# GB300 NVL72

GB300 NVL72 is an NVIDIA rack-scale system built for AI reasoning.

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

## Lambda Boosts AI Factory Resiliency and Efficiency With NVIDIA Photonics

DevFeed: [Lambda Boosts AI Factory Resiliency and Efficiency With NVIDIA Photonics](<https://devfeed.tech/articles/lambda-boosts-ai-factory-resiliency-and-efficiency-with-nvidia-photonics-136623.md>)

Original publisher: [Read original article](<https://blogs.nvidia.com/blog/lambda-photonics/>)

Author: Brian Sparks

Published: 2026-11-16T16:00:51Z

Content type: article

Language: en

Sources: [NVIDIA](<https://devfeed.tech/sources/nvidia-blog.md>)

Topics: [communications](<https://devfeed.tech/topics/communications.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>), [Cloud](<https://devfeed.tech/topics/cloud.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-factory](<https://devfeed.tech/tags/ai-factory.md>), [ai-model](<https://devfeed.tech/tags/ai-model.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [cloud-provider](<https://devfeed.tech/tags/cloud-provider.md>), [co-packaged-optics](<https://devfeed.tech/tags/co-packaged-optics.md>), [cpo](<https://devfeed.tech/tags/cpo.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [lambda](<https://devfeed.tech/tags/lambda.md>), [network](<https://devfeed.tech/tags/network.md>), [networking](<https://devfeed.tech/tags/networking.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvidia-gb300-nvl72](<https://devfeed.tech/tags/nvidia-gb300-nvl72.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [nvl72-racks](<https://devfeed.tech/tags/nvl72-racks.md>), [optics](<https://devfeed.tech/tags/optics.md>), [photonics](<https://devfeed.tech/tags/photonics.md>), [resiliency](<https://devfeed.tech/tags/resiliency.md>), [supercomputing](<https://devfeed.tech/tags/supercomputing.md>), [switches](<https://devfeed.tech/tags/switches.md>)

### AI overview

Lambda is using NVIDIA Photonics co-packaged optics switches to connect GB300 NVL72 racks with more than 10,000 GPUs. In its tests, latency fell about 4% at rack scale and up to about 10% at pod scale, while throughput matched conventional pluggable optics across the configurations tested. The excerpt reports up to 5x better power efficiency and 15x higher reliability, plus 4x fewer lasers. Liquid cooling kept median optics temperature at 36°C versus 51°C for pluggable optics, with more stable signal integrity in tests. Each 144-port, 800Gb/s switch uses 18 field-replaceable external light sources; co-packaging shortens the electrical path and removes retimer chips. These are reported deployment and test findings, not evidence of outcomes beyond the described configurations.

### Source excerpt

To train an AI model, thousands of GPUs have to work together, requiring a network that's fast, efficient and resilient. Lambda is tackling the network piece of the puzzle with NVIDIA Photonics co-packaged optics (CPO) switches. The cloud provider is using the switches to connect NVIDIA GB300 NVL72 racks with more than 10,000 GPUs, delivering [...]

## 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](<https://devfeed.tech/articles/vertiv-smartit-ships-factory-validated-mgx-and-orv3-racks-for-gb300-and-rubin-120-to-142kw-per-rack-242kw-n-1-liquid-cooling-105445.md>)

Original publisher: [Read original article](<https://www.storagereview.com/news/vertiv-smartit-mgx-orv3-ai-rack-gb300-vera-rubin-nvl72-120-142kw-242kw-liquid-cooling>)

Author: Brian Beeler

Published: 2026-10-28T15:11:21Z

Content type: news

Language: en

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

Topics: [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [Vera Rubin NVL72](<https://devfeed.tech/topics/vera-rubin-nvl72.md>), [NVIDIA Vera Rubin](<https://devfeed.tech/topics/nvidia-vera-rubin.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [chip](<https://devfeed.tech/tags/chip.md>), [conversion](<https://devfeed.tech/tags/conversion.md>), [cooling](<https://devfeed.tech/tags/cooling.md>), [dc](<https://devfeed.tech/tags/dc.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [factory](<https://devfeed.tech/tags/factory.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [liquid](<https://devfeed.tech/tags/liquid.md>), [liquid-cooling](<https://devfeed.tech/tags/liquid-cooling.md>), [mgx](<https://devfeed.tech/tags/mgx.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [nvl8](<https://devfeed.tech/tags/nvl8.md>), [ocp](<https://devfeed.tech/tags/ocp.md>), [power](<https://devfeed.tech/tags/power.md>), [rack](<https://devfeed.tech/tags/rack.md>), [rubin](<https://devfeed.tech/tags/rubin.md>), [server-rack](<https://devfeed.tech/tags/server-rack.md>)

### 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.

## 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](<https://devfeed.tech/articles/serving-deepseek-v4-1-flash-at-scale-vllm-matches-deepinfer-and-scales-to-gb300-nvl72-98978.md>)

Original publisher: [Read original article](<https://vllm.ai/blog/2026-10-23-deepseek-v41-flash-pd-wide-ep>)

