# AWS data center network architecture

AWS data center network architecture is a flat, quasi-random network design using resilient network graphs, passive optical devices, and specialized routing for large-scale data centers.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## A Wavelength Borrowing Architecture for Optical Data Center Networks - Extended Version

DevFeed: [A Wavelength Borrowing Architecture for Optical Data Center Networks - Extended Version](<https://devfeed.tech/articles/a-wavelength-borrowing-architecture-for-optical-data-center-networks-extended-version-123724.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.04874>)

Author: Andrea Detti, Chiara Lodovisi, Silvello Betti

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [communications](<https://devfeed.tech/topics/communications.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [mesh networking](<https://devfeed.tech/topics/mesh-networking.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [borrowing](<https://devfeed.tech/tags/borrowing.md>), [capability](<https://devfeed.tech/tags/capability.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [components](<https://devfeed.tech/tags/components.md>), [configuration](<https://devfeed.tech/tags/configuration.md>), [control](<https://devfeed.tech/tags/control.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [data-center-networks](<https://devfeed.tech/tags/data-center-networks.md>), [demand](<https://devfeed.tech/tags/demand.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [fabric](<https://devfeed.tech/tags/fabric.md>), [feature](<https://devfeed.tech/tags/feature.md>), [full](<https://devfeed.tech/tags/full.md>), [groups](<https://devfeed.tech/tags/groups.md>), [integration](<https://devfeed.tech/tags/integration.md>), [optical](<https://devfeed.tech/tags/optical.md>), [version](<https://devfeed.tech/tags/version.md>)

### AI overview

The proposed spine-leaf optical network reallocates unused or lightly loaded wavelengths from one leaf to another with greater demand. It combines this wavelength borrowing with two-hop detouring to manage wavelength load while limiting detoured traffic. A single borrowing-degree parameter, B, ranges from no reconfiguration to full reconfiguration; the abstract reports performance nearly equal to a fully reconfigurable design at values well below the maximum, reducing the need for full reconfiguration complexity. The fabric uses AWGs, AWGRs, colorless OXCs, and combiners. Its reconfiguration speed and dynamic control policy suit tiers where traffic changes over tens of minutes or longer, such as between rack groups (pods).

### Source excerpt

arXiv:2609.04874v2 Announce Type: replace Abstract: The growth of east-west traffic, along with the cost and power consumption of electronic switching, is motivating the integration of a low-power, high-rate, all-optical layer within the data center network. This paper presents a spine-leaf all-optical architecture in which the default wavelength configuration, one wavelength per source--destination leaf pair, can be reconfigured to accommodate unbalanced traffic demand: wavelengths that are unused or lightly loaded at one leaf are borrowed by another leaf with higher demand. This topology-engineering capability is combined with a traffic-engineering scheme, based on two-hop detouring, enabling the control of wavelength load while limiting the amount of detoured traffic. A key feature of the architecture is that its degree of wavelength reconfigurability is set by a single tunable parameter, the borrowing degree $B$, ranging from none to full; performance evaluation shows that performance nearly equal to that of a fully reconfigurable architecture is achieved well below the maximum $B$, saving the complexity of a fully reconfigurable solutions. The optical fabric relies on mature, data-center-grade components, namely AWGs, AWGRs, colorless OXCs, and combiners. The reconfiguration speed and the adopted dynamic control policy make the architecture suitable for network tiers where the traffic varies on timescales of tens of minutes or more, e.g., among groups of racks (pods).

## Who Needs DRAM? We Have Fiber

DevFeed: [Who Needs DRAM? We Have Fiber](<https://devfeed.tech/articles/who-needs-dram-we-have-fiber-119086.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.08407>)

Author: Hannah Atmer, Yuan Yao, Thiemo Voigt, Stefanos Kaxiras

Published: 2026-11-05T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [DRAM price increase](<https://devfeed.tech/topics/dram-price-increase.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [accelerators](<https://devfeed.tech/tags/accelerators.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-accelerators](<https://devfeed.tech/tags/ai-accelerators.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [availability](<https://devfeed.tech/tags/availability.md>), [broadcast](<https://devfeed.tech/tags/broadcast.md>), [co-packaged-optics](<https://devfeed.tech/tags/co-packaged-optics.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-et](<https://devfeed.tech/tags/cs-et.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [data](<https://devfeed.tech/tags/data.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [delay](<https://devfeed.tech/tags/delay.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [dram](<https://devfeed.tech/tags/dram.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [fiber](<https://devfeed.tech/tags/fiber.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [global](<https://devfeed.tech/tags/global.md>), [hbm3e](<https://devfeed.tech/tags/hbm3e.md>), [high-performance](<https://devfeed.tech/tags/high-performance.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [hyperscale](<https://devfeed.tech/tags/hyperscale.md>)

### AI overview

Fiber Memory proposes using optical fiber as an active, recirculating delay-line memory for immutable data such as large language model weights. Its hybrid data- and pipeline-parallel optical broadcast design combines multi-core fiber, passive optical tap-and-amplify interfaces, co-packaged optics, and centralized in-ring re-amplify, reshape, and retime regeneration. In a case study, the authors estimate it could eliminate redundant weight storage across tens of thousands of AI accelerators and reduce weight-delivery power by over 75% versus traditional HBM3e configurations. These are case study results, not evidence of a deployed system.

