--- 格式版本: 2 标题: "Cerebras unveils four trillion-transistor giant chip, targets generative AI" 原文链接: "https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/" 发布日期: "2026-08-12" 发布时间校准状态: "found" 发布时间需复核: "否" 发布时间来源: "llm:local:strict_original_body" 发布时间证据: "Published Time: Wed, 12 Aug 2026 01:44:03 GMT" 发布时间校准原因: "正文中明确标注的发布时间，符合优先级最高的候选条件。" 发布时间校准置信度: "1" 发布时间候选数量: 38 发布时间严格候选数量: 10 发布时间原页读取状态: "source template page reused from URL open" 发布时间未找到原因: "" 发布时间校准时间: "2026-08-12T11:51:45+08:00" 发布时间仲裁状态: "confirmed" 发布时间仲裁尝试次数: 1 发布时间仲裁耗时毫秒: 5315 发现时间: "2026-08-12T11:49:22+08:00" 入库时间: "2026-08-12T03:51:50.422Z" 来源平台: "SDxCentral 搜索" 搜索渠道: "source_template" 搜索词: "site:sdxcentral.com SRAM(Nvidia)" 匹配关键词: - "AI" - "GPU" - "SRAM" - "delivery" - "performance" - "bandwidth" 相关厂家: - "NVIDIA" - "AMD" - "Google" 相关专家: [] 内容类型: "网页" 抓取工具: "Jina Reader" 清洗工具: "Jina Reader Markdown + Defuddle/Readability 正文提取" 原始附件: [] AI优质: "否" AI打分: 53 AI分档: "非优质" AI质检状态: "不通过" AI打分理由: "本文介绍Cerebras晶圆级芯片及超算，非超节点/AI Rack/机柜级系统，主题相关性低；技术细节限于芯片参数，未涉及项目关注的高速互连、供电、散热等；商业信号为超算建设与高通合作，但非项目重点。" AI质检模型: "ali-deepseek-v4-flash" AI质检时间: "2026-08-12T11:51:58+08:00" AI主题相关性: 5 AI来源权威性: 10 AI新颖性: 12 AI技术细节: 10 AI商业部署信号: 8 AI完整性: 8 AI摘要: "Cerebras发布新一代晶圆级芯片WSE-3，集成四万亿晶体管，单CS-3系统宣称可提供125峰值AI petaflops。" AI摘要模型: "ali-deepseek-v4-flash" AI摘要时间: "2026-09-07T03:32:10.345Z" 采集批次: "2026年8月12日2点04分20秒" 采集批次ID: "20260812-020420-289" 去重键: "https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai" --- Title: Cerebras unveils four trillion-transistor giant chip, targets generative AI URL Source: https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/ Published Time: Wed, 12 Aug 2026 01:44:03 GMT Markdown Content: This website uses cookies to ensure you get the best experience on our website. [Learn more](https://www.cookiesandyou.com/) [Got it!](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/) # [![Image 1: SDxCentral](https://www.sdxcentral.com/static/img/sdx-full.2fa50292c52c.png)](https://www.sdxcentral.com/) * Channels * [The AI Channel](https://www.sdxcentral.com/ai/) * [The Networking Channel](https://www.sdxcentral.com/network-sdn/) * [The Cybersecurity Channel](https://www.sdxcentral.com/security/) * [The Distributed Infrastructure Channel](https://www.sdxcentral.com/distributed-infrastructure/) * [The Telecom Channel](https://www.sdxcentral.com/telecom-nfv/) * [The Silicon & Supercomputing Channel](https://www.sdxcentral.com/storage-silicon/) * [The Cloud & Storage Channel](https://www.sdxcentral.com/cloud/) * Media * [News](https://www.sdxcentral.com/news/) * [Analysis](https://www.sdxcentral.com/analysis/) * [Editors' Views](https://www.sdxcentral.com/control-plane/) * [Industry Views](https://www.sdxcentral.com/opinions/) * [Video Interviews](https://www.sdxcentral.com/video-interviews/) * [Broadcasts](https://www.sdxcentral.com/broadcasts/) * [Explainers](https://www.sdxcentral.com/sdx-explainers/) * [MarketWatch](https://www.sdxcentral.com/marketwatch/) * [Press Releases](https://www.sdxcentral.com/press-releases/) * [Resources](https://www.sdxcentral.com/resources/) * Live Events * Xcelerated Compute Series * [The Xcelerated Compute Show | LDN 2026](https://www.xceleratedcompute.com/london/2026/) * [The Xcelerated Compute Show | MEX 2026](https://www.xceleratedcompute.com/cancun/2026/) * [The Xcelerated Compute Show | NYC 2027](https://www.xceleratedcompute.com/new-york/2027/) * Digital Infrastructure Series * [Yotta 