# rack Published articles for rack. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Smartbird Adds Supermicro DCBBS Racks to Its Managed Single-Tenant AI Clusters, Two Weeks After WEKA DevFeed: [Smartbird Adds Supermicro DCBBS Racks to Its Managed Single-Tenant AI Clusters, Two Weeks After WEKA]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-11-08T19:07:39Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [AI Infrastructure](), [Managed Services](), [digital sovereignty](), [tenant data protection]() Tags: [added](), [agentic](), [agentic-ai](), [ai](), [ai-clusters](), [ai-data-centers](), [ai-infrastructure](), [availability](), [data-center](), [deployment](), [design](), [enterprise](), [managed](), [network](), [operations](), [partner](), [procurement](), [rack](), [rack-scale](), [solutions](), [supermicro]() ### AI overview Smartbird and Supermicro will offer dedicated, single-tenant AI clusters built with Supermicro DCBBS and rack-scale systems. Smartbird is responsible for design, procurement, integration, deployment, and ongoing operations; customers can choose on-premises or colocation deployments tailored to their processor, capacity, performance, and workload needs, including training, fine-tuning, inference, agentic AI, and HPC. The arrangement combines Supermicro's compute, storage, networking, power, cooling, and supplier network with Smartbird's managed service, reducing vendors and handoffs. Single tenancy provides isolation and operational control, while deployment location can support data residency requirements; the announcement does not identify compliance certifications or establish that configurations meet specific regulations. It also gives no configurations, GPU platforms, availability dates, customers, or regions. ### Source excerpt Smartbird, the managed AI infrastructure company that launched in June and trades on Nasdaq as BIRD, has added Supermicro to its partner network. The two will deliver dedicated, single-tenant AI clusters built from Supermicro's Data Center Building Block Solutions and rack-scale systems, with Smartbird handling design, procurement, integration, deployment, and day-to-day operations as the single The post Smartbird Adds Supermicro DCBBS Racks to Its Managed Single-Tenant AI Clusters, Two Weeks After WEKA appeared first on StorageReview.com. ## Blog Review: Nov. 4 DevFeed: [Blog Review: Nov. 4]() Original publisher: [Read original article]() Author: Jesse Allen Published: 2026-11-04T08:01:02Z Content type: news Language: en Sources: [Semiconductor Engineering]() Topics: [Chip design](), [Physical AI](), [data centers]() Tags: [agents](), [ai](), [ai-data-center](), [ai-data-centers](), [ai-software](), [architecture](), [arm](), [blog](), [blog-reviews](), [bugs](), [business](), [cadence](), [challenges](), [chiplet](), [chiplets](), [chips](), [complexity](), [data-center](), [data-centers](), [keysight](), [matrix](), [news](), [physical-ai](), [rack](), [review](), [semi](), [siemens](), [synopsys](), [system-level-design](), [verification]() ### AI overview The post surveys semiconductor engineering topics spanning physical AI, rack-scale design, verification, matrix workloads, and data center infrastructure. Cadence describes an OCP FCSA workstream intended to add requirements for real-time, safety-critical physical systems while retaining the existing architecture. Synopsys notes that chip and package choices shape rack power and cooling, while rack constraints feed back into silicon design. Siemens warns that growing interaction diversity is challenging established approaches that manage verification complexity by decomposition. Arm describes SME2 quarter-tile and structured sparsity outer product instructions as improving support for small, dense, quantized, and sparse matrix workloads. Other entries cover AI data center energy and liquid cooling, projected wafer shipment growth, and newsletter pieces on verification, thermal analysis, regulation, debugging, and engineering workflows. ### Source excerpt Chiplets for physical AI; rack-level considerations; verification complexity; sparse matrix workloads; AI data centers. The post Blog Review: Nov. 4 appeared first on Semiconductor Engineering. ## Home Lab Beginners Guide: Hardware (2026) DevFeed: [Home Lab Beginners Guide: Hardware (2026)]() Original publisher: [Read original article]() Author: Hayden James Published: 2026-11-03T06:00:02Z Content type: article Language: en Sources: [Hayden James]() Topics: [Homelab](), [Tutorial](), [virtual machines]() Tags: [2026](), [beginners](), [beginners-guide](), [blog](), [hardware](), [home-lab](), [homelab](), [linux](), [rack](), [server](), [sysadmins]() ### AI overview The guide recommends starting a home lab small, choosing its location and equipment around heat, ventilation, space, access, cable runs and power, and planning room to grow. It explains when a rack or cabinet may suit different needs, then shares hardware lessons from the author's 12U setup: a UPS sized for clean shutdown, careful attention to power use and rack depth, and passive cooling that worked in an air-conditioned, low-dust room after fan failures. For networking, it offers example router and switch options while advising readers to research fit for their needs. The author reports that Cat6 has handled a 2.5G LAN and shorter 10G runs, and suggests used servers or a low-power mini PC as starting points. The author's setup has evolved from a cable modem and EdgeRouter to 10G uplinks, a pfSense firewall, 2.5G LAN and Checkmk RAW monitoring. ### Source excerpt My home lab in a 12U rack as of March 2026. What is a Home Lab? Think of a home lab as a place where you can fail in the privacy of your own home. Continue reading... ## Liquid-Cooled NVIDIA Servers at the ASRock Rack OCP 2026 Booth DevFeed: [Liquid-Cooled NVIDIA Servers at the ASRock Rack OCP 