Author: Inferact and the vLLM Team

Published: 2026-10-23T00:00:00Z

Content type: article

Language: en

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

Topics: [vllm](<https://devfeed.tech/topics/vllm.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [async](<https://devfeed.tech/tags/async.md>), [attention](<https://devfeed.tech/tags/attention.md>), [b300](<https://devfeed.tech/tags/b300.md>), [balancing](<https://devfeed.tech/tags/balancing.md>), [deepinfer](<https://devfeed.tech/tags/deepinfer.md>), [deepseek](<https://devfeed.tech/tags/deepseek.md>), [deepseek-v4-1-flash](<https://devfeed.tech/tags/deepseek-v4-1-flash.md>), [disaggregation](<https://devfeed.tech/tags/disaggregation.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [large-scale-serving](<https://devfeed.tech/tags/large-scale-serving.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [model-serving](<https://devfeed.tech/tags/model-serving.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [open-source-ai](<https://devfeed.tech/tags/open-source-ai.md>), [performance](<https://devfeed.tech/tags/performance.md>), [rack](<https://devfeed.tech/tags/rack.md>), [scale](<https://devfeed.tech/tags/scale.md>), [tokens](<https://devfeed.tech/tags/tokens.md>), [user](<https://devfeed.tech/tags/user.md>), [v4-1-flash](<https://devfeed.tech/tags/v4-1-flash.md>), [vllm](<https://devfeed.tech/tags/vllm.md>), [vllm-blog](<https://devfeed.tech/tags/vllm-blog.md>), [wide](<https://devfeed.tech/tags/wide.md>)

### 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.

## CoreWeave Leads the Way with First NVIDIA GB300 NVL72 Deployment

DevFeed: [CoreWeave Leads the Way with First NVIDIA GB300 NVL72 Deployment](<https://devfeed.tech/articles/coreweave-leads-the-way-with-first-nvidia-gb300-nvl72-deployment-96115.md>)

Original publisher: [Read original article](<https://wf.coreweave.com/blog/coreweave-leads-the-way-with-first-nvidia-gb300-nvl72-deployment>)

Author: CoreWeave

Published: 2026-10-22T16:27:39Z

Content type: release

Language: en

Sources: [Coreweave Blog Posts](<https://devfeed.tech/sources/coreweave-blog-posts.md>)

Topics: [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>), [DOCA](<https://devfeed.tech/topics/doca.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-applications](<https://devfeed.tech/tags/ai-applications.md>), [ai-cloud](<https://devfeed.tech/tags/ai-cloud.md>), [ai-inference](<https://devfeed.tech/tags/ai-inference.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [content-type](<https://devfeed.tech/tags/content-type.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [first](<https://devfeed.tech/tags/first.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [inference](<https://devfeed.tech/tags/inference.md>), [inference-performance](<https://devfeed.tech/tags/inference-performance.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [launches](<https://devfeed.tech/tags/launches.md>), [next](<https://devfeed.tech/tags/next.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [performance](<https://devfeed.tech/tags/performance.md>), [scale](<https://devfeed.tech/tags/scale.md>)

### AI overview

CoreWeave says it has brought up the first NVIDIA GB300 NVL72 deployment on its cloud, housed in Dell's rack-scale system, with broader availability still forthcoming. The liquid-cooled rack combines 72 Blackwell Ultra GPUs, 36 Grace CPUs and 18 BlueField-3 DPUs, with up to 21 TB of GPU memory, 130 TB/s aggregate NVLink bandwidth and 800 Gb/s networking per GPU. CoreWeave cites up to 10x user responsiveness, 5x throughput per watt and 50x reasoning inference output versus Hopper; these are company-reported comparisons. Its cloud stack includes Kubernetes and Slurm offerings, rack-level monitoring and streamed hardware health events through Weights & Biases to help correlate infrastructure issues with training runs. Dell Technologies, Switch and Vertiv collaborated on the deployment.

### Source excerpt

CoreWeave launches the NVIDIA GB300 NVL72, delivering record-breaking AI inference performance and setting a new standard for next-gen, large-scale AI infrastructure.

## Get Started with Kimi K3 on CoreWeave Dedicated Inference

DevFeed: [Get Started with Kimi K3 on CoreWeave Dedicated Inference](<https://devfeed.tech/articles/get-started-with-kimi-k3-on-coreweave-dedicated-inference-96129.md>)

Original publisher: [Read original article](<https://wf.coreweave.com/blog/get-started-with-kimi-k3-dedicated-inference>)

Author: CoreWeave

Published: 2026-10-22T16:27:39Z

Content type: tutorial

Language: en

Sources: [Coreweave Blog Posts](<https://devfeed.tech/sources/coreweave-blog-posts.md>)

Topics: [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [model-serving](<https://devfeed.tech/topics/model-serving.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [vllm](<https://devfeed.tech/topics/vllm.md>), [Dynamo](<https://devfeed.tech/topics/dynamo.md>), [GB200](<https://devfeed.tech/topics/gb200.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [AI Factory](<https://devfeed.tech/topics/ai-factory.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-compute](<https://devfeed.tech/tags/ai-compute.md>), [all](<https://devfeed.tech/tags/all.md>), [content-type](<https://devfeed.tech/tags/content-type.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [dedicated-inference](<https://devfeed.tech/tags/dedicated-inference.md>), [deploy](<https://devfeed.tech/tags/deploy.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [guide](<https://devfeed.tech/tags/guide.md>), [inference](<https://devfeed.tech/tags/inference.md>), [k3](<https://devfeed.tech/tags/k3.md>), [kimi](<https://devfeed.tech/tags/kimi.md>), [kimi-k3](<https://devfeed.tech/tags/kimi-k3.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [this](<https://devfeed.tech/tags/this.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

CoreWeave Dedicated Inference deploys customer-provided Kimi K3 weights on NVIDIA GB300 NVL72, with a tenant-isolated gateway, autoscaling, and vLLM serving on NVIDIA Dynamo. Setup involves staging Kimi K3 and a DSpark draft model in CoreWeave AI Object Storage, granting access, creating a gateway and deployment with a validated engine configuration, verifying the endpoint, and then cleaning up. Customers must supply a valid Kimi K3 license and weights; billing is per GPU-hour, with no ingress or egress fees. In CoreWeave's 16-GPU tests of multi-turn agentic workloads with a sustained 95% cache hit rate, GB300 scaled to concurrency 32 at 198,000 output tokens per minute per GPU. At concurrency 32, that was 209% higher throughput than GB200, with 133% higher p50 decode throughput; GB200 peaked at concurrency 16 and became unusable at 32 for latency-sensitive interactive workloads. The stack also supports KV-cache-aware routing, speculative decoding, and prefill/decode disaggregation.

