复杂。 戴尔PowerRack配合NVIDIA Vera Rubin为客户简化了操作。 使用PowerRack，你不是买零件然后指望它们能协同工作。你部署的是一个完全集成的系统，计算、网络和存储被设计成一个整体，电源和散热设计从一开始就得到优化。每个系统都作为一个整体设计、测试和交付。这是一个一站式解决方案，旨在加速部署——客户从交付到生产的过程比以往更快（不到6.5小时）。 Efficiency内置于Dell PowerRacks，采用Dell PowerCool技术。这些热管理进步解决了AI驱动的冷却挑战，确保高效、可靠的基础设施，满足空间和电力需求。戴尔经过工厂验证的系统提供优化的气流、直流冷却和集成电力分配，非常适合现代人工智能部署。 拥有硬件和大规模运行AI的区别在于执行力。第一枚令牌的时间，可靠性和整合很重要。这正是戴尔与众不同的地方。
大规模重写人工智能的经济学 随着组织进入生产，讨论的焦点从投入转向产出。重要的单位是代币。你能生成多少，多快，以及付出多少代价。 采用NVIDIA Vera Rubin平台构建的PowerRack在大规模代理AI推理中，每枚代币成本比Blackwell低多达10倍。 这不是渐进式的改进。它改变了人工智能的部署方式。 戴尔与CoreWeave携手，帮助AI创新者部署日益复杂的工作负载，满足生产环境所需的性能、效率和可靠性。对客户来说，这意味着能够训练更大规模的模型，承担更多推理工作负载，并以前所未有的规模支持新兴的代理型人工智能应用。 在代币需求加速增长的世界里，代币成本成为限制因素。降低成本会解锁全新的采用层次。 在人工智能时代，表现赢得头条新闻。每个代币的成本决定了谁能扩展。 为AI构建器提供动力 我们的合作反映了更广泛的转变。前沿人工智能公司、企业、新云以及那些开始主权倡议的公司，正在将人工智能建设为核心基础设施，而非实验。他们需要专门为这种现实设计的系统。这正是戴尔和CoreWeave所提供的。 超过5000名客户正在基于戴尔AI工厂进行开发，从数据到代币再到结果。在这些部署中，需求保持一致：集成系统、可预测的性能、可扩展的基础设施和安全环境。 PowerRack 和 NVIDIA Vera Rubin 将这一模型扩展到了下一代。 情报正在成为基础设施 准备好在为边境提供动力的同一基础设施上建设了吗？了解更多关于戴尔如何通过NVIDIA的戴尔AI工厂实现大规模AI交付的信息。
CoreWeave完成业界首例NVIDIA Vera Rubin的升级与验证 NVL72
代理人工智能正在重塑基础设施需求。随着模型达到万亿参数，上下文窗口扩展到数百万个令牌，持续推理会话成为标准，推理性能已成为限制AI公司运营和增长速度的关键因素。 NVIDIA Vera Rubin NVL72——每机架配备72个NVIDIA Rubin显卡和36个NVIDIA Vera CPU，通过260 TB/s的NVIDIA NVLink第六代结构连接——每瓦推理效率提升多达10×，GPU数量减少多达四分之一，成本为每百万代币的十分之一。有了Vera Rubin，CoreWeave将为客户带来更好的成果。 Jane Street定量研究负责人Craig Falls表示：“我们的研究依赖于既强大又可靠的基础设施，CoreWeave在扩展到NVIDIA Hopper和Blackwell的过程中做到了这一点。”“他们能够交付高效能且具备全集群可观测性的数据集群，以及深入参与棘手问题的支持团队，这让我们有信心与他们合作完成Vera Rubin项目。我们对机架规模的效率提升感到兴奋，这为我们的研究人员带来了更快的训练运行和更短的迭代周期。”
专为机架级人工智能打造的基础设施，由CoreWeave任务控制驱动 为了让客户在生产规模上更好地利用Vera Rubin，CoreWeave开发了一套全新的专用创新产品： 软件定义液冷：Valvey是CoreWeave的可编程每机架阀组件，将冷却从被动机械系统转变为软件定义的机架级控制面。作为CoreWeave任务控制的一部分，Valvey实时监测流量、温度、压力和泄漏检测，实现自动隔离、紧急停机和维护，同时不干扰共享冷却回路上的相邻机架。 “Vera Rubin是NVIDIA迄今为止构建的最强大AI平台，”NVIDIA超大规模与高性能计算（HPC）副总裁Ian Buck表示。“CoreWeave一直站在大规模部署每一代NVIDIA架构的前沿，他们对Vera Rubin的全栈、端到端方法，从冷却到编排，正是世界上最有雄心的AI团队推动下一个AI前沿的方式。” 建立在深厚的技术合作基础上 将像Vera Rubin NVL72这样的机架级平台投入生产需要整个基础设施栈的紧密协作。CoreWeave的技术合作伙伴生态系统是Vera Rubin快速且大规模接触客户的核心。戴尔科技通过其高性能 PowerEdge XE9812 服务器为该平台提供了架构骨干。此次升级还配备了美光7600固态硬盘，通过首批机架级部署的液冷NVMe存储解决方案之一，实现了更高的能效。 戴尔科技董事长兼首席执行官迈克尔·戴尔表示：“戴尔科技与CoreWeave共同致力于提供在人工智能需求前沿的创新。”“PowerEdge XE9812正是为这种密度和精度设计的。与CoreWeave合作推出首款NVIDIA Vera Rubin NVL72机架，直接验证了企业级硬件在配合合适的运营专长下所能实现的效果。”
Agentic AI is reshaping infrastructure requirements. As models reach a trillion parameters, context windows extend to millions of tokens, and persistent reasoning sessions become standard, inference performance has emerged as the defining constraint on how quickly AI companies can operate and grow. NVIDIA Vera Rubin NVL72 — featuring 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs per rack, connected via a 260 TB/s NVIDIA NVLink 6th-generation fabric — delivers up to 10× better inference per watt, up to one-fourth fewer GPUs, and one-tenth the cost per million tokens compared to NVIDIA Blackwell. With Vera Rubin, CoreWeave will deliver better results for customers. “Our research depends on infrastructure that’s both powerful and reliable, and CoreWeave has delivered on this as we’ve scaled across NVIDIA Hopper and Blackwell,” said Craig Falls, head of Quantitative Research at Jane Street. “Their ability to deliver highly performant clusters with full cluster observability and a support team that engages deeply on hard problems gives us the confidence to partner with