2026 Booth]() Original publisher: [Read original article]() Author: Patrick Kennedy Published: 2026-11-02T04:00:03Z Content type: news Language: en Sources: [ServeTheHome]() Topics: [Vera Rubin NVL72](), [NVIDIA Vera Rubin](), [Nvidia](), [NVLink](), [RTX PRO](), [NVIDIA Spectrum-X Ethernet]() Tags: [2026](), [ai](), [asrock-rack](), [blackwell](), [bluefield-4](), [bluefield-4-dpus](), [board](), [chip](), [compute](), [connectx](), [cooling](), [cpus](), [dc](), [liquid](), [liquid-cooling](), [nvidia](), [nvl72](), [nvl72-racks](), [ocp](), [rack](), [rubin](), [server](), [server-systems](), [servers](), [spectrum-x](), [switches](), [vera]() ### AI overview At OCP 2026, ASRock Rack showed NVIDIA systems spanning a 48RU Vera Rubin NVL72 rack, HGX Rubin NVL8 servers, RTX PRO systems, a Vera CPU server, and a liquid-cooled Spectrum-X switch. Liquid cooling was a central feature: the NVL72 compute trays and HGX systems use direct-to-chip cooling. The NVL72 rack combines 18 compute trays and nine NVLink 6 switch trays, totaling 36 Vera CPUs and 72 Rubin GPUs; ASRock Rack lists several cooling-distribution options, including in-rack and liquid-to-air sidecar CDUs. The 2U HGX Rubin NVL8 system fits eight Rubin GPUs in a fanless chassis and liquid-cools its drives, CPUs, and GPUs. It uses a dual-socket Intel Xeon 6 host, a 54V DC busbar, eight 800Gbps ConnectX-9 ports, and eight hot-swap E1.S NVMe bays. The excerpt gives NVIDIA performance figures for both platforms, but does not detail the other systems mentioned. ### Source excerpt We checked out the ASRock Rack booth at OCP 2026 and saw servers ranging from fully integrated NVIDIA NVL72 racks to liquid-cooled switches The post Liquid-Cooled NVIDIA Servers at the ASRock Rack OCP 2026 Booth appeared first on ServeTheHome. ## 6 insights from the GSA Women's Leadership Initiative: AI semiconductor industry trends turn the chip race into a systems race DevFeed: [6 insights from the GSA Women's Leadership Initiative: AI semiconductor industry trends turn the chip race into a systems race]() Original publisher: [Read original article]() Author: Jonathan Anthony Published: 2026-10-30T20:28:11Z Content type: news Language: en Sources: [SiliconANGLE]() Topics: [Chip design](), [Physical AI](), [Machine learning](), [communications](), [reliability-engineering]() Tags: [advanced-computing](), [advanced-packaging](), [advanced-semiconductor-engineering-inc](), [ai](), [ai-infrastructure](), [alisa-wang](), [alison-kosik](), [ansys](), [arm-ltd](), [artificial-intelligence](), [ase](), [career-advancement](), [chip](), [chip-design](), [chips](), [components](), [cost](), [credo-technology-group-holding-ltd](), [cube-event-coverage](), [cxl](), [decades](), [dmtf](), [emib-bridge-technology](), [engineering-leadership](), [executive-sponsorship](), [female-executives](), [gina-longoria](), [globalfoundries-inc](), [gsa-womens-leadership-initiative-2026](), [gsawomensleadershipinitiative](), [high-bandwith-compute](), [iliana-chen](), [industry](), [industry-trends](), [infra](), [initiative](), [insights](), [intel-corp](), [john-furrier](), [lakshmi-ramagiri](), [leadership](), [news](), [package](), [performance](), [rack](), [reliability](), [semiconductor](), [semiconductor-industry](), [systems](), [thecube](), [trends](), [women]() ### AI overview AI semiconductor competition is shifting from optimizing individual chips to coordinating whole systems: package, rack, connectivity and supply chain now shape performance, cost and reliability from the start. The article highlights six related developments: data movement is becoming an infrastructure bottleneck, prompting optical links to move closer to compute; advanced packaging and chiplet connections are becoming part of system architecture; AI analysis of design and manufacturing data can catch yield problems earlier; physical AI requires coordinated architecture, firmware, security and standards; sourcing and second suppliers are becoming product decisions under constrained capacity; and AI tools are entering chip design and verification as development cycles shorten. The reported gains and strategies are attributed to specific industry leaders and companies, rather than presented as universal results. ### Source excerpt For decades, the semiconductor industry competed on one thing: the individual chip. One of the biggest AI semiconductor industry trends now under way is the shift toward treating chips as components of much larger systems. Performance, cost and reliability are now determined across the package, the rack and the supply chain, not inside a single [...] The post 6 insights from the GSA Women's Leadership Initiative: AI semiconductor industry trends turn the chip race into a systems race appeared first on SiliconANGLE. ## Vertiv SmartIT Ships Factory-Validated MGX and ORV3 Racks for GB300 and Rubin: 120 to 142kW per Rack, 242kW N+1 Liquid Cooling DevFeed: [Vertiv SmartIT Ships Factory-Validated MGX and ORV3 Racks for GB300 and Rubin: 120 to 142kW per Rack, 242kW N+1 Liquid Cooling]() Original publisher: [Read original article]() Author: Brian Beeler Published: 2026-10-28T15:11:21Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [GB300 NVL72](), [Vera Rubin NVL72](), [NVIDIA Vera Rubin]() Tags: [architecture](), [chip](), [conversion](), [cooling](), [dc](), [distribution](), [enterprise](), [factory](), [gb300](), [gb300-nvl72](), [liquid](), [liquid-cooling](), [mgx](), [monitoring](), [nvidia](), [nvl72](), [nvl8](), [ocp](), [power](), [rack](), [rubin](), [server-rack]() ### AI overview Vertiv SmartIT bundles a factory-validated rack, 50VDC power, direct-to-chip liquid cooling, coolant distribution and monitoring for NVIDIA MGX and OCP ORV3 systems. The published 120-142kW compute-rack and 242kW N+1 cooling figures apply to the Blackwell NVL72 enterprise configuration; the text gives