### Source excerpt

Follow this step-by-step guide to deploy Kimi K3 on CoreWeave Dedicated Inference on NVIDIA GB300 NVL72.

## CoreWeave's Innovation Velocity Drives MLPerf 6.0 Leadership

DevFeed: [CoreWeave's Innovation Velocity Drives MLPerf 6.0 Leadership](<https://devfeed.tech/articles/coreweave-s-innovation-velocity-drives-mlperf-6-0-leadership-96121.md>)

Original publisher: [Read original article](<https://wf.coreweave.com/blog/coreweaves-innovation-velocity-drives-mlperf-6-0-leadership>)

Author: CoreWeave

Published: 2026-10-22T16:27:39Z

Content type: news

Language: en

Sources: [Coreweave Blog Posts](<https://devfeed.tech/sources/coreweave-blog-posts.md>)

Topics: [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [GB200](<https://devfeed.tech/topics/gb200.md>), [Serverless](<https://devfeed.tech/topics/serverless.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-applications](<https://devfeed.tech/tags/ai-applications.md>), [ai-inference](<https://devfeed.tech/tags/ai-inference.md>), [ai-native](<https://devfeed.tech/tags/ai-native.md>), [ai-reasoning](<https://devfeed.tech/tags/ai-reasoning.md>), [applications](<https://devfeed.tech/tags/applications.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [content-type](<https://devfeed.tech/tags/content-type.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [deepseek](<https://devfeed.tech/tags/deepseek.md>), [deepseek-r1](<https://devfeed.tech/tags/deepseek-r1.md>), [gb200](<https://devfeed.tech/tags/gb200.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [leadership](<https://devfeed.tech/tags/leadership.md>), [mlperf](<https://devfeed.tech/tags/mlperf.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [offline](<https://devfeed.tech/tags/offline.md>), [performance](<https://devfeed.tech/tags/performance.md>), [r1](<https://devfeed.tech/tags/r1.md>)

### AI overview

In the excerpt, CoreWeave reports leading MLPerf 6.0 DeepSeek R1 throughput: highest among submitters on GB300 NVL72 in offline mode, and leading on GB200 NVL72 in both offline and server modes. Compared with MLPerf 5.1, GB300 DeepSeek R1 throughput improved 1.6x offline and 2x server over six months. The excerpt also reports GB300 gains over GB200 of 14% in server mode and 34% offline for DeepSeek R1, and up to 28% higher throughput for GPT-OSS-120B. CoreWeave describes three deployment paths: W&B Inference serverless with autoscaling and pay-per-token pricing; Dedicated Inference, in preview, where customers choose weights, GPUs, and runtimes while CoreWeave handles operations; and self-managed Inference on CKS for greater control over models and orchestration.

### Source excerpt

CoreWeave led MLPerf performance for NVIDIA GB200 NVL72 and NVIDIA GB300 NVL72 with DeepSeek R1 in offline mode.

## CoreWeave's NVIDIA GB300 NVL72 Production-Ready Instances for Enterprise AI, Featuring NVIDIA Blackwell Ultra GPUs, Deliver More Than 6x Performance Gain on DeepSeek R1

DevFeed: [CoreWeave's NVIDIA GB300 NVL72 Production-Ready Instances for Enterprise AI, Featuring NVIDIA Blackwell Ultra GPUs, Deliver More Than 6x Performance Gain on DeepSeek R1](<https://devfeed.tech/articles/coreweave-s-nvidia-gb300-nvl72-production-ready-instances-for-enterprise-ai-featuring-nvidia-blackwell-ultra-gpus-deliver-more-than-6x-performance-gain-on-deepseek-r1-96122.md>)

Original publisher: [Read original article](<https://wf.coreweave.com/blog/coreweaves-nvidia-gb300-nvl72-production-ready-instances-for-enterprise-ai-featuring-nvidia-blackwell-ultra-gpus-deliver-more-than-6x-performance-gain-on-deepseek-r1>)

Author: CoreWeave

Published: 2026-10-22T16:27:39Z

Content type: article

Language: en

Sources: [Coreweave Blog Posts](<https://devfeed.tech/sources/coreweave-blog-posts.md>)

Topics: [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [NVIDIA Blackwell](<https://devfeed.tech/topics/nvidia-blackwell.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [Multi-GPU](<https://devfeed.tech/topics/multi-gpu.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [blackwell-ultra](<https://devfeed.tech/tags/blackwell-ultra.md>), [content-type](<https://devfeed.tech/tags/content-type.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [deepseek](<https://devfeed.tech/tags/deepseek.md>), [deepseek-r1](<https://devfeed.tech/tags/deepseek-r1.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [enterprise-ai](<https://devfeed.tech/tags/enterprise-ai.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [inference](<https://devfeed.tech/tags/inference.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [overhead](<https://devfeed.tech/tags/overhead.md>), [performance](<https://devfeed.tech/tags/performance.md>), [real-time](<https://devfeed.tech/tags/real-time.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [scale](<https://devfeed.tech/tags/scale.md>), [throughput](<https://devfeed.tech/tags/throughput.md>)

### AI overview

CoreWeave reports over 6x higher raw DeepSeek R1 inference throughput per GPU on NVIDIA GB300 NVL72 than on an NVIDIA H100 system. The benchmark compared four GB300 GPUs using 4-way Tensor Parallelism (TP4) with 16 H100 GPUs requiring TP16; the article attributes the gain partly to GB300 memory and interconnect bandwidth reducing communication overhead. It describes 37 TB of system memory, fifth-generation NVLink bandwidth of 130 TB/s across 72 Blackwell Ultra GPUs, and Quantum-X800 InfiniBand. CoreWeave also says its topology-aware scheduler keeps workloads within an NVL72 domain when possible, alongside automated rack provisioning and Grafana monitoring. The reported result is a vendor benchmark; the excerpt does not provide broader workload comparisons or independent validation.