them on Vera Rubin. We are excited about the efficiency gains at rack scale translating into faster training runs and shorter iteration cycles for our researchers.” Purpose-Built Infrastructure for Rack-Scale AI, Powered by CoreWeave Mission Control To allow customer to take better advantage of Vera Rubin at production scale, CoreWeave developed a new set of purpose-built innovations: Software-Defined Liquid Cooling: Valvey is CoreWeave’s programmable per-rack valve assembly which turns cooling from a passive mechanical system into a software-defined, rack-level control surface. Part of CoreWeave Mission Control, Valvey monitors flow rate, temperature, pressure, and leak-detection in real time, enabling automated isolation, emergency shutdown, and maintenance without disrupting neighboring racks on a shared cooling loop. “Vera Rubin is the most capable AI platform NVIDIA has ever built,” said Ian Buck, vice president of Hyperscale and High-Performance Computing (HPC) at NVIDIA. “CoreWeave has consistently been at the frontier of deploying each new generation of NVIDIA architecture at scale, and their full-stack, end-to-end approach to Vera Rubin, from cooling to orchestration, is how the world’s most ambitious AI teams will push the next AI frontier.” Built on a Foundation of Deep Technical Partnerships Bringing a rack-scale platform like Vera Rubin NVL72 to production requires tight collaboration across the entire infrastructure stack. CoreWeave’s ecosystem of technology partners is central to how Vera Rubin reaches customers at speed and scale. Dell Technologies provided the architectural backbone for the platform through its high-performance PowerEdge XE9812 servers. The bring up also features Micron 7600 SSDs, delivering improved energy efficiency through one of the first liquid-cooled NVMe storage solutions deployed at rack-scale. “Dell Technologies and CoreWeave share a commitment to delivering innovation that performs at the frontier of what AI demands,” said Michael Dell, chairman and chief executive at Dell Technologies. “The PowerEdge XE9812 was engineered for exactly this kind of density and precision. Working with CoreWeave to bring up the first NVIDIA Vera Rubin NVL72 rack is a direct validation of what enterprise-grade hardware can do when it’s paired with the right operational expertise.” About CoreWeave CoreWeave is The Essential Cloud for AI. Built for pioneers by pioneers, CoreWeave delivers a platform of technology, tools, and teams that enables innovators to move at the pace of innovation, building and scaling AI with confidence. Trusted by leading AI labs, startups, and global enterprises, CoreWeave serves as a force multiplier by combining superior infrastructure performance with deep technical expertise to accelerate breakthroughs. Established in 2017, CoreWeave completed its public listing on Nasdaq (CRWV) in March 2025. Learn more at www.coreweave.com. Dell First to Ship Systems Built on NVIDIA Vera Rubin Platform to CoreWeave This builds on our legacy of delivering the first GB200 and GB300 systems to CoreWeave, reinforcing a collaboration built on speed, execution and a shared focus on delivering AI at scale. It also reflects a fundamental shift in how AI infrastructure is designed, built and deployed at scale. From components to architectures What used to be a conversation about GPUs has become a