no power or cooling envelope for Vera Rubin NVL72. A support rack holds three 121kW CDUs, with N+1 capacity of 242kW, and each of three racks adds a 47kW rear-door heat exchanger. The system scales from one rack to a three-rack configuration. Vertiv includes installation and commissioning, while site preparation, upstream power distribution, building water and structural work remain outside the package. ### Source excerpt Vertiv announced SmartIT, a pre-engineered rack architecture that packages the rack, DC power conversion and distribution, direct-to-chip liquid cooling, in-rack coolant distribution, and monitoring into a factory-validated unit for NVIDIA MGX and OCP ORV3 systems. The supported list is NVIDIA GB300 NVL72, Vera Rubin NVL72, DGX Rubin NVL8, and HGX Rubin NVL8, and the product The post Vertiv SmartIT Ships Factory-Validated MGX and ORV3 Racks for GB300 and Rubin: 120 to 142kW per Rack, 242kW N+1 Liquid Cooling appeared first on StorageReview.com. ## Sparxity, Formerly SpiNNcloud, Deploys JUBIS Sparse-Native Rack at Jülich Supercomputing Centre With 10x Solver and 18x Energy Claims DevFeed: [Sparxity, Formerly SpiNNcloud, Deploys JUBIS Sparse-Native Rack at Jülich Supercomputing Centre With 10x Solver and 18x Energy Claims]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-10-28T14:48:03Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [Supercomputing](), [AI Infrastructure]() Tags: [accelerators](), [ai](), [ai-inference](), [ai-workloads](), [announcement](), [architecture](), [arm](), [attention](), [base](), [based](), [board](), [computing](), [deployment](), [energy](), [enterprise](), [europe](), [installation](), [machine](), [native](), [rack](), [supercomputing]() ### AI overview Jülich Supercomputing Centre is adding Sparxity's JUBIS rack, built on event-based SpiNNaker2 chips, alongside its CPU and GPU systems for sparse machine learning and HPC simulations. The system runs units only when inputs change, aiming to avoid work and data movement for inactive model parts. Sparxity claims 10x faster sparse solver execution, 18x lower energy use for event-based machine learning, and 20% shorter time to an acceptable optimization solution; the excerpt provides no baselines or workload details, and says the figures were not independently tested. JUBIS is part of JSC's three-year ESAB effort to add event-based computing to its modular architecture for AI, neuroscience simulation, and other research. The rack's board count is undisclosed. ### Source excerpt Sparxity, the Dresden company that until today was SpiNNcloud, announced the deployment of a sparse-native computing rack system at the Jülich Supercomputing Centre (JSC), an institute of Forschungszentrum Jülich and home of JUPITER, Europe's first exascale machine. The system is named JUBIS, the installation is described as multi-million-dollar, and JSC will use it to run The post Sparxity, Formerly SpiNNcloud, Deploys JUBIS Sparse-Native Rack at Jülich Supercomputing Centre With 10x Solver and 18x Energy Claims appeared first on StorageReview.com. ## Modeling Liquid-Cooled Data Centers in EnergyPlus DevFeed: [Modeling Liquid-Cooled Data Centers in EnergyPlus]() Original publisher: [Read original article]() Author: Yun-Dam Ko, Dipashreya A. Sur, Cade Napier, Carlos D. D\\'iaz-Mar\\'in, Rishee K. Jain Published: 2026-10-27T04:00:00Z Content type: article Language: en Sources: [arXiv: Distributed Computing]() Topics: [Simulation](), [clean energy data centers]() Tags: [ai](), [ai-infrastructure](), [ai-workloads](), [cooling](), [cs-ce](), [cs-dc](), [cs-lg](), [cs-sy](), [data-center](), [data-centers](), [eess-sy](), [efficiency](), [energy](), [framework](), [high-density](), [infrastructure](), [it](), [liquid](), [liquid-cooling](), [loops](), [modeling](), [models](), [native](), [open-source](), [rack]() ### AI overview The paper introduces EnergyPlus's ElectricEquipment:ITE:LiquidCooled object to model direct-to-chip cooling across chilled-water and warm-water architectures. It links facility and rack-side water loops, coolant distribution units, and chip-junction temperatures in one solver. Comparison with an independent Modelica implementation across 21 operating conditions, including two full-year runs at one-minute resolution, reports loop-temperature RMSE no greater than 0.021 K, below the 0.1 K acceptance threshold. In the case study, warm-water cooling lowered annual PUE from 1.43 to 1.25, eliminated on-site water use, and reduced total facility energy by 37.9% versus an air-cooled baseline. ### Source excerpt arXiv:2610.33148v1 Announce Type: cross Abstract: As AI workloads accelerate data center transition toward high-density racks with direct-to-chip (DTC) liquid cooling, existing simulation tools remain limited in evaluating whole-building and multi-year efficiency. To bridge this gap, we develop a native EnergyPlus object, ElectricEquipment:ITE:LiquidCooled, that models DTC liquid cooling for chilled-water and warm-water architectures as plant-loop components. It couples facility-water loops, rack-side secondary loops, coolant distribution units, and chip-junction temperatures within a single solver. Validation against an independent Modelica implementation across 21 operating conditions, including two full-year runs compared at one-minute resolution, yields loop-temperature RMSE at or below 0.021 K, well within the 0.1 K acceptance threshold. Our case study demonstrates that warm-water liquid cooling reduces annual PUE from 1.43 to 1.25, eliminates on-site water use, and cuts total facility energy by 37.9% relative to an air-cooled baseline. By embedding liquid-cooling logic into EnergyPlus, our work provides whole-building, long-term energy modeling for high-density AI infrastructure within an open-source framework. ## An Analog TV Station For Anyone DevFeed: [An Analog TV Station For