搜索“NVIDIA”

“作为一个国家，我们承受不起让自己陷入那种境地。” Garman还试图戳破他所说的围绕数据中心的四大迷思：数据中心消耗过多水资源、推高电力成本、排放大量污染，以及对所在社区没有任何好处。 Garman援引Amazon关于自身用水情况的报告称，“数据中心的直接用水量”仅占美国全部工业用水量的0.5%，“比高尔夫球场、杏仁种植和许多其他行业低了好几个数量级。” Nvidia最近表示，其新的冷却系统消除了数据中心内部“几乎所有的用水”，但这些说法——和Amazon的一样——似乎忽略了发电和芯片制造所涉及的更广泛用水。科学家们也表示，他们需要独立研究数据中心的用水和能源使用情况，因为联邦或州层面都没有关于科技公司如何报告这些数据的要求。 至于电价，Garman表示，电价只在数据中心数量众多的一些州上涨，而在其他州则下降或至少增长更慢。他还辩称：“在电价上涨的情况下，主要是因为电网老旧，而且在需求到来之前没有进行投资和扩建。” 另一方面，一家独立监督机构最近表示，数据中心是美国最大电网电价同比上涨76%的主要元凶。 谈到污染问题——NAACP目前正就此起诉Elon Musk的SpaceX/xAI

匹配字段正文

periment: Qwen-3.6-35B-A3B inference Setup In kicking off the experiment, I gave Claude Code with Fable 5 access to the paper and the MetaInfer Repository, as well as access to an NVIDIA B200 workstation via SSH. I told it the target hardware and provided a link to the Hugging Face repo for both the NVFP4 quantized weights I wanted to use for Qwen-3.6-35B-A3B, as well as the original full-precision weights to be used as an accuracy oracle. The agent was also able to deploy engine candidates to B

匹配字段正文

ex (@dair ai).Context compression: Microsoft’s training-free FOCUS cuts peak context by up to 48% and raises task success by up to 8.9 points (@dair ai); Long-context degradation: NVIDIA’s Long-Transduction study measures a 62.8% accuracy drop from 4K to 128K context across seven open models (@dair ai); Multi-agent coordination: In AgentWorld, fewer than a third of multi-agent actions help complete the task, and coordination tasks reach only 12% success (@omarsar0); Apple LoopCD: The method halv

匹配字段正文

正在变得更加多元化。为了让大家更容易理解我们的业务，我们现在用两个市场平台来描述英伟达：数据中心和边缘计算。 边缘计算包括PC、工作站、游戏、AI基站、机器人和汽车。我们能够服务所有这些市场，是因为我们拥有一个统一架构、一个软件栈，以及一个丰富的生态系统。 英伟达开创了这一领域，并构建了完整闭环。AI工厂训练模型；Omniverse在虚拟世界中对其进行仿真；NVIDIA Jetson计算机在机器人中运行；Cosmos则是驱动这一切的世界基础模型。 机器人和机器人系统正在各个行业被打造出来，从交通、制造、手术机器人，到酒店和服务业。 过去几个月，AI出现了巨大的加速。最后我想强调几点：有用的AI已经到来，而且它是盈利的。因此，算力就是收入。Vera Rubin已经全面投产。Vera正在打开一个全新的、为智能体打造的CPU市场。每一家企业都正在成为智能体公司，而它们都运行在英伟达之上。 随着我们的成长，我们将继续扩大研发投入，投资我们的生态系统，并向股东返还资本。我们最近宣布大幅提高股息，并扩大股票回购计划。我们计划在今年、明年以及未来，把50%或更多自由现金流返还给股东。 谢谢。 部分Q

匹配字段正文

用SEC凭证、访问账户或非公开信息、SEC数据或系统变更，或有入侵或漏洞的证据。 美联社称，这是OpenAI三个月内第二次暂停模型开发。公司表示，将在确认额外防护措施到位后恢复训练。 新闻来源：OpenAI官网、央视新闻、美联社、卫报、加拿大广播公司 图片来源：卫报 英伟达推出开放智能体安全平台 据美国消费者新闻与商业频道（CNBC）9月28日报道，英伟达（NVIDIA）推出开放智能体安全平台（Open Agent Safety Platform），通过软件与硬件控制人工智能智能体的访问权限，降低其绕过安全限制的风险。 英伟达在发布的公告中表示，平台包括开源运行时软件OpenShell和Sentry参考系统设计。前者记录智能体操作并执行安全策略，可扩展至第三方计算平台；后者运行于BlueField-4数据处理单元（Data Processing Unit，DPU），独立监测智能体行为。公司称，Sentry可在毫秒内隔离并停止试图越界的智能体。目前，已有逾100家机构参与相关技术合作。 网络安全媒体CSO Online报道援引分析人士观点称，将控制措施置于智能体之外，有助于加强安全管理，