### Source excerpt

arXiv:2607.08407v3 Announce Type: replace-cross Abstract: The 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 weights. We present a hybrid data- and pipeline-parallel optical broadcast delay-line memory architecture that accounts for fiber's physical realities. By incorporating space-division multiplexed multi-core fibers, passive optical tap-and-amplify interfaces, co-packaged optics, and centralized in-ring 3R (re-amplify, reshape, retime) regeneration, our case study evaluation suggests that Fiber Memory can eliminate redundant weight storage across tens of thousands of AI accelerators and reduce weight-delivery power by over 75% compared to traditional HBM3e configurations.

## Celestica DS6005 Puts 102.4Tbps of Co-Packaged Optics and 64 1.6TbE Ports in a 1.5OU Liquid-Cooled ORv3 Switch

DevFeed: [Celestica DS6005 Puts 102.4Tbps of Co-Packaged Optics and 64 1.6TbE Ports in a 1.5OU Liquid-Cooled ORv3 Switch](<https://devfeed.tech/articles/celestica-ds6005-puts-102-4tbps-of-co-packaged-optics-and-64-1-6tbe-ports-in-a-1-5ou-liquid-cooled-orv3-switch-88005.md>)

Original publisher: [Read original article](<https://www.storagereview.com/news/celestica-ds6005-102-4tbps-co-packaged-optics-1-5ou-liquid-cooled-orv3-switch>)

Author: Harold Fritts

Published: 2026-10-15T16:50:02Z

Content type: news

Language: en

Sources: [StorageReview.com](<https://devfeed.tech/sources/storagereview-com.md>)

Topics: [quasi-random network topology](<https://devfeed.tech/topics/quasi-random-network-topology.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>)

Tags: [102-4t](<https://devfeed.tech/tags/102-4t.md>), [amd](<https://devfeed.tech/tags/amd.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [co-packaged-optics](<https://devfeed.tech/tags/co-packaged-optics.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [hardware-root-of-trust](<https://devfeed.tech/tags/hardware-root-of-trust.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [networking](<https://devfeed.tech/tags/networking.md>), [ocp](<https://devfeed.tech/tags/ocp.md>), [optics](<https://devfeed.tech/tags/optics.md>), [switch](<https://devfeed.tech/tags/switch.md>), [v3](<https://devfeed.tech/tags/v3.md>)

### AI overview

Celestica introduced the DS6005, a 102.4 Tbps co-packaged optics data center switch for Open Rack v3, with 64 optically enabled 1.6 TbE ports and liquid cooling. It uses Broadcom's Tomahawk 6 Davisson package, supports field-replaceable external laser modules, and runs SONiC. Celestica has not published power or latency figures or availability dates.

### Source excerpt

Celestica introduced the DS6005 at the OCP Global Summit, a co-packaged optics (CPO) data center switch that puts 102.4Tbps of switching capacity and 64 optically enabled 1.6TbE ports in a 1.5OU, fully liquid-cooled chassis for Open Rack v3. It is built on Broadcom's Tomahawk 6 Davisson, the CPO version of the 102.4T Tomahawk 6, which The post Celestica DS6005 Puts 102.4Tbps of Co-Packaged Optics and 64 1.6TbE Ports in a 1.5OU Liquid-Cooled ORv3 Switch appeared first on StorageReview.com.

## Open specs could cut AI data center hardware integration from a year to months

DevFeed: [Open specs could cut AI data center hardware integration from a year to months](<https://devfeed.tech/articles/open-specs-could-cut-ai-data-center-hardware-integration-from-a-year-to-months-87335.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/14/hardware-interoperability-open-specs-speed-ai-data-center-builds-aidatapipelineforum/>)

Author: Sloane Kali Faye

Published: 2026-10-14T23:30:09Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [GPUDirect](<https://devfeed.tech/topics/gpudirect.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>)

Tags: [ai-data-center](<https://devfeed.tech/tags/ai-data-center.md>), [ai-data-pipeline-forum-2026](<https://devfeed.tech/tags/ai-data-pipeline-forum-2026.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-infrastructure-bottlenecks](<https://devfeed.tech/tags/ai-infrastructure-bottlenecks.md>), [aidatapipelineforum](<https://devfeed.tech/tags/aidatapipelineforum.md>), [amd](<https://devfeed.tech/tags/amd.md>), [composable-memory-systems](<https://devfeed.tech/tags/composable-memory-systems.md>), [compute-express-link](<https://devfeed.tech/tags/compute-express-link.md>), [cube-event-coverage](<https://devfeed.tech/tags/cube-event-coverage.md>), [cxl](<https://devfeed.tech/tags/cxl.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [hardware-interoperability](<https://devfeed.tech/tags/hardware-interoperability.md>), [hyperscale-computing](<https://devfeed.tech/tags/hyperscale-computing.md>), [infra](<https://devfeed.tech/tags/infra.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [interoperability](<https://devfeed.tech/tags/interoperability.md>), [john-furrier](<https://devfeed.tech/tags/john-furrier.md>), [microsoft](<https://devfeed.tech/tags/microsoft.md>), [news](<https://devfeed.tech/tags/news.md>), [open-compute-project](<https://devfeed.tech/tags/open-compute-project.md>), [open-hardware](<https://devfeed.tech/tags/open-hardware.md>), [open-standards](<https://devfeed.tech/tags/open-standards.md>), [samsung](<https://devfeed.tech/tags/samsung.md>), [scale](<https://devfeed.tech/tags/scale.md>), [scaleflux](<https://devfeed.tech/tags/scaleflux.md>), [siamak-tavallaei](<https://devfeed.tech/tags/siamak-tavallaei.md>), [thecube](<https://devfeed.tech/tags/thecube.md>)