Anyone]() Original publisher: [Read original article]() Author: Jenny List Published: 2026-10-25T08:00:24Z Content type: article Language: en Sources: [Hackaday]() Topics: [3D]() Tags: [3d-printed](), [analog](), [analog-tv](), [audio](), [automatic](), [broadcast](), [china](), [clock](), [collector](), [gear](), [home-entertainment-hacks](), [inexpensive](), [monitor](), [pi](), [pi-pico](), [rack](), [raspberry-pi](), [scheduler](), [sd-card](), [small](), [tv](), [tv-studio]() ### AI overview BorisDigital built a compact analog TV station in a 10-inch rack so older televisions can receive broadcasts through their intended tuner. A transmitter board sits above Pi Pico clock and audio monitoring, Raspberry Pi 5-based vectorscope and waveform monitoring, SD-card video record/play units, and an automatic scheduler. The rack reproduces functions of broadcast equipment that once cost far more. Its front panels are 3D-printed, with lettering made from different filaments. This excerpt describes the station's components and construction but does not provide performance details. ### Source excerpt If you're a collector of older TVs, you're probably facing the problem in some form of having no transmissions to tune in to. With many sets you can use a ...read more ## Serving DeepSeek-V4.1-Flash at Scale: vLLM Matches DeepInfer and Scales to GB300 NVL72 DevFeed: [Serving DeepSeek-V4.1-Flash at Scale: vLLM Matches DeepInfer and Scales to GB300 NVL72]() Original publisher: [Read original article]() Author: Inferact and the vLLM Team Published: 2026-10-23T00:00:00Z Content type: article Language: en Sources: [vLLM Blog]() Topics: [vllm](), [GB300 NVL72](), [LLM optimization](), [LLM Techniques](), [NVLink]() Tags: [agentic](), [ai-engineering](), [async](), [attention](), [b300](), [balancing](), [deepinfer](), [deepseek](), [deepseek-v4-1-flash](), [disaggregation](), [gb300](), [gb300-nvl72](), [gpu](), [large-scale-serving](), [llm-inference](), [model-serving](), [nvl72](), [open-source-ai](), [performance](), [rack](), [scale](), [tokens](), [user](), [v4-1-flash](), [vllm](), [vllm-blog](), [wide]() ### AI overview On one GB300 NVL72 rack, vLLM's P/D-disaggregated, Wide-EP serving of DeepSeek-V4.1-Flash reaches 954K tokens/s per GPU on SemiAnalysis AgentX at a P90 of 196 tokens/s per user, reported as 7.9x the authors' earlier single-node result. Decode throughput on GB200 is within 6% of DeepSeek's DeepInfer across batch sizes 8-256; the prefill curves track, though differing setups prevent a direct speedup comparison. The reported gains come from fitting prefill to DEP2, extending bounded replay to layer 20, using larger GPU pages for shared Engram tables and DeepSelect for prefill ranking, plus Wide-EP with EPLB for decode. At DEP32, EPLB substantially improves expert-load balance, while CPU overhead can erase GPU-side scaling gains. A specialized decode path reaches 826 P90 tokens/s for one user on four B300 GPUs, 44% above the standard path. Results depend on the stated hardware, configurations and workloads. ### Source excerpt With P/D disaggregation and Wide-EP on a GB300 NVL72 rack, vLLM serves DeepSeek-V4.1-Flash on SemiAnalysis AgentX at 954K tokens/s per GPU at a P90 of 196 tokens/s per user, 7.9x our aggregated single-node result. ## How Mixx Technologies Is Lighting The Way To Fully Optical Scale-up Networks DevFeed: [How Mixx Technologies Is Lighting The Way To Fully Optical Scale-up Networks]() Original publisher: [Read original article]() Author: Tobias Mann Published: 2026-10-22T20:14:12Z Content type: news Language: en Sources: [The Next Platform: In-depth coverage of high end computing]() Topics: [communications](), [NVLink](), [NVIDIA Vera Rubin](), [networking news]() Tags: [2023](), [3d](), [about](), [accelerators](), [adoption](), [cables](), [connect](), [copper](), [datacenter](), [form](), [gpu](), [lighting](), [like](), [modern](), [networks](), [nvidia](), [optical-interconnects](), [photonics](), [rack](), [scale](), [scale-up](), [systems](), [technologies](), [up](), [with]() ### AI overview Mixx Technologies is pursuing a staged path to optical scale-up networks, starting with dense fiber connections and moving toward near-packaged optics (NPO), co-packaged optics (CPO), optical circuit switches (OCS), and software to manage connections. Its SxC connector packs 64 fibers into one connector; the company says it can fit 1,000 fibers in 800 mm². Mixx plans an NPO module around 2028-29, with CPO later, after NPO adoption helps establish the approach. NPO offers a potentially refurbishable engine if it fails, while CPO brings optics closer to the chip and could avoid copper reach limits and enable higher port counts. The article says large optical fabrics will still need dense fiber attachment and high-radix OCS hardware; reconfiguring OCS drops connections, so changes need scheduling. Mixx's fully optical vision remains a longer-term goal, with product timing tied to market demand. ### Source excerpt Modern rack systems like Nvidia's NV72 are packed with miles of copper flyover cables which form the ... ## Supermicro NVIDIA Vera CPU Rack: 256 Vera CPUs, 400TB of LPDDR5X, and 22,500 Agent Sandboxes in a 208kW Liquid-Cooled 48U DevFeed: [Supermicro NVIDIA Vera CPU Rack: 256 Vera CPUs, 400TB of LPDDR5X, and 22,500 Agent Sandboxes in a 208kW Liquid-Cooled 48U]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-10-22T14:33:04Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [NVIDIA Vera](), [Nvidia](), [Vera Rubin NVL72](), [agent observability](), [NVIDIA Vera Rubin](), [NVIDIA Spectrum-X Ethernet]() Tags: [agent](), [agent-orchestration](), [agent-workflows](), [agentic](), [agentic-ai](), [ai](), [cpu](), [cpus](), [data-center](), [enterprise](), [liquid](), [lpddr5x](), [nvidia](), [nvidia-vera](), [nvidia-vera-cpu](), [nvidia-vera-cpus](), [portfolio](), [rack](), [sandboxes](), [supermicro](), [vera](), [vera-cpu]() ### AI