### Source excerpt

CoreWeave's GB300 NVL72 cut DeepSeek R1 inference overhead with TP4, delivering 6.5x throughput gains and enabling real-time reasoning at enterprise scale.

## NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global Infrastructure

DevFeed: [NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global Infrastructure](<https://devfeed.tech/articles/nvidia-confidential-computing-expands-enabling-frontier-model-builders-to-securely-scale-across-global-infrastructure-92597.md>)

Original publisher: [Read original article](<https://blogs.nvidia.com/blog/confidential-computing-model-builders-infrastructure/>)

Author: John Fanelli

Published: 2026-10-21T10:30:02Z

Content type: release

Language: en

Sources: [NVIDIA](<https://devfeed.tech/sources/nvidia-blog.md>)

Topics: [Confidential Computing](<https://devfeed.tech/topics/confidential-computing.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>), [cloud-infrastructure](<https://devfeed.tech/topics/cloud-infrastructure.md>), [Frontier AI](<https://devfeed.tech/topics/frontier-ai.md>), [NVIDIA DSX](<https://devfeed.tech/topics/nvidia-dsx.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [Azure](<https://devfeed.tech/topics/azure.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [builds](<https://devfeed.tech/tags/builds.md>), [computing](<https://devfeed.tech/tags/computing.md>), [confidential-computing](<https://devfeed.tech/tags/confidential-computing.md>), [data](<https://devfeed.tech/tags/data.md>), [deploy](<https://devfeed.tech/tags/deploy.md>), [frontier-model](<https://devfeed.tech/tags/frontier-model.md>), [global](<https://devfeed.tech/tags/global.md>), [gtc-2026](<https://devfeed.tech/tags/gtc-2026.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [layer](<https://devfeed.tech/tags/layer.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvidia-blackwell](<https://devfeed.tech/tags/nvidia-blackwell.md>), [nvidia-confidential-computing](<https://devfeed.tech/tags/nvidia-confidential-computing.md>), [nvidia-openshell](<https://devfeed.tech/tags/nvidia-openshell.md>), [nvidia-openshell-reference-architecture](<https://devfeed.tech/tags/nvidia-openshell-reference-architecture.md>), [nvidia-vera-rubin](<https://devfeed.tech/tags/nvidia-vera-rubin.md>), [openshell](<https://devfeed.tech/tags/openshell.md>), [scale](<https://devfeed.tech/tags/scale.md>), [scale-across](<https://devfeed.tech/tags/scale-across.md>)

### AI overview

NVIDIA Confidential Bare Metal lets frontier model builders run sensitive workloads on dedicated third-party infrastructure while retaining control of their software stack, models, and data. New controls protect data in transit, measure platform integrity, and enforce hardware-level access restrictions. It joins Confidential Containers for Kubernetes and Confidential Virtual Machines for virtualized environments, allowing organizations to choose a mode based on trust boundaries and operating requirements. NVIDIA says its hardware-rooted isolation and cryptographic attestation protect models and data during execution. Support is expanding to Grace Blackwell NVL72 and x86-based HGX Rubin NVL8 and DGX Rubin NVL8 systems. Google Cloud plans A4X Max instances based on GB300 NVL72; Cohere plans to integrate the service into Model Vault. The excerpt lists additional supporting cloud providers and systems partners, and says NVIDIA DSX OS will support all three modes.

### Source excerpt

NVIDIA builds tools to enable safe agents across the stack. The NVIDIA OpenShell Reference Architecture, announced today, is part of NVIDIA's growing portfolio of technologies that enables organizations to safely deploy agents at scale. NVIDIA Confidential Computing is the infrastructure layer that protects the models and data those agents depend on while running. Today, NVIDIA [...]

## Meta Prometheus: Inside the Gigawatt Supercluster Scaling to 1 Million GPUs on GB200 and GB300 NVL72

DevFeed: [Meta Prometheus: Inside the Gigawatt Supercluster Scaling to 1 Million GPUs on GB200 and GB300 NVL72](<https://devfeed.tech/articles/meta-prometheus-inside-the-gigawatt-supercluster-scaling-to-1-million-gpus-on-gb200-and-gb300-nvl72-90236.md>)

Original publisher: [Read original article](<https://www.storagereview.com/news/meta-prometheus-gigawatt-supercluster-1-million-gpus-gb200-gb300-nvl72>)

Author: Brian Beeler

Published: 2026-10-19T16:17:27Z

Content type: news

Language: en

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

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [GB200](<https://devfeed.tech/topics/gb200.md>), [Cloud Architecture](<https://devfeed.tech/topics/cloud-architecture.md>), [Vera Rubin NVL72](<https://devfeed.tech/topics/vera-rubin-nvl72.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [400g](<https://devfeed.tech/tags/400g.md>), [51-2t](<https://devfeed.tech/tags/51-2t.md>), [accelerators](<https://devfeed.tech/tags/accelerators.md>), [ai](<https://devfeed.tech/tags/ai.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [asic](<https://devfeed.tech/tags/asic.md>), [can](<https://devfeed.tech/tags/can.md>), [complete](<https://devfeed.tech/tags/complete.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [gb200](<https://devfeed.tech/tags/gb200.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [inside](<https://devfeed.tech/tags/inside.md>), [meta](<https://devfeed.tech/tags/meta.md>), [muse-wonder](<https://devfeed.tech/tags/muse-wonder.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [published](<https://devfeed.tech/tags/published.md>), [scaling](<https://devfeed.tech/tags/scaling.md>)

### AI overview

Meta's Prometheus AI supercluster in Ohio is expanding toward roughly one million GPUs. The article details its GB200 and GB300 systems, multi-building Ethernet network, GPU spares and checkpoint recovery, reliability improvements, firmware automation, power management, and cooling infrastructure. Meta reports that interruptions fell by 99 percent from their mid-2025 peak after a year of work across the hardware and software stack.