### AI overview

Open hardware specifications and community-developed integration are presented as ways to improve interoperability among AI data center suppliers. A Samsung engineer says this approach could shorten hyperscalers' integration of new designs from roughly nine to twelve months to three to six months. The article also discusses CXL memory pooling and its incorporation into an OCP subproject.

### Source excerpt

As AI factories scale, hardware interoperability is becoming as important as raw performance. Accelerators, memory and storage now come from many different suppliers, and every mismatch between them slows deployment across the data center. Pressure is mounting as AI infrastructure bottlenecks shift attention from individual chips to entire systems. Open communities such as the Open [...] The post Open specs could cut AI data center hardware integration from a year to months appeared first on SiliconANGLE.

## Salience Labs raises $125M after showcasing world's most advanced, fully optical circuit-switching tech

DevFeed: [Salience Labs raises $125M after showcasing world's most advanced, fully optical circuit-switching tech](<https://devfeed.tech/articles/salience-labs-raises-125m-after-showcasing-world-s-most-advanced-fully-optical-circuit-switching-tech-86760.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/14/salience-labs-raises-125m-in-funding-after-showcasing-worlds-most-advanced-fully-optical-circuit-switching-tech/>)

Author: Mike Wheatley

Published: 2026-10-14T13:00:27Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [Chip design](<https://devfeed.tech/topics/chip-design.md>), [Frontier AI](<https://devfeed.tech/topics/frontier-ai.md>)

Tags: [ai-data-center](<https://devfeed.tech/tags/ai-data-center.md>), [ai-data-centers](<https://devfeed.tech/tags/ai-data-centers.md>), [ai-workloads](<https://devfeed.tech/tags/ai-workloads.md>), [amd](<https://devfeed.tech/tags/amd.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [aws](<https://devfeed.tech/tags/aws.md>), [bottlenecks](<https://devfeed.tech/tags/bottlenecks.md>), [chips](<https://devfeed.tech/tags/chips.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [compute](<https://devfeed.tech/tags/compute.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cost](<https://devfeed.tech/tags/cost.md>), [fully-optical-network-switch](<https://devfeed.tech/tags/fully-optical-network-switch.md>), [funding](<https://devfeed.tech/tags/funding.md>), [gpu-clusters](<https://devfeed.tech/tags/gpu-clusters.md>), [industry](<https://devfeed.tech/tags/industry.md>), [infra](<https://devfeed.tech/tags/infra.md>), [intel](<https://devfeed.tech/tags/intel.md>), [led](<https://devfeed.tech/tags/led.md>), [network-switches](<https://devfeed.tech/tags/network-switches.md>), [networking-chokepoints](<https://devfeed.tech/tags/networking-chokepoints.md>), [news](<https://devfeed.tech/tags/news.md>), [optical-connectivity](<https://devfeed.tech/tags/optical-connectivity.md>), [optical-networking](<https://devfeed.tech/tags/optical-networking.md>), [optical-switching](<https://devfeed.tech/tags/optical-switching.md>), [salience-labs](<https://devfeed.tech/tags/salience-labs.md>), [series-b](<https://devfeed.tech/tags/series-b.md>), [silicon-photonics](<https://devfeed.tech/tags/silicon-photonics.md>), [switching](<https://devfeed.tech/tags/switching.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>), [top-story-6](<https://devfeed.tech/tags/top-story-6.md>)

### AI overview

Salience Labs raised $125 million in a Series B round to expand manufacturing and develop higher-density optical switches for AI data centers. The company demonstrated a 64-port all-optical photonic switch and is developing 128-port and 256-port configurations. Its approach routes data through optical connections to address bandwidth bottlenecks, latency, and power consumption in AI computing clusters.

### Source excerpt

A startup called Salience Labs Inc. wants to play a bigger role in speeding up artificial intelligence workloads after raising $125 million in a Series B funding. Today's round was led by Bullhound Capital and saw the two of the chipmaking industry's most iconic names participate via their investment arms, Intel Capital AMD Ventures. SBI Group, [...] The post Salience Labs raises $125M after showcasing world's most advanced, fully optical circuit-switching tech appeared first on SiliconANGLE.