overview Supermicro's SRS-VCPU-256-M0 is a 48U, fully liquid-cooled CPU rack designed to execute agent workflows alongside GPU systems or operate as a standalone platform. It contains 256 NVIDIA Vera CPUs (22,528 cores, 45,056 threads), with up to 400TB of LPDDR5X, 300TB/s aggregate memory bandwidth, and up to 32TB of E1.S NVMe per tray. Supermicro rates it for up to 22,500 concurrent sandboxes, but gives no workload, memory allocation, or software basis for that figure, and no measured sandbox density is reported. The rack uses BlueField-4 DPUs and Spectrum-X Ethernet; OpenShell applies runtime policy controls while Sentry monitors and can quarantine agents out of band. Its 208kW power design supports redundant AC or 800 VDC feeds, and cooling options include an in-row CDU and liquid-to-air sidecars. It is available to quote, with no ship date stated. ### Source excerpt Supermicro announced the NVIDIA Vera CPU rack today, a 48U, 100 percent liquid-cooled rack that holds 256 NVIDIA Vera CPUs, 22,528 cores, and 45,056 threads and that Supermicro rates for up to 22,500 concurrent agent sandboxes. It is the newest member of the Data Center Building Block Solutions (DCBBS) portfolio, built to run beside the The post Supermicro NVIDIA Vera CPU Rack: 256 Vera CPUs, 400TB of LPDDR5X, and 22,500 Agent Sandboxes in a 208kW Liquid-Cooled 48U appeared first on StorageReview.com. ## How Companies Are Using NVIDIA DSX to Break the Power Bottleneck DevFeed: [How Companies Are Using NVIDIA DSX to Break the Power Bottleneck]() Original publisher: [Read original article]() Author: Vladimir Troy Published: 2026-10-22T05:00:25Z Content type: news Language: en Sources: [NVIDIA]() Topics: [NVIDIA DSX](), [AI Infrastructure](), [Nvidia](), [NVIDIA Vera Rubin](), [Omniverse](), [Isaac](), [Metropolis]() Tags: [3](), [about](), [adoption](), [ai](), [ai-factory](), [ai-infrastructure](), [building](), [dsx](), [factory](), [grid](), [gtc-2026](), [infrastructure](), [nvidia](), [nvidia-dsx](), [nvidia-vera-rubin](), [omniverse](), [plan](), [platform](), [power](), [rack](), [run](), [system](), [using](), [world]() ### AI overview NVIDIA DSX aims to ease AI infrastructure power constraints by coordinating grid, facility and compute planning and operation. DSX Sim and SimReady assets support joint simulation of grid, substation, factory and cooling before construction; DSX Flex aligns demand with grid conditions and onsite generation; MaxLPS and Exchange target compute efficiency and signal sharing. Company examples include Hitachi Energy's DSX Ready battery system, Schneider Electric/AVEVA/ETAP's validated power and cooling architectures and digital twin, and Siemens Energy's grid-aware reference design. Phaidra reports about 10% more compute within the same power envelope using DSX integrations; Emerald AI's approach is modeled to unlock up to 100 GW of existing U.S. grid capacity. These are described outcomes or modeled potential, not guarantees. ### Source excerpt Companies around the world are building on the NVIDIA DSX AI Factory Platform to plan, simulate and run the grid and the AI factory as one system, from the substation to the rack. Their adoption of the platform responds to an acute problem. Power is now the biggest bottleneck in infrastructure. More than $300 billion [...] ## Dell AI Factory with NVIDIA Adds a 256-CPU Vera Rack, a Groq 3 LPU Inference Rack, and a 64GB GB10 Desktop DevFeed: [Dell AI Factory with NVIDIA Adds a 256-CPU Vera Rack, a Groq 3 LPU Inference Rack, and a 64GB GB10 Desktop]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-10-21T16:16:56Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [dell](), [AI Infrastructure](), [AI Factory](), [groq](), [NVIDIA Vera](), [NVIDIA DSX](), [NemoClaw](), [OpenShell](), [AI Architecture]() Tags: [agentic](), [ai](), [ai-factory](), [berlin](), [cpu](), [cpus](), [dell](), [dell-ai-factory](), [dell-technologies](), [desktop](), [enterprise](), [gb10](), [groq](), [gtc](), [inference](), [nvidia](), [nvidia-gtc](), [nvidia-gtc-berlin](), [nvidia-vera](), [nvidia-vera-cpus](), [rack](), [vera]() ### AI overview Dell's expansion adds infrastructure for separate stages of agentic AI: the 256-CPU PowerEdge M9882 targets orchestration and code execution while leaving inference to GPU clusters; the liquid-cooled PowerEdge XE9810 uses Groq 3 LPUs for rack-scale inference. Dell also lists an air-cooled two-socket Vera server for conventional racks, a 64GB GB10 desktop configuration for local model work, and validated deployment templates. Availability ranges from this month for the R8822 to early 2027 for several systems. Important details remain unpublished: Dell has not specified the M9882's memory or networking, or the XE9810's LPU count, throughput, and full configuration. The excerpt also describes integrated containerized data-center deployments, software validation, and layered agent governance; Dell's cost and deployment claims are vendor-reported and unaudited here. ### Source excerpt Dell Technologies expanded the Dell AI Factory with NVIDIA today at NVIDIA GTC Berlin with a 48U PowerEdge M9882 rack holding 256 NVIDIA Vera CPUs for agentic workload execution, the PowerEdge XE9810 rack-scale inference system built on Groq 3 LPUs, an air-cooled two-socket Vera server, a 64GB Dell Pro Precision with GB10, and a set The post Dell AI Factory with NVIDIA Adds a 256-CPU Vera Rack, a Groq 3 LPU Inference Rack, and a 64GB GB10 Desktop appeared first on StorageReview.com. ## HPE Private Cloud AI Adds MCP Servers, Mixed H200 and RTX PRO 6000 Clusters, and a 256-CPU NVIDIA Vera Rack DevFeed: [HPE Private Cloud AI Adds MCP Servers, Mixed H200 and RTX PRO 6000 Clusters, and a 256-CPU NVIDIA Vera Rack]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-10-21T11:30:00Z Content type: release Language: en Sources: [StorageReview.com]() Topics: [RTX PRO](), [NVIDIA Vera](), [MSP MCP](), [Infrastructure as a Service (IaaS)]() Tags: [agents](), [ai](), [ai-agents](), [ai-developers](), [ai-factory](), [cloud](), [clusters](), [cpu](), [enterprise](), [hpe](), [hpe-private-cloud-ai](), [mcp](), [mcp-servers](), [nvidia](), [nvidia-vera](), [private-cloud](), [rack](), [rtx](), [rtx-pro](), [servers](), [vera]() ### AI overview HPE Private Cloud AI adds governed APIs and MCP servers for developers, ISVs, and agents to access inference, models, knowledge services, and pipelines. Access is managed by identity, role, tenant, and project, with versioning and audit logging; API-first onboarding is planned for connected and air-gapped deployments. Mixed NVIDIA H200 and RTX PRO 6000 Blackwell Server Edition clusters and KAI quota- and priority-based scheduling are generally available. HPE is also introducing a liquid-cooled NVIDIA Vera CPU Rack with 256 CPUs for agent sandboxes, reinforcement-learning evaluations, and data processing; HPE claims 80% faster agentic CPU performance but gives no baseline or test configuration. Private Cloud AI updates are scheduled to roll out in late 2026, with self-service application and agent registration, API and MCP discovery, and credential management planned for 2027. ### Source excerpt HPE announced updates to HPE AI Factory with NVIDIA today at NVIDIA GTC Berlin, and most of the changes land in HPE Private Cloud AI. Developers, independent software vendors, and AI agents get governed access to the platform's inference, models, and pipelines through production APIs and Model Context Protocol (MCP) servers, clusters can now mix The post HPE Private Cloud AI Adds MCP Servers, Mixed H200 and RTX PRO 6000 Clusters, and a 256-CPU NVIDIA Vera Rack appeared first on StorageReview.com. ## Blog Review: Oct. 21 DevFeed: [Blog Review: Oct. 21]() Original publisher: [Read original article]() Author: Jesse Allen Published: 2026-10-21T07:01:51Z Content type: article Language: en Sources: [Semiconductor Engineering]() Topics: [Chip design](), [communications]() Tags: [ai](), [ai-development](), [amba](), [arm](), [blog](), [blog-reviews](), [business](), [cadence](), [challenges](), [development](), [digital-twins](), [edge](), [edge-ai](), [edge-ai-development](), [global](), [keysight](), [news](), [power](), [rack](), [review](), [satellite](), [semi](), [semiconductor](), [server](), [siemens](), [stanton-chase](), [synopsys](), [system-level-design](), [talent]() ### AI overview The review highlights design constraints and engineering approaches across AI systems, interconnects, digital twins, and semiconductor infrastructure. AI workloads are shifting optimization from the server to the rack, with designs starting from target workloads and aiming to improve useful AI work per watt and dollar. AMBA CXS is presented as a scalable shared transport for multi-protocol, multi-node traffic, with controlled interleaving that preserves ordering, route identity, packet reconstruction, and contiguity. Other featured topics include digital twins from chip design through fab automation, an edge AI development loop that feeds target information and real-execution evidence into iteration, and lab power emulation for satellite power systems. The excerpt also notes that regional semiconductor talent shortages have differing causes and that subsidies alone cannot supply needed expertise. Additional newsletter items cover PCIe retimer reach, resilient vias, open interfaces, chiplet connectivity, and optical engines in packages with high-power AI processors. ### Source excerpt From server to rack; AMBA CXS; digital twins; edge AI development loop; satellite power; global talent challenges. The post Blog Review: Oct. 21 appeared first on Semiconductor Engineering. ## Arm at the OCP Global Summit 2026: How Arm AGI CPU is scaling agentic AI infrastructure from CPU to rack DevFeed: [Arm at the OCP Global Summit 2026: How Arm AGI CPU is scaling agentic AI infrastructure from CPU to rack]() Original publisher: [Read original article]() Author: Arm Editorial Team Published: 2026-10-15T19:36:18Z Content type: news Language: en Sources: [Arm Newsroom]() Topics: [AI Infrastructure](), [Distributed Systems & Parallel Computing](), [Cloud Architecture](), [Arm](), [Amazon AGI Lab]() Tags: [2026](), [agentic-ai-infrastructure](), [agentic-workflow](), [ai-infrastructure](), [ai-systems](), [ai-workloads](), [amd-epyc](), [arm](), [rack](), [scaling]() ### AI overview At OCP Global Summit 2026, Arm described how Arm AGI CPU and partners are building open rack-scale infrastructure for agentic AI. The systems combine CPUs, accelerators, memory, networking and software, while shared standards aim to ease deployment across hardware designs. Arm also presented workload-specific CPU measurement methods; a Signal65 analysis reported 1.99x modeled performance per rack versus AMD EPYC 9755 across seven representative agentic AI workloads. ### Source excerpt At OCP Global 2026, Arm AGI CPU and a growing partner ecosystem show how open infrastructure can bring CPUs, accelerators, memory, software and rack design together for scale The post Arm at the OCP Global Summit 2026: How Arm AGI CPU is scaling agentic AI infrastructure from CPU to rack appeared first on Arm Newsroom. ## Supermicro OCP Accepted Portfolio Adds 48V Busbar ORv3 Racks, a 250 kW In-Rack CDU, and a 102.4 Tbps CPO Switch DevFeed: [Supermicro OCP Accepted Portfolio Adds 48V Busbar ORv3 Racks, a 250 kW In-Rack CDU, and a 102.4 Tbps CPO Switch]() Original publisher: [Read original article]() Author: Harold Fritts Published: 2026-10-14T15:26:03Z Content type: news Language: en Sources: [StorageReview.com]() Topics: [NVIDIA Spectrum-X Ethernet](), [AI Infrastructure](), [NVIDIA Vera Rubin]() Tags: [amd](), [amd-epyc](), [arm](), [asic](), [broadcom](), [co-packaged-optics](), [component](), [compute](), [coolant-distribution-unit](), [cooling](), [cpu](), [cpus](), [data](), [data-center](), [dc](), [enterprise](), [ocp](), [open-compute-project](), [rack](), [server-rack](), [switch]() ### AI overview Supermicro announced an expanded OCP Accepted portfolio of ORv3 and ORW racks, liquid-cooling equipment, compute systems, and networking products. The portfolio includes a 48V busbar, an in-rack coolant distribution unit rated for up to 250 kW, and a co-packaged optics switch rated at 102.4 Tbps. The article also describes Supermicro's management and security stack, while noting that some performance figures are vendor claims and that facility suitability requires site-specific calculations. ### Source excerpt Supermicro announced an expanded line of Open Compute Project Accepted systems, racks, and rear door heat exchangers, all delivered through its Data Center Building Block Solutions (DCBBS) program and on display at booth B4 at the OCP Global Summit in San Jose through October 15. The common element is the rack: the 21-inch ORv3 and The post Supermicro OCP Accepted Portfolio Adds 48V Busbar ORv3 Racks, a 250 kW In-Rack CDU, and a 102.4 Tbps CPO Switch appeared first on StorageReview.com. ## Nvidia-backed Upscale takes on AI networking silos with Token Fabric DevFeed: [Nvidia-backed Upscale takes on AI networking silos with Token Fabric]() Original publisher: [Read original article]() Author: Sean Michael Kerner Published: 2026-10-08T18:44:41Z Content type: news Language: en Sources: [Network World]() Topics: [Cloud Networking](), [Spectrum Ethernet](), [NVLink](), [Nvidia]() Tags: [ai](), [ai-clusters](), [artificial-intelligence](), [artificial-intelligence-networking](), [data-center](), [edition-us](), [ethernet](), [language-en](), [networking](), [nvidia](), [rack](), [storytype-news]() ### AI overview Upscale introduced Token Fabric, a hardware and software architecture that combines scale-up networking within racks with scale-out networking across data centers for AI clusters. It pairs SkyFabriX switch silicon and Nvidia Spectrum-X Ethernet systems with SkyOS and SkyCMD software for shared operations and orchestration. General availability is planned for early 2027, with early access and joint validation underway. ### Source excerpt AI clusters depend on two types of networking. Scale-up networking connects accelerators inside a rack. Scale-out networking connects racks across the data center. A single AI job spans both, which is a challenge for many organizations that tend to treat the two as separate networks. Networking startup Upscale today introduced Token Fabric to bring the two together. The architecture combines the company's own SkyFabriX scale-up silicon, scale-out systems built on Nvidia Spectrum-X Ethernet, and a common software layer for operations. General availability is planned for early 2027. Early-access and joint-validation programs are running now. The announcement builds on two earlier milestones. Upscale, based in Santa Clara, Calif., emerged from stealth in September 2025 with a $100 million seed round. In June, it raised an additional $190 million and outlined plans for SkyHammer, its custom scale-up switch chip. SkyFabriX is based on that SkyHammer architecture. Nvidia, which supplies the Spectrum-X silicon, joined the June round. Upscale has now raised about $500 million and has more than 300 employees. "If you look at compute, compute is ramping up on a rapid scale, and networking has fallen way behind," Barun Kar, CEO of Upscale, told Network World. What is inside Token Fabric Token Fabric is a combination of hardware and software for both scale-up and scale-out. SkyFabriX silicon. SkyFabriX is the scale-up switch silicon. It supports ESUN (Ethernet for Scale-Up Networking) and standard Ethernet/IP. It is designed for evolving protocols such as UALink (Ultra Accelerator Link). Switch capacity is 115.2 Tbps, with a roadmap to multiple petabits per second. Scale-up switch trays. The trays come in custom and standard rack form factors. They pool GPUs and other accelerators into large scale-up domains. Scale-out systems. Upscale builds these around Nvidia Spectrum-X silicon at speeds from 400G and 800G up to 1.6T. With SkyOS, they integrate into Nvidia and other accelera ## Gentoo Needs Your Help! Several Sponsors Have Pulled Out DevFeed: [Gentoo Needs Your Help! Several Sponsors Have Pulled Out]() Original publisher: [Read original article]() Author: Sourav Rudra Published: 2026-10-08T09:17:29Z Content type: news Language: en Sources: [It's FOSS]() Topics: [hosting](), [virtual machines](), [Infrastructure as a Service (IaaS)](), [free software]() Tags: [cloud](), [colocation](), [connectivity](), [contributions](), [dell](), [firmware](), [framework](), [funds](), [hardware](), [help](), [hetzner](), [hosting](), [hp](), [infrastructure](), [machines](), [needs](), [news](), [rack](), [servers](), [space](), [sponsors]() ### AI overview Gentoo is seeking replacement infrastructure sponsors after several longtime supporters withdrew. Its five options are donated machines for US hosting locations; colocation, with capacity outside the US most urgently needed; virtual machines with at least 2 vCPUs, 4 GB RAM and 25 GB storage; hosted physical servers with Gentoo receiving remote access; or funds for hosting and colocation. Gentoo says many services are monolithic, making Kubernetes a poor fit, while it is using cloud services more where practical. For sponsor queries, the article gives infra@gentoo.org. ### Source excerpt They will accept VMs, physical servers, rack space, donated machines, or funds for hosting. ## AI infrastructure reliability moves beyond the rack DevFeed: [AI infrastructure reliability moves beyond the rack]() Original publisher: [Read original article]() Author: Victoria Gayton Published: 