2026](https://www.yotta-event.com/) * [](https://www.sdxcentral.com/sdx-en/account/login/?next=%2Fnews%2Fcerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai%2F) * [](https://www.sdxcentral.com/sdx-en/search/) Search * Show Menu * Channels * Back * [The AI Channel](https://www.sdxcentral.com/ai/) * [The Networking Channel](https://www.sdxcentral.com/network-sdn/) * [The Cybersecurity Channel](https://www.sdxcentral.com/security/) * [The Distributed Infrastructure Channel](https://www.sdxcentral.com/distributed-infrastructure/) * [The Telecom Channel](https://www.sdxcentral.com/telecom-nfv/) * [The Silicon & Supercomputing Channel](https://www.sdxcentral.com/storage-silicon/) * [The Cloud & Storage Channel](https://www.sdxcentral.com/cloud/) * Media * Back * [News](https://www.sdxcentral.com/news/) * [Analysis](https://www.sdxcentral.com/analysis/) * [Editors' Views](https://www.sdxcentral.com/control-plane/) * [Industry Views](https://www.sdxcentral.com/opinions/) * [Video Interviews](https://www.sdxcentral.com/video-interviews/) * [Broadcasts](https://www.sdxcentral.com/broadcasts/) * [Explainers](https://www.sdxcentral.com/sdx-explainers/) * [MarketWatch](https://www.sdxcentral.com/marketwatch/) * [Press Releases](https://www.sdxcentral.com/press-releases/) * [Resources](https://www.sdxcentral.com/resources/) * Live Events * Back * Xcelerated Compute Series * [The Xcelerated Compute Show | LDN 2026](https://www.xceleratedcompute.com/london/2026/) * [The Xcelerated Compute Show | MEX 2026](https://www.xceleratedcompute.com/cancun/2026/) * [The Xcelerated Compute Show | NYC 2027](https://www.xceleratedcompute.com/new-york/2027/) * Digital Infrastructure Series * [Yotta 2026](https://www.yotta-event.com/) Close modal ## Search Suggestions [Get the Report: How leading storage architects are solving the AI memory crunch](https://www.sdxcentral.com/resources/the-storage-supplement/) * [Home](https://www.sdxcentral.com/) * [News](https://www.sdxcentral.com/news/) * [The Silicon & Supercomputing Channel](https://www.sdxcentral.com/storage-silicon/) # Cerebras unveils four trillion-transistor giant chip, targets generative AI And is building its third supercomputer, alongside plans for Qualcomm inference chip deployments March 13, 2024 By[Sebastian Moss](https://www.sdxcentral.com/profile/sebastian-moss/)[Have your say](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/#comments) [](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.addtoany.com/share)[![Image 2: Preferred source](https://www.sdxcentral.com/static/img/google_preferred_source.172e670ec003.png)](https://www.google.com/preferences/source?q=sdxcentral.com) Cerebras Systems has introduced the latest version of its wafer-sized semiconductor. The Wafer Scale Engine 3 boasts four trillion transistors and 900,000 'AI cores,' alongside 44GB of on-chip SRAM. When combined into the Cerebras CS-3 system, the company claims the chip is capable of 125 peak AI petaflops. That can be scaled up to 2,048 CS-3s in a single cluster, for a purported 256 exaflops of AI compute. [](https://media.datacenterdynamics.com/media/images/Cerebras_WSE-3.original.jpg)![Image 3: Cerebras WSE-3](https://media.datacenterdynamics.com/media/images/Cerebras_WSE-3.width-358.jpg) Cerebras WSE-3– Cerebras "We didn't build it big because big was easy or big was sexy, but not only are we 52 times bigger [than an Nvidia GPU], but we bring 52 times the number of cores, more than 800x on-chip memory, and more than three orders of magnitude more memory bandwidth and communication bandwidth," Cerebras' VP of product Dr. Andy Hock told _DCD_. Compared to the WSE-2, "it's the same biggest square we can cut out of the biggest circular primary silicon we can buy," Hock said. "It's going to be 2× the performance for AI work - within the same power envelope and price." The company has never officially disclosed the price of its chips, but they are believed to cost around $2-3 million. Hock said that a single CS-3 uses 23kW. Alongside the new chip, the company said that it had completed the construction of Condor Galaxy 2, its second cloud supercomputer built as part of [a $900 million deal with G42](https://www.datacenterdynamics.com/en/news/cerebras-signs-900m-deal-with-uaes-g42-to-build-up-to-nine-ai-supercomputers/). "We're now building the third Condor Galaxy machine," Hock said, with this one featuring CS-3s. Cerebras has also partnered with Qualcomm to deploy the latter company's Cloud AI 100 Ultra chips for inference work next to the training supercomputers. Hock said the company was specifically training models to perform well on the inference chips, which would "eliminate a large fraction of unnecessary computation of multiply by zero operations." Stay ahead of the latest compute, storage & networking channel trends. Subscribe to SDxCentral. [Create an Account to Subscribe Now](https://www.sdxcentral.com/create-your-free-sdx-central-account-today/) [](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/)[](https://www.addtoany.com/share)[![Image 4: Preferred source](https://www.sdxcentral.com/static/img/google_preferred_source.172e670ec003.png)](https://www.google.com/preferences/source?q=sdxcentral.com) ### Tags * [Wafer Scale Engine 3](https://www.sdxcentral.com/tags/wafer-scale-engine-3/) * [CS-3](https://www.sdxcentral.com/tags/cs-3/) * [Qualcomm](https://www.sdxcentral.com/tags/qualcomm/) * [Cerebras](https://www.sdxcentral.com/tags/cerebras/) * [Andy Hock](https://www.sdxcentral.com/tags/andy-hock/) * [Condor Galaxy](https://www.sdxcentral.com/tags/condor-galaxy/) * [Qualcomm Cloud AI 100 Ultra](https://www.sdxcentral.com/tags/qualcomm-cloud-ai-100-ultra/) ## Comments [](https://www.sdxcentral.com/news/cerebras-unveils-four-trillion-transistor-giant-chip-targets-generative-ai/) ## Resources [More](https://www.sdxcentral.com/resources/) * ![Image 5: Why Verification Matters in Modern Chip Design](https://media.datacenterdynamics.com/media/images/Screenshot_2026-08-10_135256.height-150.png) # [Why verification matters in modern chip design](https://www.sdxcentral.com/resources/why-verification-matters-in-modern-chip-design/) 10 Aug 2026 * # [Architecting scalable inference on AMD AI Platforms](https://www.sdxcentral.com/resources/architecting-scalable-inference-on-amd-ai-platforms/) 07 Aug 2026 * # [Beyond the rack: Reinventing high-efficiency infrastructure to power the next generation of hosting and online gaming](https://www.sdxcentral.com/resources/reinventing-high-efficiency-infrastructure-to-power-the-next-generation-of-hosting-and-online-gaming/) 29 Jul 2026 * # [The agentic AI journey to self-driving networks](https://www.sdxcentral.com/resources/the-agentic-ai-journey-to-self-driving-networks/) 28 Jul 2026 * [More](https://www.sdxcentral.com/resources/) ![Image 6: SDx Talks Thumbnails (1280 x 720 mm) (5)](https://media.datacenterdynamics.com/media/images/SDx_Talks_Thumbnails_1280_x_720_mm_Y16FcDn.2e16d0ba.fill-280x185.png) 17:39 # [SDxInterviews: AI in football - Beyond the 90 minutes with Andy Rhodes, Lenovo](https://www.sdxcentral.com/video-interviews/sdxinterviews-ai-in-football-beyond-the-90-minutes-with-andy-rhodes-lenovo/) Thank