## Cisco Systems scales programmable AI network capacity with new silicon and repluggable optics innovations

DevFeed: [Cisco Systems scales programmable AI network capacity with new silicon and repluggable optics innovations](<https://devfeed.tech/articles/cisco-systems-scales-programmable-ai-network-capacity-with-new-silicon-and-repluggable-optics-innovations-85289.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/12/cisco-systems-scales-programmable-ai-network-capacity-with-new-silicon-and-repluggable-optics-innovations/>)

Author: Mike Wheatley

Published: 2026-10-12T15:00:03Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [future of AI](<https://devfeed.tech/topics/future-of-ai.md>)

Tags: [ai-cloud](<https://devfeed.tech/tags/ai-cloud.md>), [ai-clusters](<https://devfeed.tech/tags/ai-clusters.md>), [ai-data-centers](<https://devfeed.tech/tags/ai-data-centers.md>), [ai-factories](<https://devfeed.tech/tags/ai-factories.md>), [ai-networking](<https://devfeed.tech/tags/ai-networking.md>), [ai-networks](<https://devfeed.tech/tags/ai-networks.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [aws](<https://devfeed.tech/tags/aws.md>), [chip](<https://devfeed.tech/tags/chip.md>), [chips](<https://devfeed.tech/tags/chips.md>), [cisco](<https://devfeed.tech/tags/cisco.md>), [cisco-systems](<https://devfeed.tech/tags/cisco-systems.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [coherent-optics](<https://devfeed.tech/tags/coherent-optics.md>), [compute](<https://devfeed.tech/tags/compute.md>), [data](<https://devfeed.tech/tags/data.md>), [infra](<https://devfeed.tech/tags/infra.md>), [network](<https://devfeed.tech/tags/network.md>), [networking](<https://devfeed.tech/tags/networking.md>), [news](<https://devfeed.tech/tags/news.md>), [open-compute-project](<https://devfeed.tech/tags/open-compute-project.md>), [open-compute-project-global-summit](<https://devfeed.tech/tags/open-compute-project-global-summit.md>), [optics](<https://devfeed.tech/tags/optics.md>), [pluggable-optics](<https://devfeed.tech/tags/pluggable-optics.md>), [scale-across-architecture](<https://devfeed.tech/tags/scale-across-architecture.md>), [scale-across-networks](<https://devfeed.tech/tags/scale-across-networks.md>), [silicon-one-a100](<https://devfeed.tech/tags/silicon-one-a100.md>), [silicon-one-g300](<https://devfeed.tech/tags/silicon-one-g300.md>), [silicon-one-g302](<https://devfeed.tech/tags/silicon-one-g302.md>), [silicon-one-p200](<https://devfeed.tech/tags/silicon-one-p200.md>), [sonic](<https://devfeed.tech/tags/sonic.md>), [stage](<https://devfeed.tech/tags/stage.md>), [summit](<https://devfeed.tech/tags/summit.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>), [top-story-4](<https://devfeed.tech/tags/top-story-4.md>), [ultra-dense-small-form-factor-repluggable](<https://devfeed.tech/tags/ultra-dense-small-form-factor-repluggable.md>)

### AI overview

Cisco announced networking silicon, optical modules, a programmable management system, and a SONiC security feature aimed at increasing capacity and resilience for AI data center networks. The announcements include the 102.4-terabit Silicon One G300 shipping at volume, the 51.2-terabit G302 expected before year-end, and a 12.8-terabit-per-second optical module developed with Nvidia.

### Source excerpt

Cisco Systems Inc. took to the stage at the annual Open Compute Project Global Summit today, unveiling a new suite of open silicon and optical networking systems as it delivered one of the event's most eagerly anticipated keynotes. The new innovations are designed to meet the growing demands of power- and data-hungry artificial intelligence workloads. [...] The post Cisco Systems scales programmable AI network capacity with new silicon and repluggable optics innovations appeared first on SiliconANGLE.

## Powering predictable costs with AWS Direct Connect flat-rate pricing

DevFeed: [Powering predictable costs with AWS Direct Connect flat-rate pricing](<https://devfeed.tech/articles/powering-predictable-costs-with-aws-direct-connect-flat-rate-pricing-65226.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/networking-and-content-delivery/powering-predictable-costs-with-aws-direct-connect-flat-rate-pricing/>)

Author: Donald Quindardo

Published: 2026-10-05T19:56:23Z

Content type: tutorial

Language: en

Sources: [Networking & Content Delivery](<https://devfeed.tech/sources/networking-content-delivery.md>)

Topics: [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [Amazon Web Services](<https://devfeed.tech/topics/aws.md>), [cost-optimization](<https://devfeed.tech/topics/cost-optimization.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [API Governance](<https://devfeed.tech/topics/api-governance.md>)

