CoreWeave and NVIDIA reported on September 30, 2026, that NVIDIA Vera Rubin NVL72 systems were available on CoreWeave Cloud, with Cognition running production workloads on the system. That is a later status than CoreWeave’s January 5 announcement, which said it expected to add Rubin in the second half of 2026. Vera Rubin NVL72 is rack-scale data-center infrastructure accessed as a cloud service—not a retail GPU—and CoreWeave pairs it with Kubernetes, operations, sandboxing, and inference services.
What is NVIDIA Vera Rubin NVL72?
Vera Rubin NVL72 is a rack-scale AI system built around NVIDIA Rubin GPUs and Vera CPUs. CoreWeave’s June 1, 2026, description specifies 72 Rubin GPUs and 36 Vera CPUs per rack, connected by sixth-generation NVLink with a stated fabric bandwidth of 260 TB/s. A later CoreWeave announcement also describes ConnectX-9 SuperNICs and BlueField-4 DPUs in each rack. Multiple racks can be connected using Spectrum-X Ethernet to form larger clusters.
That scale matters: NVL72 is not a standalone card a user installs in a workstation. CoreWeave offers it as infrastructure through its cloud platform, with the provider managing the underlying rack and cluster operations. CoreWeave describes its Mission Control layer as providing observability and operations, and its Kubernetes-native Rack Lifecycle Controller as coordinating rack provisioning, power operations, and hardware validation. CoreWeave’s June 1 system announcement and its September cluster announcement provide the hardware details.
Is Vera Rubin available on CoreWeave?
Yes. In its September 30, 2026, announcement, CoreWeave reported Vera Rubin NVL72 availability on CoreWeave Cloud and named Cognition as the first customer running production workloads on it. NVIDIA published a separate report the same day describing availability and the platform services used to operate the capacity. These announcements update, rather than contradict, the January plan: at the time, CoreWeave said it expected Rubin deployment in the second half of 2026.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
- PLEASE NOTE: Exporting an NVIDIA RTX Pro 6000 GPU outside the US requires strict adherence to the U.S. Export Administration Regulations (EAR) and issuance of an export license from the Bureau of Industry and Security (BIS). Compliance and Know Your Customer (KYC) screening may be required as a condition of order acceptance. [NVIDIA Blackwell Streaming Multiprocessor] The new SM features increased processing throughput, and new neural shaders that integrate neural networks inside of programmable shaders | DLSS 4: Multi Frame Generation ensures ultra-smooth frame pacing for lifelike simulations.
- [Double-Flow-Through Design] The RTX PRO 6000 Blackwell features a double-flow-through cooling design, optimizing efficiency and airflow to sustain peak performance under 600W power loads. | [5th Gen Tensor Cores] Deliver up to 3X the performance of the previous generation and support for FP4 precision for faster AI model processing times with reduced memory usage, enabling local fine-tuning of LLMs and generative AI | [4th Gen Ray Tracing Cores] Double the ray-triangle intersection rate of the previous generation to create photoreal, physically accurate scenes and immersive 3D designs with RTX Mega Geometry, which enables up to 100X more ray-traced triangles.
- [PCIe Gen 5] Support for PCIe Gen 5 provides double the bandwidth of PCIe Gen 4, improving data-transfer speeds from CPU memory and unlocking faster performance for data-intensive tasks like AI, data science, and 3D modeling. | [GDDR7 Memory] With 96 GB of GPU memory and 1.8 TB ps bandwidth, it can tackle massive 3D and AI projects, fine-tune AI models locally, explore large-scale VR environments, and drive larger multi-app workflows.
- [DisplayPort 2.1] Achieve unparalleled visual clarity and performance, driving high resolution displays at up to 8K at 240 Hz and 16K at 60 Hz. Increased bandwidth enables seamless multi-monitor setups while HDR and higher color depth support ensures superior color accuracy for precision work, such as video editing, 3D design, and live broadcasting.
- [Universal MIG] Divide a single RTX PRO 6000 Blackwell into multiple isolated instances, each with dedicated resources, allowing for concurrent execution of multiple workloads, optimized GPU utilization, and secure isolation of different applications or users. [WARRANTY] 3 YR Manufacturer's Warranty. Bulk OEM Packaging. Retail Packaging is NOT included.
| Date | What was announced |
|---|---|
| January 5, 2026 | CoreWeave said it expected to add Rubin systems in the second half of 2026; NVIDIA described a similar partner availability window. These were plans, not reports of production customer use. CoreWeave and NVIDIA |
| June 1, 2026 | CoreWeave said it had brought up and completed system-level validation for a Vera Rubin NVL72 rack. CoreWeave |
| August 20, 2026 | CoreWeave announced a multi-year Hudson River Trading agreement for AI research and model training on CoreWeave, including Vera Rubin infrastructure. CoreWeave |
| September 16, 2026 | CoreWeave announced a multi-rack cluster connecting hundreds of Rubin GPUs. CoreWeave |
| September 30, 2026 | CoreWeave and NVIDIA reported Vera Rubin NVL72 availability; Cognition was identified as the first production customer on the system. CoreWeave and NVIDIA |
The announcements establish that production use had begun, but they do not specify current pricing, a minimum capacity commitment, eligibility for new customers, or the general access process. A named production deployment should not be read as a guarantee that every prospective customer can immediately reserve capacity.