you! You will receive a confirmation email shortly. ## SDx Explainers [More](https://www.sdxcentral.com/sdx-explainers/) * ![Image 7: AI agent concept](https://media.datacenterdynamics.com/media/images/GettyImages-2215349358.2ef912c2.fill-280x185.jpg)18 Nov 2025 # [What is agentic AI and what impact will it have on networks?](https://www.sdxcentral.com/sdx-explainers/what-is-agentic-ai-and-what-impact-will-it-have-on-networks/) * # [What is network functions virtualization (NFV)? The definitive guide to NFV](https://www.sdxcentral.com/sdx-explainers/what-is-network-functions-virtualization-nfv-the-definitive-guide-to-nfv/) 30 Oct 2025 * # [What is the Model Context Protocol (MCP)? The definitive guide to AI’s new integration standard](https://www.sdxcentral.com/sdx-explainers/what-is-the-model-context-protocol-mcp-the-definitive-guide-to-ais-new-integration-standard/) 28 Oct 2025 * # [NFV and SDN: What's the difference?](https://www.sdxcentral.com/sdx-explainers/nfv-and-sdn-whats-the-difference/) 30 Mar 2020 * [More](https://www.sdxcentral.com/sdx-explainers/) Browse by Topic * [Security Operations](https://www.sdxcentral.com/topics/security-operations/) * [Network Threat Detection & Response](https://www.sdxcentral.com/topics/security-operations/network-threat-detection-response/) * [Cloud Security & Posture Management](https://www.sdxcentral.com/topics/security-operations/cloud-security-posture-management/) * [Security Automation](https://www.sdxcentral.com/topics/security-operations/security-automation/) * [Storage Security & Encryption](https://www.sdxcentral.com/topics/security-operations/storage-security-encryption/) * [CVE & Vulnerability Tracker](https://www.sdxcentral.com/topics/security-operations/cve-vulnerability-tracker/) * [AI & Machine Learning](https://www.sdxcentral.com/topics/ai-machine-learning/) * [Agentic AI](https://www.sdxcentral.com/topics/ai-machine-learning/agentic-ai/) * [AI & HPC](https://www.sdxcentral.com/topics/ai-machine-learning/ai-hpc/) * [Analytics & Monitoring](https://www.sdxcentral.com/topics/ai-machine-learning/analytics-monitoring/) * [AI Hardware](https://www.sdxcentral.com/topics/ai-machine-learning/ai-hardware/) * [Cloud & Hyperscale](https://www.sdxcentral.com/topics/cloud-hyperscale/) * [Multi-Cloud Network Connectivity](https://www.sdxcentral.com/topics/cloud-hyperscale/multi-cloud-network-connectivity/) * [Multi-Cloud Management](https://www.sdxcentral.com/topics/cloud-hyperscale/multi-cloud-management/) * [Hybrid Cloud Connectivity](https://www.sdxcentral.com/topics/cloud-hyperscale/hybrid-cloud-connectivity/) * [Distributed Cloud Models](https://www.sdxcentral.com/topics/cloud-hyperscale/distributed-cloud-models/) * [Cloud-Native Networking](https://www.sdxcentral.com/topics/cloud-hyperscale/cloud-native-networking/) * [Kubernetes](https://www.sdxcentral.com/topics/cloud-hyperscale/kubernetes/) * [Storage & Memory](https://www.sdxcentral.com/topics/cloud-hyperscale/storage-memory/) * [Connectivity](https://www.sdxcentral.com/topics/connectivity/) * [Network Programmability & APIs](https://www.sdxcentral.com/topics/connectivity/network-programmability-apis/) * [SD-WAN & SASE](https://www.sdxcentral.com/topics/connectivity/sd-wan-sase/) * [Automation & Virtualization](https://www.sdxcentral.com/topics/connectivity/automation-virtualization/) * [Open Networking Standards](https://www.sdxcentral.com/topics/connectivity/open-networking-standards/) * [Edge & IoT](https://www.sdxcentral.com/topics/edge-iot/) * [Content