Tags: [amazon-cloudwatch](<https://devfeed.tech/tags/amazon-cloudwatch.md>), [amazon-vpc](<https://devfeed.tech/tags/amazon-vpc.md>), [aws](<https://devfeed.tech/tags/aws.md>), [aws-cost-explorer](<https://devfeed.tech/tags/aws-cost-explorer.md>), [aws-direct-connect](<https://devfeed.tech/tags/aws-direct-connect.md>), [aws-support](<https://devfeed.tech/tags/aws-support.md>), [aws-transit-gateway](<https://devfeed.tech/tags/aws-transit-gateway.md>), [cost-optimization](<https://devfeed.tech/tags/cost-optimization.md>), [costs](<https://devfeed.tech/tags/costs.md>), [hybrid-connectivity](<https://devfeed.tech/tags/hybrid-connectivity.md>), [intermediate-200](<https://devfeed.tech/tags/intermediate-200.md>), [networking](<https://devfeed.tech/tags/networking.md>), [networking-content-delivery](<https://devfeed.tech/tags/networking-content-delivery.md>), [pricing](<https://devfeed.tech/tags/pricing.md>), [resiliency](<https://devfeed.tech/tags/resiliency.md>)

### AI overview

The article explains how AWS Direct Connect flat-rate pricing applies a fixed hourly charge with no per-gigabyte data transfer out charge for covered Regions. It walks through configuring connections and resiliency groups, selecting geographic tiers, converting existing connections, and evaluating costs with CloudWatch and Cost Explorer. Examples cover single-Region and multi-Region workloads, failover, and separate billing modes across accounts. It also describes limitations, including standard charges for traffic outside a subscribed tier and exclusions such as hosted connections, SiteLink, and provider fees.

### Source excerpt

In a traditional on-premises network, predictable costs have always been central to network connectivity budgets. Organizations relied on leased lines, Multiprotocol Label Switching (MPLS) links, and dedicated cross-connects. No matter how much data they moved, the bill stayed the same. As they rebuilt their architecture in the cloud, they gained the flexibility of pay-as-you-go pricing. [...]

## HPE's Rahim: Pace of change in the data center is 'extraordinary'

DevFeed: [HPE's Rahim: Pace of change in the data center is 'extraordinary'](<https://devfeed.tech/articles/hpe-s-rahim-pace-of-change-in-the-data-center-is-extraordinary-63627.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4229791/hpes-rahim-pace-of-change-in-the-data-center-is-extraordinary.html>)

Author: Michael Cooney

Published: 2026-10-01T22:22:45Z

Content type: news

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [scalable-architecture](<https://devfeed.tech/topics/scalable-architecture.md>), [Software-Defined Data Center](<https://devfeed.tech/topics/software-defined-data-center.md>), [NVIDIA Spectrum-X Ethernet](<https://devfeed.tech/topics/nvidia-spectrum-x-ethernet.md>), [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [Load Balancing](<https://devfeed.tech/topics/load-balancing.md>), [Project Jupiter](<https://devfeed.tech/topics/project-jupiter.md>), [GPUDirect](<https://devfeed.tech/topics/gpudirect.md>), [quasi-random network topology](<https://devfeed.tech/topics/quasi-random-network-topology.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [amd](<https://devfeed.tech/tags/amd.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-data-center-enterprise-routers-networking-networking-devices-sd-branch](<https://devfeed.tech/tags/artificial-intelligence-data-center-enterprise-routers-networking-networking-devices-sd-branch.md>), [broadcom](<https://devfeed.tech/tags/broadcom.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [data-center-networking](<https://devfeed.tech/tags/data-center-networking.md>), [enterprise-routers](<https://devfeed.tech/tags/enterprise-routers.md>), [ethernet](<https://devfeed.tech/tags/ethernet.md>), [helios](<https://devfeed.tech/tags/helios.md>), [hpe](<https://devfeed.tech/tags/hpe.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [juniper-networks](<https://devfeed.tech/tags/juniper-networks.md>), [networking](<https://devfeed.tech/tags/networking.md>), [networking-devices](<https://devfeed.tech/tags/networking-devices.md>), [wan](<https://devfeed.tech/tags/wan.md>)

### AI overview

At HPE's Networking Investor Day, executives described rising demand for AI data center networking and outlined growth expectations across networking, routing, campus and branch, and security. HPE highlighted an integrated AMD Helios rack scale system using scale-up Ethernet, while citing network orders and projected revenue growth as evidence of the opportunity.