匹配字段正文

periment: Qwen-3.6-35B-A3B inference Setup In kicking off the experiment, I gave Claude Code with Fable 5 access to the paper and the MetaInfer Repository, as well as access to an NVIDIA B200 workstation via SSH. I told it the target hardware and provided a link to the Hugging Face repo for both the NVFP4 quantized weights I wanted to use for Qwen-3.6-35B-A3B, as well as the original full-precision weights to be used as an accuracy oracle. The agent was also able to deploy engine candidates to B

匹配字段正文

cal (like iCloud) would flow into the “US Information/ICT” indices of GDP calculation, but its sales of iPhones, Macs and iPads would flow into manufacturing, the same place where NVIDIA’s GPU sales would go. ICT also doesn’t include consultancy revenue or IT services from companies like Accenture, but that isn’t really relevant to the analysis. In any case, the ICT industry’s contribution to GDP is actually very, very useful for this calculation, because it specifically includes sales of AI sof

匹配字段正文

le-dominated industry, it is both striking and exciting that AMD now has two women in its most senior technical roles as the firm continues to rebuild its claim to being a serious NVIDIA competitor. The World Labs acquisition is the news of the week, but the background story which bubbles along is that initial shipments of AMD’s first rack-scale AI solution, Helios, will begin by the year’s end. Investors are excited to see these systems out in the world; HPE stock rose sharply this week after i

匹配字段正文

g range and acceleration on higher-end trims. U.S. authorities arrested Greg Lui, owner of California-based Earthmade Computer, for smuggling over $300M worth of export-controlled Nvidia AI servers to China. Reddit announced it is ending support for RSS feeds in November and shutting down public API access by March 2027 to curb large-scale scraping by AI bots. OpenAI added virtual try-on to ChatGPT, letting users select an apparel listing, upload a photo of themselves, and see how it looks on th

匹配字段正文

fornia has subpoenaed OpenAI over its rogue AI agents. ( Guardian ) + Who's liable when AI agents go rogue? ( MIT Technology Review ) 2 A US man allegedly smuggled $300 million of Nvidia chips to China He’s been arrested in California on federal charges. ( Reuters $) + The chip smuggling cases expose Nvidia’s blind spots. ( Bloomberg $) + China plans to teach AI in every school by 2030. ( WP $) + But it already has an overuse problem. ( NYT $) 3 Florida officials have found 11 Flock cameras with

匹配字段正文

xperienced talent they will ultimately need. AI may not eliminate the corporate career ladder, but it could take away many of the rungs people have traditionally used to climb it. Nvidia launches AI safety platform after agent security breaches: The AI chipmaker introduced an open source safety platform designed to add runtime guardrails as enterprises give AI agents more autonomy. AI industry copes with out-of-control agents. Here’s OpenAI’s response: The generative AI vendor delayed its Astra

匹配字段正文

nd the revenues of Micron Technology, Samsung, and SK Hynix are skyrocketing much faster than the IBM mainframe ever did. This is a thing to behold, and it underpins the growth of Nvidia and AMD. Not to put too fine a point on it, but without the high bandwidth stacked DRAM these companies make, the GenAI revolution would not – and could not – have happened. I have been covering Micron’s financials since this boom started, and I will be adding Samsung and SK Hynix in the next round of numbers to

匹配字段正文

he first time the tech giant has sent one of its advanced chips into space. Built by Planet Labs , the satellite will prove that a Google Tensor Processing Unit, its competitor to Nvidia’s GPUs, can function in space. That means supplying a kilowatt of continuous power, cooling the chip, and running a series of models through their paces to see if anything goes wrong. “We’ve done testing on the ground, but you know, there’s no test that’s completely as good as the real thing,” said Travis Beals,

匹配字段正文

院选举中不断上升，两大政党的候选人也越来越多地采取反对数据中心建设的立场。例如，密歇根州共和党候选人Mike Rogers呼吁暂停数据中心建设一年；特朗普总统在得克萨斯州的提名人Ken Paxton则誓言取消此类项目的补贴。 AIIC的联合主席Sinema将部分反对情绪归因于“一个极其复杂且有针对性的虚假信息机器”，但她补充道，许多美国人所表达的“担忧或恐惧是完全自然的”。 她表示，“真正重要的是要认识到，美国民众提出的个人关切是合理的，需要得到回应，并且应以积极主动、基于社区的方式加以解决。”她指出，AIIC委托进行的一项民意调查表明，当选民看到新的“原则”时，他们更有可能支持数据中心建设。 与此同时，特朗普本人则无视了对限制数据中心建设的呼声，称反对者希望将美国科技产业拖入“落后与贫困”的境地。他还否认了有关人工智能安全问题的说法。 当地时间，周二（9月30日），在与Anthropic、OpenAI、Nvidia、Palantir等科技公司的高管共进午餐后，特朗普表示，这些高管已承诺“与当地社区合作”，并补充说，“这可能意味着提供石油和天然气、提供教育，以及帮助人们改善经济状况”。