2026-10-05T21:23:26Z Content type: news Language: en Sources: [SiliconANGLE]() Topics: [AI Infrastructure](), [Production Engineering](), [NVIDIA Vera Rubin](), [Vera Rubin NVL72](), [AWS Lambda]() Tags: [ai-infrastructure](), [ai-infrastructure-reliability](), [compute](), [cooling](), [coreweave](), [coreweaves-racky-rack-manager](), [cube-event-coverage](), [data-center](), [data-centers](), [dave-vellante](), [dell](), [dell-technologies](), [deployments](), [engineering](), [fully-connected-2026](), [fullyconnected](), [hardware](), [infra](), [infrastructure](), [jacob-yundt](), [john-furrier](), [liquid-cooling](), [nasdaq-crwv](), [networking](), [news](), [nvidia](), [nyse-dell](), [partnership](), [power](), [rack](), [rack-scale-infrastructure](), [reliability](), [sarat-krishnan](), [thecube](), [vera-rubin]() ### AI overview Dell and CoreWeave describe how joint engineering and manufacturing work supports reliable AI infrastructure deployments. They move complex diagnostics earlier in production, integrate CoreWeave's Racky rack manager in Dell factories, and coordinate compute, networking, power, and liquid cooling across data center facilities. The article also discusses Vera Rubin systems, where coordination may need to extend from individual racks to entire data halls or data centers. ### Source excerpt AI infrastructure reliability depends on how well compute, networking, power and cooling operate together. Turning early hardware breakthroughs into repeatable deployments requires coordination across engineering, manufacturing and operations. Dell Technologies Inc.'s partnership with CoreWeave Inc. has helped it adapt successive generations of rack-scale systems to different facilities, according to Sarat Krishnan (pictured, right), director of [...] The post AI infrastructure reliability moves beyond the rack appeared first on SiliconANGLE. ## An experimental rack of Android smartphones DevFeed: [An experimental rack of Android smartphones]() Original publisher: [Read original article]() Author: Anne Barela Published: 2026-09-29T17:58:34Z Content type: article Language: en Sources: [Adafruit Industries]() Topics: [Android](), [adb]() Tags: [adb](), [android](), [android-smartphones](), [cluster](), [computers](), [devices](), [experiment](), [experimental](), [hardware](), [hub](), [networking](), [rack](), [retired](), [scripting](), [smartphone](), [smartphones](), [telephony](), [usb]() ### AI overview The rack groups Android smartphones for experimentation, putting retired handsets to work as a cluster rather than using them one at a time. A commenter suggested driving the phones from a host with adb over USB; the excerpt describes this as a usual scripting route that can support several devices through a hub. The post excerpt does not specify the project's goal or confirm how the phones are controlled. ### Source excerpt sulfuroid posted on X about building a rack of Android smartphones. The post is short: the phones are mounted together as a cluster to experiment with, rather than used one at a time. A commenter asked what the goal is and how the phones are driven, suggesting adb over USB. That is the usual route [...] ## Network Rack for Small Office (Build Guide) DevFeed: [Network Rack for Small Office (Build Guide)]() Original publisher: [Read original article]() Author: Hayden James Published: 2026-06-22T12:58:14Z Content type: tutorial Language: en Sources: [Hayden James]() Topics: [CIDR and subnets]() Tags: [access-points](), [admin](), [airflow](), [ata](), [blog](), [build](), [community-forums](), [core](), [functional](), [future](), [general](), [growth](), [guide](), [homelab](), [isp](), [network](), [office](), [online](), [power](), [rack](), [room](), [small](), [space](), [storage](), [sysadmins]() ### AI overview The guide describes a small-office network rack built to keep core services running through brief brown-outs and outages while leaving room to expand. It uses a wheeled, open-frame 20U rack, a UniFi Dream Machine Pro router/firewall with dual-WAN, and a fan-free UniFi Pro Max 16 PoE switch rated at 180 W. CyberPower CP1500PFCLCD units are said to provide roughly two hours of backup for the router, switch, and Wi-Fi; a separate 900 VA unit powers VoIP gear. A modular 24-port Cat-6 patch panel, labeled patch leads, cable-management panels, and a 10G fiber router-to-switch uplink complete the setup. The excerpt says SFP+ ports provide expansion headroom, including for future cameras, a second ISP, or a Linux server. It cuts off while describing the battery layout, so further details are unavailable. ### Source excerpt Last week I walked into an empty 5 x 4 ft storage room. By the end of the week, that blank space held a fully functional network core designed to keep a small professional office online through power cuts, ISP hiccups, and future growth. Continue reading... ## IonQ | IonQ Participates in Quantum World Congress 2023 DevFeed: [IonQ | IonQ Participates in Quantum World Congress 2023]() Original publisher: [Read original article]() Author: IonQ Staff Published: 2026-01-15T23:19:20Z Content type: news Language: en Sources: [IonQ]() Topics: [Quantum Computing](), [Computing]() Tags: [2025](), [data-center](), [drive](), [ionq](), [launch](), [quantum-community](), [quantum-computing](), [rack](), [supercomputers]() ### AI overview IonQ participated in the Quantum World Congress 2023 and announced its roadmap for enterprise-grade quantum computing. Forte Enterprise was expected to launch in 2024 with #AQ 35, faster gate speeds, a rack-mountable footprint, and data-center compatibility. IonQ also said its Tempo system was expected in 2025 with #AQ 64, targeting problems that classical supercomputers cannot solve. ### Source excerpt Working to build the world's best quantum computers to solve the world's most complex problems [Next page]()