### Source excerpt

Following its record-setting third quarter, HPE said it expects cumulative AI network orders for fiscal 2026 to top $3 billion, exceeding its previous forecasts. The company also raised its fiscal 2027 networking revenue target, projecting growth percentages in the high teens to low 20s. Specific to data center networking, HPE is estimating revenue to grow at roughly 50% CAGR through fiscal 2029. "The pace of change in the data center is, quite frankly, extraordinary. AI has fundamentally reshaped how infrastructure is designed and deployed," Rami Rahim, executive vice president, president and general manager of HPE Networking, told analysts and others at the annual HPE Networking Investor Day. "And the reason is simple: You can invest billions of dollars into GPUs, but if the network can't keep up with the traffic and the performance demands, those resources are not fully utilized. So, as a result, networking has become a much more strategic part of the AI infrastructure stack." "We are taking standards-based Ethernet to a new layer of the data center stack for the first time, so it's unlocking a massive TAM [total addressable market]. And if you think about the capacity requirements within the data center, scale-out is orders of magnitude greater than scale-across, and scale-up is orders of magnitude greater than scale-out," said Rahim (pictured above at a 2025 event). "The capacity requirements are stunning in this area, so it unlocks a significant new TAM for us." Rahim pointed to HPE's integrated AMD Helios package as an example of what an integrated Juniper/HPE will deliver. In December, HPE said it will support AMD's Helios AI rack scale architecture with integrated scale-up Ethernet networking. The system basically will be a scale-up turnkey Ethernet package that will feature a purpose-built Juniper Networks switch based on Broadcom's 102.4 Tbps. Tomahawk 6 network silicon and employ the Ultra Accelerator Link over Ethernet (UALoE) specification. "We are bri

## Scale-up interconnect startup CScale launches with $188M in funding

DevFeed: [Scale-up interconnect startup CScale launches with $188M in funding](<https://devfeed.tech/articles/scale-up-interconnect-startup-cscale-launches-with-188m-in-funding-62961.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/09/30/scale-up-interconnect-startup-cscale-launches-with-188m-in-funding/>)

Author: Maria Deutscher

Published: 2026-10-01T00:35:03Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [NVLink](<https://devfeed.tech/topics/nvlink.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Product Management](<https://devfeed.tech/topics/product-management.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-workloads](<https://devfeed.tech/tags/ai-workloads.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [components](<https://devfeed.tech/tags/components.md>), [copper](<https://devfeed.tech/tags/copper.md>), [cscale](<https://devfeed.tech/tags/cscale.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [developing](<https://devfeed.tech/tags/developing.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [executive](<https://devfeed.tech/tags/executive.md>), [funding](<https://devfeed.tech/tags/funding.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [infra](<https://devfeed.tech/tags/infra.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [interconnect](<https://devfeed.tech/tags/interconnect.md>), [network](<https://devfeed.tech/tags/network.md>), [news](<https://devfeed.tech/tags/news.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [optical-interconnect](<https://devfeed.tech/tags/optical-interconnect.md>), [scale](<https://devfeed.tech/tags/scale.md>), [scale-up](<https://devfeed.tech/tags/scale-up.md>), [scale-up-architecture](<https://devfeed.tech/tags/scale-up-architecture.md>), [series-c](<https://devfeed.tech/tags/series-c.md>), [startup](<https://devfeed.tech/tags/startup.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>)

### AI overview

CScale, a startup developing optical data center interconnects, launched with $188 million in funding. Its technology is designed to link thousands of accelerators across multiple racks in large AI clusters and contain optical-link failures, though the company did not explain how it achieves that reliability.

### Source excerpt

CScale Inc., a startup that develops optical interconnects for data centers, launched today with $188 million in funding. The company raised about three-quarters of the capital through a recently closed Series C round. It was co-led by Atreides Management, Valor Equity Partners and Premji Invest. The funds were joined by several other backers including Nvidia [...] The post Scale-up interconnect startup CScale launches with $188M in funding appeared first on SiliconANGLE.

## AI-driven networking and infrastructure design at APNIC 62

DevFeed: [AI-driven networking and infrastructure design at APNIC 62](<https://devfeed.tech/articles/ai-driven-networking-and-infrastructure-design-at-apnic-62-62287.md>)

Original publisher: [Read original article](<https://blog.apnic.net/2026/09/30/ai-driven-networking-and-infrastructure-design-at-apnic-62/>)

Author: George Michaelson

Published: 2026-09-30T03:02:50Z

Content type: article

Language: en

Sources: [APNIC Blog](<https://devfeed.tech/sources/apnic-blog.md>)

Topics: [Network design](<https://devfeed.tech/topics/network-design.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [clean energy data centers](<https://devfeed.tech/topics/clean-energy-data-centers.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [apnic-62](<https://devfeed.tech/tags/apnic-62.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [network-design](<https://devfeed.tech/tags/network-design.md>), [network-operators](<https://devfeed.tech/tags/network-operators.md>), [networking](<https://devfeed.tech/tags/networking.md>), [scale](<https://devfeed.tech/tags/scale.md>), [tech-matters](<https://devfeed.tech/tags/tech-matters.md>)

### AI overview

Presenters at APNIC 62 discussed how AI workloads are changing data centre network design and operations. The session covered network architectures, autonomous management, predictive fault detection, and AI-assisted geofeed authoring, including a three-spine interconnect design used by Lightstorm.

### Source excerpt

Presenters at APNIC 62 explored how AI is reshaping network design and operations, from next-generation optical architectures to autonomous management systems, highlighting scale challenges and improved reliability, efficiency, and operational decision-making.