### Source excerpt

Meta published a detailed look inside Prometheus today, the gigawatt-scale AI supercluster it is building in New Albany, Ohio. The first phase is complete and has trained the Muse family of foundation models, including the Muse Wonder models, and Meta says the site is being scaled toward roughly 1 million GPUs. Two years ago, the The post Meta Prometheus: Inside the Gigawatt Supercluster Scaling to 1 Million GPUs on GB200 and GB300 NVL72 appeared first on StorageReview.com.

## Scale Agentic Coding Workflows with Meta's Muse Spark Open on NVIDIA GB300 NVL72

DevFeed: [Scale Agentic Coding Workflows with Meta's Muse Spark Open on NVIDIA GB300 NVL72](<https://devfeed.tech/articles/scale-agentic-coding-workflows-with-meta-s-muse-spark-open-on-nvidia-gb300-nvl72-88210.md>)

Original publisher: [Read original article](<https://developer.nvidia.com/blog/scale-agentic-coding-workflows-with-metas-muse-spark-open-on-nvidia-gb300-nvl72/>)

Author: Tanya Lenz

Published: 2026-10-15T18:00:01Z

Content type: article

Language: en

Sources: [NVIDIA Developer](<https://devfeed.tech/sources/nvidia-developer.md>), [NVIDIA Technical Blog](<https://devfeed.tech/sources/nvidia-technical-blog.md>)

Topics: [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [gpt-6-astra](<https://devfeed.tech/topics/gpt-6-astra.md>), [Mixture of Experts (MoE)](<https://devfeed.tech/topics/mixture-of-experts-moe.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>), [DGX Spark](<https://devfeed.tech/topics/dgx-spark.md>)

Tags: [accelerated-computing](<https://devfeed.tech/tags/accelerated-computing.md>), [agentic-ai-generative-ai](<https://devfeed.tech/tags/agentic-ai-generative-ai.md>), [agentic-coding](<https://devfeed.tech/tags/agentic-coding.md>), [context-window](<https://devfeed.tech/tags/context-window.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [featured](<https://devfeed.tech/tags/featured.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [megatron](<https://devfeed.tech/tags/megatron.md>), [mixture-of-experts-moe](<https://devfeed.tech/tags/mixture-of-experts-moe.md>), [nvfp4](<https://devfeed.tech/tags/nvfp4.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [pytorch](<https://devfeed.tech/tags/pytorch.md>), [scale](<https://devfeed.tech/tags/scale.md>), [top-stories](<https://devfeed.tech/tags/top-stories.md>), [training-ai-models](<https://devfeed.tech/tags/training-ai-models.md>)

### AI overview

Meta introduces Muse Spark Open, an 816-billion-parameter mixture-of-experts model with 42 billion active parameters and a one-million-token context window, trained on NVIDIA GB300 NVL72 for agentic coding. The article describes its long-running coding tasks, tool use, multimodal input, and deployment and fine-tuning options for developers.

### Source excerpt

Meta launches Muse Spark Open, a coding-optimized 816B mixture-of-experts (MoE) model with 42B active parameters and a 1M token context window. Trained on...

## Nvidia tackles AI factory scaling bottlenecks with open rack designs and networking advances

DevFeed: [Nvidia tackles AI factory scaling bottlenecks with open rack designs and networking advances](<https://devfeed.tech/articles/nvidia-tackles-ai-factory-scaling-bottlenecks-with-open-rack-designs-and-networking-advances-85293.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/12/nvidia-tackles-ai-factory-scaling-bottlenecks-with-open-rack-designs-and-networking-advances/>)

Author: Kyt Dotson

Published: 2026-10-12T15:00:53Z

Content type: news

Language: en

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

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [NVIDIA Vera Rubin](<https://devfeed.tech/topics/nvidia-vera-rubin.md>), [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [NVIDIA Spectrum-X Ethernet](<https://devfeed.tech/topics/nvidia-spectrum-x-ethernet.md>), [Spectrum Ethernet](<https://devfeed.tech/topics/spectrum-ethernet.md>), [axe-core](<https://devfeed.tech/topics/axe-core.md>), [NVIDIA Research](<https://devfeed.tech/topics/nvidia-research.md>), [cornelis networks](<https://devfeed.tech/topics/cornelis-networks.md>)

