[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82025-en":3,"doc-seo-82025-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":11,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},82025,7971461740886,"Theodore","https://ap-avatar.wpscdn.com/davatar_3d24733baf745e90a7e4bdd5f77d97b2",8,"Research & Report","Who Needs DRAM? We Have Fiber","Rising pressure on DRAM availability and contract pricing stems from generative AI’s high-performance memory demands and the growth of hyperscale data centers consuming a major share of global DRAM output. Fiber Memory proposes using optical fiber as an active, recirculating delay-line memory for immutable data such as LLM weights. A data-parallel optical broadcast delay-line architecture models fiber physical constraints. Using multi-core fibers, passive tap-and-amplify, CPO, and regional all-optical regeneration, evaluation shows eliminating redundant weight storage across 10,000 AI accelerators and cutting weight-delivery energy by over 70% versus HBM3e.","Who Needs DRAM? We Have Fiber  \nHannah Atmer  \nUppsala University Uppsala, Sweden  \nThiemo Voigt Uppsala University Uppsala, Sweden  \nYuan Yao Uppsala University Uppsala, Sweden  \nStefanos Kaxiras Uppsala University Uppsala, Sweden  \narXiv :2607 .08407v 1 [ cs .AR] 9 Jul 2026  \nAbstract  \nThe rising pressure on DRAM availability and contract pricing reflects generative AI’s massive high-performance memory requirements. This pressure is heavily compounded by hyperscale data center expansion, which now consumes a significant portion of global DRAM output. In this work, we propose a new architecture: Fiber Memory, which reimagines the role of optical fiber in a hyperscale data center, deploying it as an active, recirculating delay-line memory for immutable data, such as large language model (LLM) weights. We present a data-parallel optical broadcast delayline memory architecture that accounts for fiber’s physical realities. By incorporating space-division multiplexed multi-core fibers (MCFs), passive optical tap-and-amplify interfaces, co-packaged optics (CPO), and regional all-optical regeneration, our case study evaluation demonstrates that Fiber Memory can eliminate redundant weight storage across 10,000 AI accelerators and reduce weight-delivery energy by over 70% compared to traditional HBM3e configurations.  \n1 Introduction and Motivation  \nMemory is a bottleneck in modern computing clusters running large language models (LLMs) . Traditional hardware platforms depend on stacking high-bandwidth memory (HBM) or double-data-rate (DDR) DRAM directly adjacent to processing units to feed billions of model parameters into arithmetic pipelines. This paradigm has driven a massive surge in DRAM demand, resulting in supply constraints, high costs, and thermal and power limits within hyperscale data centers [26] . In this work, we propose a new paradigm where the optical fiber network is used as memory to avoid unnecessary data replication.  \nInsight 1: Weight replication in DRAM memory across a datacenter is immensely inefficient. Consider that in a hyperscale datacenter, the same model parameters (e.g., Attention and MLP weights) are replicated across all the nodes that serve the same LLM. Not only model parameters are replicated extensively, but far worse, the accesses (requests-responses) to such replicated data are identically  \nThis work was supported by the Swedish Foundation for Strategic Research (SSF) grant FUS21-0067 .  \nperformed by every node that serves the same model. Replication leads to excessive energy consumption.  \nInsight 2: Fiber is Memory. The sheer amount of fiber ina hyperscale datacenter (10,000 to 100,000 km of fiber strand) holds an immense number of bits at any time. A simple loop of fiber that spans the datacenter has a capacity of multiple TB of data that circulate past every compute node in the loop at 2/3 the speed of light (􀀲) . Effectively, we can turn fiber into a Delay-line Memory, one of the first types of memory used in electronic computers [34], albeit with a tremendous speed, capacity, and length, compared to the Mercury delay-line memories of the 1940’s and early 1950’s.  \nInstead of thousands of nodes consuming local electrical energy to fetch identical weights from HBM or DDR, a single centralized optical transmitter can stream the model parameters once into the shared fiber network. Inference nodes passively “tap” the data circulating continuously in the fiber, extracting weights on the fly and eliminating redundant weight storage and fetch energy across the cluster.  \nOur proposal is supported by the growing adoption of co-packaged optics (CPO) which places silicon photonics engines directly onto the processor substrate to bypass powerhungry electrical transceivers [12]. In co-package optics, the silicon switch chip and the optical engines (silicon photonics chips) are placed on the same package substrate. The electrical signal only has to travel millimeters instead of centimeters, which av","cbCaivHl4CA3HC4I","https://ap.wps.com/l/cbCaivHl4CA3HC4I","pdf",985870,4,1,"English","en",105,"# Introduction and Motivation\n# Fiber as Memory\n## Bit Capacity of Fiber and the Bandwidth-Delay Product","[{\"question\":\"Why is DRAM demand becoming a critical bottleneck for generative AI systems?\",\"answer\":\"DRAM demand rises sharply because LLM parameters must be fed into compute pipelines using stacked HBM or nearby DDR. Hyperscale data center expansion compounds supply constraints, cost increases, and thermal/power limits.\"},{\"question\":\"What is the core idea behind “Fiber Memory”?\",\"answer\":\"Fiber Memory reimagines optical fiber in a hyperscale data center as an active, recirculating delay-line memory. Model weights are streamed once into shared fiber, and inference nodes passively tap the circulating data when needed.\"},{\"question\":\"How does Fiber Memory reduce energy compared with traditional HBM3e configurations?\",\"answer\":\"By avoiding redundant local weight storage and fetch energy across many accelerators, Fiber Memory supports passive weight extraction from a shared optical network. The evaluation in the work reports over 70% reduction in weight-delivery energy versus HBM3e while eliminating redundant weight storage across 10,000 AI accelerators.\"}]","Who Needs DRAM? We Have Fiber | PDF",1784177656,20,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"who-needs-dram-we-have-fiber","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":20},"https://docshare.wps.com/document/who-needs-dram-we-have-fiber/82025/",{"url":52,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-29","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is DRAM demand becoming a critical bottleneck for generative AI systems?","Question",{"text":75,"@type":76},"DRAM demand rises sharply because LLM parameters must be fed into compute pipelines using stacked HBM or nearby DDR. Hyperscale data center expansion compounds supply constraints, cost increases, and thermal/power limits.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the core idea behind “Fiber Memory”?",{"text":80,"@type":76},"Fiber Memory reimagines optical fiber in a hyperscale data center as an active, recirculating delay-line memory. Model weights are streamed once into shared fiber, and inference nodes passively tap the circulating data when needed.",{"name":82,"@type":73,"acceptedAnswer":83},"How does Fiber Memory reduce energy compared with traditional HBM3e configurations?",{"text":84,"@type":76},"By avoiding redundant local weight storage and fetch energy across many accelerators, Fiber Memory supports passive weight extraction from a shared optical network. The evaluation in the work reports over 70% reduction in weight-delivery energy versus HBM3e while eliminating redundant weight storage across 10,000 AI accelerators.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,127,130,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":29,"slug":126},9,"Religion & Spirituality","religion-spirituality",{"id":29,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":29,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]