# Rent NVIDIA Vera Rubin NVL72: reserve rack-scale capacity > Reserve NVIDIA Vera Rubin NVL72 capacity: 72 Rubin GPUs with 288 GB of HBM4 each and 36 Vera CPUs. Price on request; terms confirmed before any payment. Source: https://overgpu.com/gpus/nvidia-vera-rubin-nvl72 Updated: 2026-10-23 Publisher: OverGPU (https://overgpu.com/) **In short:** OverGPU takes reservations for the NVIDIA Vera Rubin NVL72: one rack of 72 GPUs with 288 GB of HBM4 each, price on request. Reservations are made online, and nothing is charged until OverGPU confirms the price and the term. ## Vera Rubin NVL72 configurations and monthly prices at OverGPU | Configuration | Type | GPU memory | CPU | RAM | Storage | Network | Price per month (USD) | Per GPU-hour | Availability | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | [NVIDIA Vera Rubin NVL72](https://overgpu.com/gpus/nvidia-vera-rubin-nvl72) | Rack-scale | 72 × 288 GB HBM4 | 36× NVIDIA Vera | Up to 54 TB LPDDR5X | Storage sized per order | NVLink 6 · ConnectX-9 | On request | — | Reservations open, by reservation | Per GPU-hour = monthly price ÷ number of GPUs ÷ 730 hours, shown only to compare with hourly offers: every plan is billed as one flat monthly price. ## NVIDIA Vera Rubin NVL72 specifications (vendor-published) | Specification | Value | Note | | --- | --- | --- | | Architecture | Rubin | NVIDIA | | GPU memory | 288 GB HBM4 | Preliminary figures published by NVIDIA, subject to change. | | Memory bandwidth | 19.2 TB/s | | | FP4 | 35,000 TFLOPS | NVFP4 · 50,000 TFLOPS with sparsity | | FP8 | 17,500 TFLOPS | Dense | | BF16 / FP16 | 4,000 TFLOPS | Dense | | GPU interconnect | NVLink 6 | 72-GPU domain | | Form factor | Vera Rubin rack · liquid-cooled | | | Max power | Rack-level | | ## What the Vera Rubin NVL72 is best at - **Securing future capacity:** Register interest early for an allocation while rack-scale GPU capacity stays tight. - **288 GB of HBM4 per GPU:** About 20.7 TB of HBM4 across 72 GPUs for frontier-scale training and inference. - **One NVLink 6 domain:** 72 Rubin GPUs and 36 Vera CPUs connected in a single rack-scale NVLink 6 system. - **Next-generation training plans:** Teams planning their next large training runs on NVIDIA's Rubin generation. ## When another GPU fits better - [GB300 NVL72 rental from $7,990 per GPU-month](https://overgpu.com/gpus/nvidia-gb300-nvl72): When you need a 72-GPU NVLink domain now, on a 12 to 36 month term. - [Rent NVIDIA B300 and HGX B300 servers from $5,190 per month](https://overgpu.com/gpus/nvidia-b300): When you need capacity today on a one-month term, with the same 288 GB per GPU. ## Overview A Vera Rubin NVL72 reservation lets you claim a place for NVIDIA's next rack-scale system before it ships. Reservations are open now: you register interest, and we confirm allocation and terms in writing before any payment. If you want to rent Vera Rubin capacity for training or inference, this page covers what is known, how reserving works, and what to run in the meantime. ## What we can say about Vera Rubin NVL72 today - **GPUs:** 72 Rubin GPUs per rack, with 288 GB of HBM4 each, about 20.7 TB in total - **CPUs:** 36 Vera CPUs - **Interconnect:** NVLink 6, the generation after the NVLink 5 used in Blackwell systems - **Form:** Liquid-cooled, rack-scale system - **Availability:** Limited, expected in late 2026 That list is deliberately short. The compute and bandwidth figures in the specifications above are preliminary numbers published by NVIDIA and may change before systems ship; we confirm the exact configuration of the systems we rent in writing. What you can plan with today is memory: 288 GB per GPU, the same capacity as the B300, now in HBM4, a newer memory generation than the HBM3e of Blackwell systems. ## How a Vera Rubin NVL72 reservation works 1. **Register interest.** [Create a free account](https://overgpu.com/register) and [send a reservation request](https://overgpu.com/deploy) from the deployment page, with these details in the notes: what you plan to run, how many GPUs or racks you need, the term you have in mind, and your earliest and latest acceptable start dates. 2. **We confirm in writing.** When we can allocate capacity, we send the allocation, price and term in writing. Nothing is due before this confirmation. 3. **You decide.** If the terms work for you, you accept them; if not, you walk away without having paid anything. 4. **Payment.** You pay from your account balance, which you top up in crypto. For an order of this size, we may ask for verification of your company and intended use before confirming, as described in our [verification policy](https://overgpu.com/legal/verification). Sharing how flexible your dates are helps us match you with the capacity we can allocate. ## Should you reserve Vera Rubin or rent what ships now? A reservation makes sense when your plans already point past what is available today: - You have a training roadmap that runs into 2027 and calls for a 72-GPU NVLink domain on the newest generation. - Your models, or their training state, already outgrow an 8-GPU server, so you will need rack-scale capacity in any case. - You want an allocation and a price in writing before you commit budget. It makes less sense if you need GPUs this quarter, if your jobs fit in eight GPUs, or if you are still deciding what to train; monthly servers keep your options open in those cases. If you do reserve, a useful request states the model size and precision you plan to run, the number of GPUs, the term, your start window and any constraints on location, so that our written answer can be specific. ## Planning a Vera Rubin workload now Because memory per GPU matches the B300, you can