Tags: [accelerated-computing](<https://devfeed.tech/tags/accelerated-computing.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-ai-infrastructure](<https://devfeed.tech/tags/agentic-ai-infrastructure.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-factories](<https://devfeed.tech/tags/ai-factories.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-systems](<https://devfeed.tech/tags/ai-systems.md>), [ai-training-and-inference](<https://devfeed.tech/tags/ai-training-and-inference.md>), [artificial-inteligence](<https://devfeed.tech/tags/artificial-inteligence.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [aws](<https://devfeed.tech/tags/aws.md>), [aws-marketplace](<https://devfeed.tech/tags/aws-marketplace.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [infra](<https://devfeed.tech/tags/infra.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [joe-delaere](<https://devfeed.tech/tags/joe-delaere.md>), [nasdaq-csco](<https://devfeed.tech/tags/nasdaq-csco.md>), [nasdaq-goog](<https://devfeed.tech/tags/nasdaq-goog.md>), [nasdaq-meta](<https://devfeed.tech/tags/nasdaq-meta.md>), [nasdaq-nvda](<https://devfeed.tech/tags/nasdaq-nvda.md>), [news](<https://devfeed.tech/tags/news.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvidia-vera-cpu](<https://devfeed.tech/tags/nvidia-vera-cpu.md>), [nvidia-vera-rubin](<https://devfeed.tech/tags/nvidia-vera-rubin.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>), [open-source](<https://devfeed.tech/tags/open-source.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>), [top-story-3](<https://devfeed.tech/tags/top-story-3.md>), [vera-rubin](<https://devfeed.tech/tags/vera-rubin.md>)

### AI overview

Nvidia announced open Vera Rubin rack and compute tray designs through the Open Compute Project, alongside updates to its MGX architecture for AI factory systems. The announcements also include networking partnerships with Meta, Google and Cisco, covering Spectrum-6 Ethernet, Falcon Transport support and a proposed optical module format. The article describes how these infrastructure changes are intended to support large-scale AI training, inference and agent workloads.

### Source excerpt

Nvidia Corp. today announced new partnerships and contributions to the open-source community at the Open Compute Project Global Summit in San Jose, aimed at laying the foundation for the future of agentic artificial intelligence infrastructure. At the core of these announcements, the chip and hardware giant contributed key Vera Rubin rack and compute tray designs [...] The post Nvidia tackles AI factory scaling bottlenecks with open rack designs and networking advances appeared first on SiliconANGLE.

## Gigabyte W775-V10-L01 Hands-on Bringing NVIDIA GB300 Deskside

DevFeed: [Gigabyte W775-V10-L01 Hands-on Bringing NVIDIA GB300 Deskside](<https://devfeed.tech/articles/gigabyte-w775-v10-l01-hands-on-bringing-nvidia-gb300-deskside-65692.md>)

Original publisher: [Read original article](<https://www.servethehome.com/gigabyte-w775-v10-l01-hands-on-bringing-nvidia-gb300-deskside/>)

Author: Patrick Kennedy

Published: 2026-10-06T17:00:41Z

Content type: comparison

Language: en

Sources: [ServeTheHome](<https://devfeed.tech/sources/servethehome.md>)

Topics: [DGX Station](<https://devfeed.tech/topics/dgx-station.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [gpudirect rdma](<https://devfeed.tech/topics/gpudirect-rdma.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>)

Tags: [10gbe](<https://devfeed.tech/tags/10gbe.md>), [ai](<https://devfeed.tech/tags/ai.md>), [blackwell](<https://devfeed.tech/tags/blackwell.md>), [connectx](<https://devfeed.tech/tags/connectx.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [dgx-station](<https://devfeed.tech/tags/dgx-station.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gigabyte](<https://devfeed.tech/tags/gigabyte.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [grace-cpu](<https://devfeed.tech/tags/grace-cpu.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [workstation](<https://devfeed.tech/tags/workstation.md>)

### AI overview

The article reviews Gigabyte's W775-V10-L01, a desk-side NVIDIA DGX Station built around the GB300 platform. It describes the system's liquid cooling, power requirements, networking, and component layout, and reports testing Qwen3.6 inference performance, ConnectX-8 RDMA networking, and the Grace CPU-to-GPU C2C link.

### Source excerpt

We test the Gigabyte W775-V10-L01, including getting over 2.2B tokens/ day on the Blackwell Ultra GPU, 800Gbps from the ConnectX-8 SuperNIC, and testing the NVIDIA Grace CPU with the C2C link between it and the GPU The post Gigabyte W775-V10-L01 Hands-on Bringing NVIDIA GB300 Deskside appeared first on ServeTheHome.

## CoreWeave brings Nvidia Vera Rubin, AI tools to its cloud services

DevFeed: [CoreWeave brings Nvidia Vera Rubin, AI tools to its cloud services](<https://devfeed.tech/articles/coreweave-brings-nvidia-vera-rubin-ai-tools-to-its-cloud-services-63210.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4229739/coreweave-brings-nvidia-vera-rubin-ai-tools-to-its-cloud-services.html>)

Author: Andy Patrizio

Published: 2026-10-01T17:10:46Z

Content type: news

Language: en

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

Topics: [NVIDIA Vera Rubin](<https://devfeed.tech/topics/nvidia-vera-rubin.md>), [Vera Rubin NVL72](<https://devfeed.tech/topics/vera-rubin-nvl72.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>), [GB200](<https://devfeed.tech/topics/gb200.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [NVIDIA Vera](<https://devfeed.tech/topics/nvidia-vera.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-tools](<https://devfeed.tech/tags/ai-tools.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-cloud-computing-cpus-and-processors](<https://devfeed.tech/tags/artificial-intelligence-cloud-computing-cpus-and-processors.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [cloud-computing](<https://devfeed.tech/tags/cloud-computing.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [cpus-and-processors](<https://devfeed.tech/tags/cpus-and-processors.md>), [devin](<https://devfeed.tech/tags/devin.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvidia-vera-rubin](<https://devfeed.tech/tags/nvidia-vera-rubin.md>), [vera-rubin](<https://devfeed.tech/tags/vera-rubin.md>)

### AI overview

CoreWeave announced that Nvidia's Vera Rubin NVL72 rack-scale AI platform is available in its cloud, with Cognition as its first cluster customer. Cognition reported throughput gains over a GB200 NVL72 baseline for inference and reinforcement-learning workloads. CoreWeave also introduced Forge, an AI development platform for model training, inference, evaluation, and agent development, and added support for Vera CPUs alongside its GPU infrastructure.