## Tackling the three hidden mistakes when planning a GCC

DevFeed: [Tackling the three hidden mistakes when planning a GCC](<https://devfeed.tech/articles/tackling-the-three-hidden-mistakes-when-planning-a-gcc-61017.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4223892/tackling-the-three-hidden-mistakes-when-planning-a-gcc.html>)

Author: jwood

Published: 2026-09-28T08:26:32Z

Content type: article

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [it modernization](<https://devfeed.tech/topics/it-modernization.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [AI Strategy](<https://devfeed.tech/topics/ai-strategy.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>)

Tags: [boston](<https://devfeed.tech/tags/boston.md>), [data-infrastructure](<https://devfeed.tech/tags/data-infrastructure.md>), [data-residency](<https://devfeed.tech/tags/data-residency.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [genai](<https://devfeed.tech/tags/genai.md>), [india](<https://devfeed.tech/tags/india.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [it-leadership](<https://devfeed.tech/tags/it-leadership.md>), [it-leadership-it-strategy](<https://devfeed.tech/tags/it-leadership-it-strategy.md>), [it-strategy](<https://devfeed.tech/tags/it-strategy.md>), [mistakes](<https://devfeed.tech/tags/mistakes.md>), [mlops](<https://devfeed.tech/tags/mlops.md>)

### AI overview

The article argues that enterprises should plan global capability centers (GCCs) as strategic business platforms rather than offshore offices. It highlights digital foundation concerns including data residency, network capacity for AI workloads, security risks from third-party access, architecture, connectivity, and vendor governance, and says these decisions should be addressed early.

### Source excerpt

The decisions that determine whether a global capability center (GCC) succeeds or stalls are rarely the ones that get boardroom attention. With the number of GCCs in India growing by nearly a third since 2021 - more than 500 of the Forbes Global 2000 operate in the country - the choices that underpin a successful launch are treated as operational detail and left too late in the process. Consider that nearly two thirds (65%) of GCCs are "average performers" according to a Boston Consulting Group study, with a further 8% underperforming. Just four out of every 50 qualify as "top performers." "How do you bring everything under one roof?" asks Murali Krishnan, AVP & Head of Business - Enterprise Network at Tata Communications. "Otherwise, when these things are highly distributed, it has a dilution factor with respect to what you potentially want to achieve - for example, if you are going to go through an artificial intelligence transformation. Having your entire knowledge in one place makes a lot more sense for an enterprise rather than having a distributed global workforce." This reflects the shift in what a GCC is expected to deliver. While cost arbitrage got the GCC operating model started, transformation and speed-to-market are now driving investment. The centers capturing that shift are the ones that sequence decisions correctly before they break ground. "Design your GCC as a strategic business platform, not an offshore office," says Anubhav Bhatnagar, Associate Director - Enterprise Connectivity Products at Tata Communications. He counts six fault lines in a GCC's digital foundation: data residency rules, AI workloads on networks never designed for them, a security perimeter that dissolves the moment a third party gets access. These sit alongside architecture, connectivity and vendor governance. In this article we are going to explore the three biggest mistakes which expose these fault lines, and how enterprises can correct them to set up a GCC for future success.

## The AI race built faster chips. Now the chips are too fast for their own wires.

DevFeed: [The AI race built faster chips. Now the chips are too fast for their own wires.](<https://devfeed.tech/articles/the-ai-race-built-faster-chips-now-the-chips-are-too-fast-for-their-own-wires-82934.md>)

Original publisher: [Read original article](<https://thenextweb.com/news/eliyan-145m-series-c-unicorn-ai-chip-interconnects>)

Author: ssanders

Published: 2026-07-30T18:04:28Z

Content type: news

Language: en

Sources: [Eliyan](<https://devfeed.tech/sources/eliyan.md>)

Topics: [nuclear energy for AI](<https://devfeed.tech/topics/nuclear-energy-for-ai.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>)

Tags: [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [amd](<https://devfeed.tech/tags/amd.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [ayar-labs](<https://devfeed.tech/tags/ayar-labs.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [deep-tech](<https://devfeed.tech/tags/deep-tech.md>), [eliyan-in-the-news](<https://devfeed.tech/tags/eliyan-in-the-news.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [interconnect](<https://devfeed.tech/tags/interconnect.md>), [investors-and-funding](<https://devfeed.tech/tags/investors-and-funding.md>), [semiconductors](<https://devfeed.tech/tags/semiconductors.md>), [startups-and-technology](<https://devfeed.tech/tags/startups-and-technology.md>), [united-states](<https://devfeed.tech/tags/united-states.md>)

### AI overview

Eliyan raised $145 million in a Series C at a $1 billion valuation to develop interconnect technology that moves data between AI chips and data center racks. The company is expanding from electrical links to optical interconnects, with funding for development, partnerships and manufacturing.

### Source excerpt

Eliyan raised a $145m Series C at a $1bn valuation to fix the data bottleneck between AI chips, adding Cisco and Lumentum for its optical push.