size Vera Rubin deployments today with the same arithmetic. Weights take roughly parameters × bytes per parameter (2 in BF16, 1 in FP8, 0.5 in FP4), plus KV cache and 10–30% headroom, and full training with Adam needs roughly 16–20 bytes per parameter before activations. - **Inference of a one-trillion-parameter model in FP8:** about 1 TB of weights, so at least four GPUs for the weights alone; a full rack leaves roughly 19 TB for KV cache, activations and other models. - **Full training of a 405B-class model:** about 6.5–8.1 TB of training state, or 23 to 29 GPUs' worth of memory before activations, well within one 72-GPU rack. - **Full training of a one-trillion-parameter model:** about 16–20 TB, close to the rack's entire 20.7 TB of HBM4, so plan for activation memory on top, either across racks or with offloading. Our [VRAM calculator](https://overgpu.com/tools/vram-calculator) runs these numbers for other sizes. On the software side, a new GPU architecture usually means new driver, CUDA and framework versions, so keep your stack containerized and pinned, and plan time to validate it on the new systems before you move production work. ## Need capacity before Vera Rubin ships? Two options are available today, with up to 288 GB of memory per GPU: - **[HGX B300 servers](https://overgpu.com/gpus/nvidia-b300)**: eight Blackwell Ultra GPUs per server at $39,900 per month, with a one-month minimum term. Because plans run month to month, you can keep them until your Vera Rubin capacity is delivered and then [cancel with no fee](https://overgpu.com/help/renewals-and-cancellation). Stock is limited. - **[GB300 NVL72](https://overgpu.com/gpus/nvidia-gb300-nvl72)**: 72 Blackwell Ultra GPUs in one NVLink domain, from $7,990 per GPU per month on 12 to 36 month terms. Choose it if you need a rack-scale NVLink domain now, and set the term length with your Vera Rubin timeline in mind, since the two commitments would otherwise overlap. One practical approach is to start training on HGX B300 servers, keep checkpoints and data pipelines portable, and move the next run to Vera Rubin once it is delivered. HGX B300 servers are ordered directly on the deployment page and delivered in under 10 minutes after payment; your Vera Rubin reservation request is sent from the same page. ## Vera Rubin NVL72 pricing There is no list price for Vera Rubin NVL72 yet. Each reservation is confirmed in writing with its configuration, price and term, and nothing is payable until you accept it. Prices are in US dollars and exclude VAT and other taxes where applicable. When the time comes, you pay from your account balance, topped up in crypto: Bitcoin (BTC), Ethereum (ETH), Monero (XMR), USDT on TRON (TRC-20) or Solana (SOL), at an exchange rate locked for a limited time for each top-up. Card payments are coming soon, and every payment comes with an invoice in your dashboard. For an overview of everything we offer at this scale, see [rack-scale systems](https://overgpu.com/products/rack-scale-systems). ## Frequently asked questions ### How do I make a Vera Rubin NVL72 reservation? Create a free account and send a reservation request from the deployment page, with your workload, GPU count, term and preferred start dates in the notes. We then confirm allocation and terms in writing before any payment. ### Do I pay anything to reserve? No payment is taken when you register interest. You pay from your account balance only after accepting the written confirmation of allocation, price and terms. ### When will Vera Rubin NVL72 be available? Limited availability is expected in late 2026. ### How much does it cost to rent Vera Rubin NVL72? There is no list price yet. Each reservation is confirmed in writing with its configuration, price and term before you pay. ### Are the Vera Rubin figures final? No. The compute and bandwidth figures on this page are preliminary numbers published by NVIDIA and may change before systems ship. Plan with the memory capacity, 288 GB of HBM4 per GPU; we confirm the exact configuration in writing before you pay. ### What can I use until Vera Rubin arrives? HGX B300 servers have the same 288 GB per GPU on a one-month minimum term, and GB300 NVL72 racks are available on 12 to 36 month terms. Monthly HGX B300 plans can be cancelled with no fee once your Vera Rubin capacity is delivered. ## Related pages - [GPU servers for LLM training on dedicated 8-GPU nodes](https://overgpu.com/solutions/llm-training): Bare-metal HGX B300, B200 and H200 nodes with NVLink and 400–800G networking, rented monthly. See the memory math first. - [Dedicated GPU servers for enterprise, with a written SLA](https://overgpu.com/solutions/enterprise): Whole GPUs, bare-metal 8-GPU nodes, a 99.9% SLA with credits, VAT invoices and term pricing. Plan a pilot. - [Rack-scale GPU systems: GB300 NVL72 and Vera Rubin NVL72](https://overgpu.com/products/rack-scale-systems): Rent GB300 NVL72 from $7,990 per GPU-month on 12–36 month terms, or reserve Vera Rubin NVL72. See who needs a 72-GPU NVLink domain. - [GPU server setup help from our engineers](https://overgpu.com/products/expert-onboarding): First-boot checks, environment and migration advice from our engineers, and sizing tools before you order. - [GPU shortage 2026: what buyers face and how to secure GPUs](https://overgpu.com/blog/gpu-shortage-2026): On-demand stock sells out, providers push 1-year contracts and prices are up. Why it happens, and how teams still secure GPUs. - [B200 vs B300: is Blackwell Ultra worth the upgrade?](https://overgpu.com/compare/b200-vs-b300): 180 vs 288 GB of HBM3e, the FP4 uplift and 400G vs 800G networking. See what fits, the cost per GB and when the B200 is enough.