### Source excerpt

CoreWeave has made a series of announcements starting with Nvidia's new Vera Rubin NVL72 rack-scale AI platform available in its cloud. At its CoreWeave's Fully Connected conference in San Francisco, the company announced Cognition, the developer of the Devin AI software-engineering agent, is the first customer for a Vera Rubin cluster. Cognition brought up the Vera Rubin cluster in early September and ran what CoreWeave described as the first customer-executed inference benchmark on the new platform. In announcing the deal, Cognition said it measured up to a 4.8X increase in total token throughput for SWE-2 inference workloads on Vera Rubin NVL72 compared with an Nvidia GB200 NVL72 baseline. The company also reported a 3.8X gain in output-token throughput for reinforcement-learning workloads. Chen Goldberg, executive vice president of product and engineering at CoreWeave, said the company's engineering investment was intended to allow customers to bring new rack-scale systems into production within days rather than rebuilding infrastructure processes for each GPU generation. "When it comes to agentic tasks, long contexts, repeated model calls and thousands of concurrent tasks put pressure on the entire platform," Goldberg said in a statement. "Our job is to make compute, networking and software work as a single system." He said the throughput gains could translate into more concurrent Devin sessions per GPU, faster research cycles and a lower per-session cost without reducing generation speed. CoreWeave has already deployed Nvidia GB200 and GB300 NVL72 systems, including Dell PowerRack configurations developed with Dell Technologies. The cloud provider said its existing customers will be able to operate Vera Rubin capacity using the same tooling and operating approach used for their GB200 and GB300 fleets. In addition to the Vera Rubin support, CoreWeave is adding support for the Vera CPU, positioning the processor as a way to address the growing amount of CPU-inte

## Securing AI Agents at Scale with NVIDIA

DevFeed: [Securing AI Agents at Scale with NVIDIA](<https://devfeed.tech/articles/securing-ai-agents-at-scale-with-nvidia-80443.md>)

Original publisher: [Read original article](<https://www.paloaltonetworks.com/blog/2026/09/securing-ai-agents-at-scale-with-nvidia/>)

Author: Anand Oswal

Published: 2026-09-28T09:00:50Z

Content type: article

Language: en

Sources: [Palo Alto Networks](<https://devfeed.tech/sources/palo-alto-networks.md>)

Topics: [AI in cybersecurity](<https://devfeed.tech/topics/ai-in-cybersecurity.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [NVIDIA Vera Rubin](<https://devfeed.tech/topics/nvidia-vera-rubin.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [DOCA](<https://devfeed.tech/topics/doca.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>), [Local AI](<https://devfeed.tech/topics/local-ai.md>)

Tags: [agent-security](<https://devfeed.tech/tags/agent-security.md>), [announcement](<https://devfeed.tech/tags/announcement.md>), [bluefield-dpu](<https://devfeed.tech/tags/bluefield-dpu.md>), [doca](<https://devfeed.tech/tags/doca.md>), [partners](<https://devfeed.tech/tags/partners.md>), [products-and-services](<https://devfeed.tech/tags/products-and-services.md>), [scale](<https://devfeed.tech/tags/scale.md>), [securing-ai](<https://devfeed.tech/tags/securing-ai.md>), [securing-ai-agents](<https://devfeed.tech/tags/securing-ai-agents.md>)

### AI overview

Palo Alto Networks describes its expanded collaboration with NVIDIA to secure enterprise AI agents. The plans cover AI governance through Prisma AIRS AI Gateway, runtime protection on NVIDIA BlueField DPUs, and agent identity controls, including least-privilege access and monitoring.

### Source excerpt

AI agents are crossing an important threshold. They are moving from assistants that respond to us to autonomous systems that reason, plan, and act on our behalf. They can write and execute ... The post Securing AI Agents at Scale with NVIDIA appeared first on Palo Alto Networks Blog.

## Giga Computing Opens GAIFA, a 700 kW Liquid-Cooled AI Factory Test Bed With Two GB300 NVL72 Racks

DevFeed: [Giga Computing Opens GAIFA, a 700 kW Liquid-Cooled AI Factory Test Bed With Two GB300 NVL72 Racks](<https://devfeed.tech/articles/giga-computing-opens-gaifa-a-700-kw-liquid-cooled-ai-factory-test-bed-with-two-gb300-nvl72-racks-58777.md>)

Original publisher: [Read original article](<https://www.storagereview.com/news/giga-computing-opens-gaifa-a-700-kw-liquid-cooled-ai-factory-test-bed-with-two-gb300-nvl72-racks>)

Author: Harold Fritts

Published: 2026-09-23T16:34:09Z

Content type: news

Language: en

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

Topics: [AI Factory](<https://devfeed.tech/topics/ai-factory.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [Computing](<https://devfeed.tech/topics/computing.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-factory](<https://devfeed.tech/tags/ai-factory.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [amd-epyc](<https://devfeed.tech/tags/amd-epyc.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [computing](<https://devfeed.tech/tags/computing.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [liquid-cooling](<https://devfeed.tech/tags/liquid-cooling.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [power](<https://devfeed.tech/tags/power.md>), [rack-scale](<https://devfeed.tech/tags/rack-scale.md>)

### AI overview

Giga Computing has opened GAIFA, a six-rack, direct-liquid-cooled AI computing center in Taiwan. The facility is designed to validate rack-scale infrastructure, including compute, power, thermal management, networking, and system compatibility, with customers before deployment. Its first phase includes two NVIDIA GB300 NVL72 racks and a planned 700 kW power envelope.