What AI tools does CoreWeave offer with Rubin?
The named services are separate parts of CoreWeave’s cloud platform; the announcements do not establish that every one is bundled into every customer’s service tier.
- CoreWeave Kubernetes Service: Kubernetes-based environment for deploying and managing workloads.
- SUNK: CoreWeave’s Kubernetes integration for NVIDIA GPUs.
- Mission Control: provider-described observability and operations capabilities for infrastructure.
- CoreWeave Sandboxes: isolated environments for working with AI workloads.
- CoreWeave Inference: inference service for running model workloads.
- CoreWeave Forge: NVIDIA’s September report describes this as combining Weights & Biases, OpenPipe post-training expertise, and the open-source marimo notebook project.
NVIDIA’s September 30 report describes the services available for operating Rubin capacity. Details on which services are included, optional, or separately provisioned depend on the customer’s arrangement.
Rank #2
- VD8465 Japanese Authorized Distributor Product
- The speed of FP32 calculation is twice as fast as previous generations, which greatly improves the complex 3D processing and graphics simulation workflow
- Up to 2X the throughput compared to previous generations and significantly faster workloads such as video content rendering, architectural design assessments, and virtual prototypes of product design
- Achieve more than twice the previous generation AI performance improvement, support faster FP8 precision data and accelerate the execution of mixed flotation decimal and whole numbers
- It has a large capacity of memory necessary for working with a vast array of data sets and workloads such as rendering, data science, and simulation
How much faster is Vera Rubin than GB200?
There is no single speedup that applies to every model or workload. The published figures are company- and customer-reported results for specific workloads, with GB200 NVL72 as the comparison system.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errors| Workload and result | What the figure means | Attribution and source |
|---|---|---|
| SWE-2 software-engineering inference: up to 4.8× total token throughput | Cognition’s reported result on Vera Rubin NVL72 compared with a GB200 NVL72 baseline. “Up to” and the named workload are essential qualifications; the figure is not a general speedup for all inference. | Cognition, as reported by CoreWeave and NVIDIA |
| Cognition reinforcement-learning workloads: 3.8× output-token throughput | CoreWeave’s reported result for those workloads; it is not the same metric as total token throughput. | CoreWeave |
| DeepSeek R1 reasoning: 10× token throughput per megawatt | CoreWeave’s comparison with GB200 NVL72 at matched interactivity. Throughput per megawatt is not a claim of 10× lower customer bills or total cost. | NVIDIA, reporting the CoreWeave result |
The available announcements do not establish independent replication or provide a basis for treating these figures as general-purpose benchmarks. Performance can vary with model, serving setup, workload, and latency or interactivity requirements.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which workloads are these systems intended for?
CoreWeave and NVIDIA position Rubin for large-scale model training, inference, reasoning, mixture-of-experts models, and agentic AI. CoreWeave’s January announcement also named drug discovery, genomic research, climate simulation, and fusion-energy modeling as potential workloads. Those are stated target areas, not evidence that each has already been validated on the announced production deployment.
Rank #3
- Small in Size, Serious in Performance — a space-saving design delivering professional-class performance, enterprise-grade security and reliability, flexible deployment options, and a MIL-STD-810H–certified build engineered for demanding work environments.
- Extreme AI and professional graphics performance — The ThinkStation P3 Ultra SFF Gen 2 combines an integrated Intel NPU with NVIDIA RTX 4000 SFF Ada Generation graphics (20GB GDDR6) to deliver up to 335 TOPS of AI performance across CPU and GPU. Ideal for AI inferencing, deep learning, 3D animation, content creation, advanced imaging, 3D modeling, and BIM software—all in a compact, energy-efficient workstation.
- Fast, secure storage with next gen memory & business-ready OS — 2TB PCIe Gen 5 TLC Opal SSD for ultra fast boot and load times, MAXED OUT 128GB DDR5-6400MHz memory, and Windows 11 Professional preinstalled.
- Easy-access front connectivity — USB-A (USB 10Gbps), 2 x USB-C (USB4 20Gbps) – data transfer only, Headphone/mic combo
- Warranty — Factory Sealed. 1 Year Lenovo Warranty
Cognition, the applied AI lab behind Devin, is the clearest announced production example, using the system for software-engineering inference and reinforcement learning. Hudson River Trading provides a separate enterprise research example: its August multi-year agreement covers AI-driven trading research and model development, with Vera Rubin and other systems in its CoreWeave platform mix. The agreement does not establish access terms for other buyers.
What should a buyer verify before comparing cloud options?
Published system specifications and selected workload results are useful starting points, but they do not answer the commercial and operational questions that determine fit. Compare providers against the same workload and ask for comparable terms on:
- Capacity and access: confirmed availability, reservation timing, and any minimum commitment.
- Workload performance: throughput and latency for the models and serving conditions you actually use.
- Efficiency: performance per watt measured under comparable workload and interactivity conditions.
- Data path: networking and storage configuration, including how data reaches the compute cluster.
- Software and operations: orchestration, observability, deployment tools, and the support model.
- Security and location: data residency, access controls, and other requirements relevant to your workload.
- Commercial terms: current price, billing basis, capacity commitment, and cancellation or scaling terms.
The announcements cited here do not provide comparable current prices, a cross-provider benchmark, or complete buying terms. Request those details directly before making a purchasing decision.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