匹配字段正文

ts Season 3, featuring 31 days of hands-on tutorials on building and securing AI agents in production with Gemini and ADK — no sign-in or paywall. OpenAI, Google, Anthropic, Meta, Nvidia, and xAI signed a voluntary White House “Superintelligence Accord” to police their top models with outside auditors. The Pentagon’s Pete Hegseth introduced Project Meridian, a 120-day hunt for futuristic weapons troops could need “from under the Earth to beyond the moon,” with Elon Musk, Newt Gingrich, and Palme

匹配字段正文

bug fix : OpenAI fixed an image-encoding bug, adding 1 Intelligence Index point to GPT-6 Luna. Ultrafast inference : OpenAI quotes up to 300 tok/s. SemiAnalysis reports it runs on NVIDIA GPUs at low batch sizes, not on Cerebras ( @kimmonismus ). Hands-on report : Generation is about 8x faster, but end-to-end agent tasks speed up only 2–4x because tool latency dominates ( @sayashk ). Computer use : Gains are largest here, since UI actions respond in milliseconds. Cost : The tester exhausted a wee

匹配字段正文

一代的架构，专为需要低延迟的 AI 工作负载而优化。GPU 需要在芯片与内存之间来回搬运数据，而 WSE 则把所有数据都保留在片上，从而消除了制约 GPU 推理的内存带宽瓶颈。 以下是你如何判断延迟债确实存在的方法：跟着钱走。 在过去六个月里，AI 领域最重要的四家公司都押下了数十亿美元的重注，赌的是更快的推理，而这些押注没有一笔落在目前已经主导 AI 的 NVIDIA 硬件上。Google 尽管是 NVIDIA 最大的客户之一，却自研了 Ironwood TPU，在推理工作负载上比 NVIDIA 的 GPU 快 4 倍。Anthropic 承诺向 Google 的 TPU 基础设施投入数百亿美元，而不是扩充其 NVIDIA 算力。NVIDIA 自己则斥资 200 亿美元，掏空了 Groq 的知识产权和顶尖人才，而这家公司的全部价值主张就是推理速度。这是 NVIDIA 迄今为止最大的一笔收购。而 OpenAI 刚刚从 Cerebras 购买了 750 兆瓦的算力。Cerebras 制造的晶圆级芯片，正是专为 AI 推理所需的那种顺序式、受内存带宽限制的工作负载而设计的。 四家在一切事务

匹配字段正文

Mar 26 2026 The entire way we run AI inference is being rearchitected right now. AWS and Cerebras just announced a partnership around it. NVIDIA spent $20 billion acquiring Groq to catch up. Jensen Huang stood on stage at GTC 2026 and effectively validated what companies like Cerebras have been saying for years: general-purpose GPUs aren't enough for inference at scale. The thing they're all converging on is called disaggregated inference . And if you're a developer building anything on top of L

匹配字段正文

工程 如何在 DeepSeek Harness 中运行任意开源模型 Alex Ker AI 工程 微调 Qwen3-TTS 以实现高质量语音克隆 Ian Carrasco AI 工程 如何在任意 harness 中运行 Kimi K3：使用 Baseten Switch 进行路由 Zak Keener 另有 1 人 AI 工程 在 Baseten 上使用 NVIDIA Nemotron 3 Ultra 与 LangChain Deep Agents Code 进行构建 Philip Kiely AI 工程 如何在任意 harness 中运行 GLM-5.2 Alex Ker AI 工程 实现零停机模型更新的滚动部署 Archit Mishra 另有 2 人 AI 工程 Harness 就是一切。以下是如何优化你的 harness。Alex Ker 另有 1 人 AI 工程 保护你的 harness：如何将 NVIDIA 的 NemoClaw 与前沿开源模型配合运行 Alex Ker 1 2 3 ... 7 requestAnimationFrame(function(){$RT=pe

匹配字段正文

r World Power大会上，来自公用事业公司、电力生产商、超大规模厂商、数据中心开发商、运营商和供应链合作伙伴的领导者们讨论了如何在扩大容量的同时保障电网可靠性、保护电费缴纳者并加速经济发展。 Data Center World Power顾问委员会执行主席、Apolo.us首席执行官兼联合创始人Bill Kleyman在早间一场会议上开场，阐述了部署Nvidia DGX级服务器的现实情况。他指出，一个标准机架可容纳六个此类系统，每个系统围绕八块Nvidia H100加速器构建，单块加速器卡的成本最高可达40,000美元。 相关：机架功率正快速攀升。这对数据中心意味着什么 “AFCOM《2026年数据中心状况》调查中69%的受访者预计，机架密度将以我们过去从未见过的速度增长，”Kleyman说。“与此同时，我们看到数据中心液冷采用率增长了36%，而且这一趋势还在持续扩大。” NRG Business执行副总裁Scott Hart描述了客户需求如何升级。NRG Business是北美主要的零售电力、天然气和需求响应提供商。“传统上，数据中心会是一个并网的20至40 MW设施，”他说

匹配字段正文