Delivery Networks](https://www.sdxcentral.com/topics/edge-iot/content-delivery-networks/) * [Distributed Data Centers](https://www.sdxcentral.com/topics/edge-iot/distributed-data-centers/) * [Edge Architectures](https://www.sdxcentral.com/topics/edge-iot/edge-architectures/) * [IoT Network Management](https://www.sdxcentral.com/topics/edge-iot/iot-network-management/) * [Space & Satellites](https://www.sdxcentral.com/topics/edge-iot/space-satellites/) * [IT Hardware & Semiconductors](https://www.sdxcentral.com/topics/it-hardware-semiconductors/) * [SmartNICs, DPUs & ASICs](https://www.sdxcentral.com/topics/it-hardware-semiconductors/smartnics-dpus-asics/) * [NVMe & Storage Protocols](https://www.sdxcentral.com/topics/it-hardware-semiconductors/nvme-storage-protocols/) * [Data Center Interconnects](https://www.sdxcentral.com/topics/it-hardware-semiconductors/data-center-interconnects/) * [Quantum](https://www.sdxcentral.com/topics/quantum/) * [Quantum Networks](https://www.sdxcentral.com/topics/quantum/quantum-networks/) * [Quantum Security & Post-Quantum Cryptography](https://www.sdxcentral.com/topics/quantum/quantum-security-post-quantum-cryptography/) * [Telecoms 5G/6G](https://www.sdxcentral.com/topics/telecoms-5g6g/) * [5G Core Network Virtualization](https://www.sdxcentral.com/topics/telecoms-5g6g/5g-core-network-virtualization/) * [Open RAN & Disaggregated Hardware](https://www.sdxcentral.com/topics/telecoms-5g6g/open-ran-disaggregated-hardware/) * [Telecom Infrastructure Automation](https://www.sdxcentral.com/topics/telecoms-5g6g/telecom-infrastructure-automation/) * [Carrier Ethernet & SDN](https://www.sdxcentral.com/topics/telecoms-5g6g/carrier-ethernet-sdn/) * [6G](https://www.sdxcentral.com/topics/telecoms-5g6g/6g/) * [Regulations & Spectrum](https://www.sdxcentral.com/topics/telecoms-5g6g/regulations-spectrum/) * [Fiber, Backhaul & Transport](https://www.sdxcentral.com/topics/telecoms-5g6g/fiber-backhaul-transport/) * [People & Appointments](https://www.sdxcentral.com/topics/people-appointments/) * [Channel & Partners](https://www.sdxcentral.com/topics/channel-partners/) * [Mergers, Acquisitions & Funding](https://www.sdxcentral.com/topics/mergers-acquisitions-funding/) Create Your Free Account * [Create your free SDxCentral Account today](https://www.sdxcentral.com/create-your-free-sdx-central-account-today/) Our Team * [Our Team](https://www.sdxcentral.com/our-sdx-team/) Platform * [News](https://www.sdxcentral.com/news/) * [Analysis](https://www.sdxcentral.com/analysis/) * [Industry Views](https://www.sdxcentral.com/opinions/) * [Broadcasts](https://www.sdxcentral.com/broadcasts/) * [MarketWatch](https://www.sdxcentral.com/marketwatch/) * [Resources](https://www.sdxcentral.com/resources/) * © 2026 SDxCentral * [Data Protection / Privacy Policy](https://www.sdxcentral.com/data-protection-privacy-policy-infrax/) * [Terms & Conditions](https://www.sdxcentral.com/terms-conditions/) * [Work With Us](https://www.sdxcentral.com/work-with-us/) * [](https://www.linkedin.com/company/sdncentral/?viewAsMember=true "Linkedin") * [](https://x.com/sdxcentral "Twitter") SDxCentral, 32-38 Saffron Hill, London, EC1N 8FH Email. [info@datacenterdynamics.com](mailto:info@datacenterdynamics.com) SDxCentral is a subsidiary of [InfraXmedia](https://www.infraxmedia.com/) **More from InfraXmedia**: [DatacenterDynamics](https://www.datacenterdynamics.com/en/) - [Data Center Nation](https://datacenternation.com/) - [The AI Cities Summit](https://www.aicitysummit.com/)