### Source excerpt

Giga Computing has opened GAIFA (GIGABYTE AI Factory Accelerator), its first proprietary AI computing center, in New Taipei City, Taiwan: a six-rack, direct-liquid-cooled facility built to validate rack-scale AI infrastructure with customers before it ships. The first phase carries a planned 700 kW power envelope, with more capacity planned in later phases as operational needs The post Giga Computing Opens GAIFA, a 700 kW Liquid-Cooled AI Factory Test Bed With Two GB300 NVL72 Racks appeared first on StorageReview.com.

## PD Serving of Qwen3.8-2.4T

DevFeed: [PD Serving of Qwen3.8-2.4T](<https://devfeed.tech/articles/pd-serving-of-qwen3-8-2-4t-79490.md>)

Original publisher: [Read original article](<https://vllm.ai/blog/2026-09-21-qwen38-pd-serving>)

Author: vLLM Team

Published: 2026-09-21T00:00:00Z

Content type: article

Language: en

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

Topics: [vllm](<https://devfeed.tech/topics/vllm.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [Amazon SageMaker HyperPod](<https://devfeed.tech/topics/amazon-sagemaker-hyperpod.md>), [Dynamo](<https://devfeed.tech/topics/dynamo.md>), [CUDA Graphs](<https://devfeed.tech/topics/cuda-graphs.md>)

Tags: [8k](<https://devfeed.tech/tags/8k.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [api](<https://devfeed.tech/tags/api.md>), [batch](<https://devfeed.tech/tags/batch.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [bottlenecks](<https://devfeed.tech/tags/bottlenecks.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [concurrency](<https://devfeed.tech/tags/concurrency.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cuda](<https://devfeed.tech/tags/cuda.md>), [cuda-graphs](<https://devfeed.tech/tags/cuda-graphs.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [disaggregation](<https://devfeed.tech/tags/disaggregation.md>), [docker](<https://devfeed.tech/tags/docker.md>), [dynamo](<https://devfeed.tech/tags/dynamo.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [model-serving](<https://devfeed.tech/tags/model-serving.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [open-source-ai](<https://devfeed.tech/tags/open-source-ai.md>), [performance](<https://devfeed.tech/tags/performance.md>), [qwen3](<https://devfeed.tech/tags/qwen3.md>), [qwen3-8](<https://devfeed.tech/tags/qwen3-8.md>), [vllm](<https://devfeed.tech/tags/vllm.md>), [vllm-blog](<https://devfeed.tech/tags/vllm-blog.md>)

### AI overview

The article explains how the vLLM team tuned disaggregated prefill and decode serving for the Qwen3.8-2.4T model on a GB300 NVL72 cluster. It describes estimating KV cache capacity, measuring prefill and decode performance, and selecting serving configurations. The reported results are 5,000 total tokens per second per GPU in a high-throughput scenario and 180 generated tokens per second per user in a low-latency scenario; it also provides recipes for reproducing the measurements.

### Source excerpt

How vLLM reaches 5K throughput and 180 interactivity on Qwen3.8-2.4T with GB300 NVL72 PD serving and how to reproduce results yourself.

## How Speculators and Mooncake enabled multi-node DSpark training for Kimi K3

DevFeed: [How Speculators and Mooncake enabled multi-node DSpark training for Kimi K3](<https://devfeed.tech/articles/how-we-trained-the-fastest-dspark-for-kimi-k3-using-gb300-nvl72-79487.md>)

Original publisher: [Read original article](<https://vllm.ai/blog/2026-09-15-kimi-k3-dspark>)

Author: Helen Zhao, Fynn Schmitt-Ulms, Yuchen Fama, Antonio J. Dominguez, and Kevin Li

Published: 2026-09-15T00:00:00Z

Content type: article

Language: en

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

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Multi-GPU](<https://devfeed.tech/topics/multi-gpu.md>), [NCCL](<https://devfeed.tech/topics/nccl.md>), [GB300 NVL72](<https://devfeed.tech/topics/gb300-nvl72.md>), [NVFP4](<https://devfeed.tech/topics/nvfp4.md>)

Tags: [accelerators](<https://devfeed.tech/tags/accelerators.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [bare-metal](<https://devfeed.tech/tags/bare-metal.md>), [blackwell](<https://devfeed.tech/tags/blackwell.md>), [cuda](<https://devfeed.tech/tags/cuda.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [dspark](<https://devfeed.tech/tags/dspark.md>), [gb300](<https://devfeed.tech/tags/gb300.md>), [gb300-nvl72](<https://devfeed.tech/tags/gb300-nvl72.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [model-serving](<https://devfeed.tech/tags/model-serving.md>), [node](<https://devfeed.tech/tags/node.md>), [nvl72](<https://devfeed.tech/tags/nvl72.md>), [open-source-ai](<https://devfeed.tech/tags/open-source-ai.md>), [performance](<https://devfeed.tech/tags/performance.md>), [speculative-decoding](<https://devfeed.tech/tags/speculative-decoding.md>), [speculators](<https://devfeed.tech/tags/speculators.md>), [training](<https://devfeed.tech/tags/training.md>), [vllm-blog](<https://devfeed.tech/tags/vllm-blog.md>)

### AI overview

The vLLM Speculators library supports training and deploying DSpark speculative decoding models, with a Kimi K3 implementation reporting higher generation speed and throughput. A Mooncake connector streams hidden states between vLLM and training processes across nodes, enabling multi-node training for the 2.8-trillion-parameter model.

### Source excerpt

How Speculators and Mooncake enabled multi-node DSpark training for Kimi K3.