The NVIDIA GB200 line listed on 26 August has sold through. The desk can still source NVIDIA GB200 off-page: send the quantity, condition and delivery window and we will check the wider market. Details below are the August listing, kept for reference.
A single verified system for teams building toward rack-scale NVL72 architecture.

The price above is ex-works. Between the supplier door and your rack sit freight and insurance at node weights, export screening, customs classification and duties, import VAT or GST and its recovery, and the last mile into the facility. We price all of it up front so two offers can be compared honestly. The mechanics are on the import and IOR desk.
Full specification and where this SKU fits is on the NVIDIA GB200 page. Everything else verified today sits on the live inventory. If this is not the right SKU for your workload, we will say so.
The stock we see on GB200 is channel-allocated: distributors and ODM partners holding built 2x superchip systems that shipped against allocations rather than end-customer orders. That is different from refurbished fleet stock. The unit is new, the warranty starts with you, and the price reflects what the channel paid NVIDIA, not a depreciation curve.
Current listed example: new 2x GB200 superchip systems at $158,000 ex-works US, 7 days to ship. Seven days tells you the box is on a shelf, not in a build queue. When lead time is measured in days rather than quarters, your leverage sits in payment terms and logistics, not in price, because the holder wants the capital freed.
We contract you directly with the supplier after KYC clears, and we acknowledge within one hour with a first sourcing pass inside one business day. Compensation comes from the supply side under disclosed arrangements.
The 2x superchip system pairs Grace CPUs with Blackwell GPUs in one module, so the buyers we see are running workloads where CPU-to-GPU memory bandwidth is the bottleneck: long-context inference, retrieval-heavy serving, and fine-tuning jobs that thrash host memory on conventional x86 hosts. It is a coherent-memory machine first and a FLOPS machine second.
Go below it if your models fit comfortably on discrete GPUs and interconnect is already your constraint; a standard Blackwell or Hopper server costs less per unit of raw compute. Go above it, to rack-scale NVL72, when you are training or serving models that need dozens of GPUs in one NVLink domain. The 2x system is the stepping stone: same architecture, single-module scale, deliverable now.
The GB200 stock we see is new channel-allocated hardware; the platform is too recent for a meaningful refurbished pool. We do handle new, refurbished and cancelled-order stock across our OEM, ODM and distributor channels, but for GB200 specifically expect new units, full warranty, and KYC completed before any supplier introduction.
Some of it is genuinely available. We currently have new 2x GB200 superchip systems listed at $158,000 ex-works US with 7 days lead time, sourced through the channel rather than a direct NVIDIA allocation. As a buyer-side procurement desk we acknowledge within one hour and run a first sourcing pass within one business day, and you contract directly with the supplier.
This line quotes 7 days from commitment, United States-located. Lead times only mean something when the allocation behind them is verified, and every line we publish is confirmed with the holder before it is listed.
Indicative $158,000 per system, ex-works USD. That excludes freight, insurance, importer-of-record, duties and local charges, all of which we price into a landed number before you commit. Volume, term and configuration move it.
New stock through the authorized channel with full KYC, which is precisely why the allocation is available. Warranty terms are confirmed for your jurisdiction before purchase.
Yes. Freight, insurance, customs and importer-of-record are handled by the same desk, subject to export screening on the destination and end user. We price the landed cost up front rather than discovering it at the border.
Twenty minutes with the desk, no pitch and no quote at the end of it. Tell us roughly what you need and we will come back within one business day.
Your enquiry has landed with the desk. Acknowledged within one hour.