conversation about full rack-scale systems and AI factories. The constraint is no longer just compute. It is how compute, data, networking, power and cooling operate together as one. Bringing all of that together is complex. Dell PowerRack with NVIDIA Vera Rubin simplifies it for our customers. With PowerRack, you are not buying components and hoping they work together. You are deploying a fully integrated system where compute, networking and storage are engineered together to work as one unit, and power and thermal design are optimized from the start. Each system is designed, tested and delivered as a single unit. It is a turnkey solution built to accelerate deployment — so customers move from delivery to production faster than ever (in under 6.5 hours). Efficiency is built into Dell PowerRacks with Dell PowerCool technology. These thermal management advancements tackle AI-driven cooling challenges and ensure efficient, reliable infrastructure tailored to space and power needs. Dell’s factory-validated systems offer optimized airflow, direct-liquid cooling and integrated power distribution ideal for modern AI deployments. The difference between having hardware and running AI at scale is execution. Time to first token, reliability and integration matter. That is where Dell differentiates. Rewriting the economics of AI at scale As organizations move into production, the conversation shifts away from inputs to outputs. The unit that matters is tokens. How many you can generate, how fast, and at what cost. PowerRacks built with NVIDIA Vera Rubin platform deliver up to 10 times lower cost per token than Blackwell for large-scale agentic AI inferencing. That is not an incremental improvement. It changes how AI gets deployed. Together, Dell and CoreWeave are helping AI innovators deploy increasingly complex workloads with the performance, efficiency and reliability required for production environments. For customers, that means the ability to train larger models, serve more inference workloads and support emerging agentic AI applications at unprecedented scale. In a world where token demand is accelerating, cost per token becomes the limiting factor. Reducing that cost unlocks entirely new levels of adoption. In the AI era, performance wins headlines. Cost per token decides who can scale. Powering the AI builders Our collaboration reflects a broader shift. Frontier AI companies, enterprises, neoclouds and those embarking on sovereign initiatives are building AI as core infrastructure, not an experiment. They require systems purpose-built for that reality. That is what Dell and CoreWeave delivers. More than 5,000 customers are building on the Dell AI Factory, moving from data to tokens to outcomes. Across these deployments, the requirement is consistent: integrated systems, predictable performance, scalable infrastructure and secure environments. PowerRack and NVIDIA Vera Rubin extend that model to the next generation. Intelligence is becoming infrastructure Ready to build on the same infrastructure powering the frontier? Learn more about how Dell delivers AI at scale with the Dell AI Factory with NVIDIA.
关于CoreWeave CoreWeave 是人工智能?的核心云。CoreWeave由先驱者为先驱打造，提供一个技术、工具和团队平台，使创新者能够以创新步伐前进，自信地构建和扩展人工智能。CoreWeave 受到领先 AI 实验室、初创企业和全球企业的信赖，结合卓越的基础设施性能与深厚的技术专长，成为加速突破的倍增器。CoreWeave成立于2017年，并于2025年3月完成在纳斯达克（CRWV）上市。更多信息请访问 www.coreweave.com。
|
|