## Optical interconnect startup Eliyan raises $145M in funding to solve the AI chip cluster connectivity crunch

DevFeed: [Optical interconnect startup Eliyan raises $145M in funding to solve the AI chip cluster connectivity crunch](<https://devfeed.tech/articles/optical-interconnect-startup-eliyan-raises-145m-in-funding-to-solve-the-ai-chip-cluster-connectivity-crunch-82932.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/07/29/optical-interconnect-startup-eliyan-raises-145m-funding-solve-ai-chip-cluster-connectivity-crunch/>)

Author: ssanders

Published: 2026-07-29T18:00:40Z

Content type: news

Language: en

Sources: [Eliyan](<https://devfeed.tech/sources/eliyan.md>)

Topics: [NVLink](<https://devfeed.tech/topics/nvlink.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>)

Tags: [ai-chip-clusters](<https://devfeed.tech/tags/ai-chip-clusters.md>), [ai-clusters](<https://devfeed.tech/tags/ai-clusters.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [chip-interconnects](<https://devfeed.tech/tags/chip-interconnects.md>), [chip-networks](<https://devfeed.tech/tags/chip-networks.md>), [data-center-interconnect](<https://devfeed.tech/tags/data-center-interconnect.md>), [data-center-networks](<https://devfeed.tech/tags/data-center-networks.md>), [eliyan](<https://devfeed.tech/tags/eliyan.md>), [eliyan-in-the-news](<https://devfeed.tech/tags/eliyan-in-the-news.md>), [funding](<https://devfeed.tech/tags/funding.md>), [infra](<https://devfeed.tech/tags/infra.md>), [interconnect](<https://devfeed.tech/tags/interconnect.md>), [interconnects](<https://devfeed.tech/tags/interconnects.md>), [networking](<https://devfeed.tech/tags/networking.md>), [news](<https://devfeed.tech/tags/news.md>), [optical-interconnect](<https://devfeed.tech/tags/optical-interconnect.md>), [ramin-farjadrad](<https://devfeed.tech/tags/ramin-farjadrad.md>), [seligman-ventures](<https://devfeed.tech/tags/seligman-ventures.md>), [series-c](<https://devfeed.tech/tags/series-c.md>), [startup](<https://devfeed.tech/tags/startup.md>)

### AI overview

Eliyan raised $145 million in a Series C round at a reported $1 billion valuation. The chip interconnect startup says its electro-optical NuLink PHYs and NuGear chiplets can improve bandwidth and power efficiency across AI compute, memory, and networking systems. It plans to use the funding for research, product development, manufacturing capacity, and partner expansion.

### Source excerpt

Optical interconnect startup Eliyan raises $145M in funding to solve the AI chip cluster connectivity crunch - SiliconANGLE

## Why didn't ngrok go down in last week's AWS outage?

DevFeed: [Why didn't ngrok go down in last week's AWS outage?](<https://devfeed.tech/articles/why-didn-t-ngrok-go-down-in-last-week-s-aws-outage-83221.md>)

Original publisher: [Read original article](<https://ngrok.com/blog/dont-use-us-east-1>)

Author: Peter Yoakum

Published: 2025-10-29T00:00:00Z

Content type: article

Language: en

Sources: [\[ngrok news\]](<https://devfeed.tech/sources/ngrok-news.md>)

Topics: [Amazon Web Services](<https://devfeed.tech/topics/aws.md>), [cloud service outage management](<https://devfeed.tech/topics/cloud-service-outage-management.md>), [Amazon Route 53](<https://devfeed.tech/topics/amazon-route-53.md>), [AWS data center network architecture](<https://devfeed.tech/topics/aws-data-center-network-architecture.md>), [Amazon EC2](<https://devfeed.tech/topics/amazon-ec2.md>), [Post Mortem](<https://devfeed.tech/topics/post-mortem.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [aws](<https://devfeed.tech/tags/aws.md>), [company](<https://devfeed.tech/tags/company.md>), [container](<https://devfeed.tech/tags/container.md>), [control-plane](<https://devfeed.tech/tags/control-plane.md>), [copy](<https://devfeed.tech/tags/copy.md>), [critical-infrastructure](<https://devfeed.tech/tags/critical-infrastructure.md>), [dashboard](<https://devfeed.tech/tags/dashboard.md>), [dns](<https://devfeed.tech/tags/dns.md>), [docker-container](<https://devfeed.tech/tags/docker-container.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [fail](<https://devfeed.tech/tags/fail.md>), [failover](<https://devfeed.tech/tags/failover.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [internet-infrastructure](<https://devfeed.tech/tags/internet-infrastructure.md>), [internet-traffic](<https://devfeed.tech/tags/internet-traffic.md>), [ngrok](<https://devfeed.tech/tags/ngrok.md>), [outage](<https://devfeed.tech/tags/outage.md>), [region](<https://devfeed.tech/tags/region.md>), [traffic](<https://devfeed.tech/tags/traffic.md>), [us](<https://devfeed.tech/tags/us.md>)

### AI overview

ngrok describes how it stayed available during an AWS us-east-1 outage by operating from us-east-2 and designing its network for regional failover. Its DNS-based load balancing removes affected points of presence from resolution, while each point of presence maintains local configuration and handles traffic policy actions in-region.

### Source excerpt

How ngrok stayed available during the recent AWS us-east-1 outage, why we avoid the region, and the design choices that keep customer traffic flowing.