96 GB on a single PCIe card
96 GB of GDDR7 with ECC at 1,597 GB/s, double the 48 GB of L40S and RTX 6000 Ada, keeps the weights of a 70B-class LLM in FP4 resident on one card with room for KV cache and long contexts, no sharding across GPUs.
// coming soon: blackwell
NVIDIA RTX PRO 6000 Blackwell GPU Server
Blackwell-generation bare metal for AI inference, agentic AI, digital twins and visual computing: 96 GB GDDR7 at 1,597 GB/s, 4 PFLOPS of FP4 Tensor performance, Multi-Instance GPU, and PCIe Gen 5 in standard servers with up to 8 cards per node. Reserve capacity now.
- Coming soon, reservations open
- 96 GB GDDR7, 4 PFLOPS FP4, MIG
- 20+ Tier III global data centers
- Crypto payments & 24/7 support included
New HardwareComing Soon
Request Sent
Thanks. An engineer will spec your RTX PRO 6000 build and reach out shortly.
Something went wrong. Please try again or email sales@redswitches.com.
// reserve capacity
Reserve NVIDIA RTX PRO 6000 Capacity
RTX PRO 6000 Blackwell Server Edition cards are landing in our data centers soon. Leave your details and our engineers will reserve a configuration for you, confirm lead times, and quote 1x, 2x, 4x and 8x PCIe builds with volume and committed-term discounts.
Request Sent
Thanks for reaching out. An engineer will spec your build and get back to you shortly.
Something went wrong. Please try again or email sales@redswitches.com.
// why redswitches for rtx pro 6000 blackwell
What You Get With Every NVIDIA RTX PRO 6000 Blackwell Server
Every RedSwitches NVIDIA RTX PRO 6000 Blackwell server is single-tenant bare metal with full root and IPMI access, unmetered 1, 10 or 25 Gbps bandwidth, no setup fee, no bandwidth overage and a flat monthly price. Stocked builds are online in about 1 hour, larger configurations up to 8 GPUs per node are built to order, volume and 6/12-month committed-term discounts apply, payments include crypto, and engineers answer 24/7.
- Free
Reserve at No Cost
Reservations for NVIDIA RTX PRO 6000 Blackwell are free and non-binding. An engineer confirms the lead time, holds the hardware for you, and sends a quote before you commit to anything.
- $0
Setup Fee, Flat Monthly
No setup fee and a flat monthly price for the whole server. No per-hour meter and no surprise line items, so GPU spend is forecastable.
- Unmetered
Bandwidth, No Egress Bills
Unmetered 1, 10 or 25 Gbps uplinks are included, and whatever port speed you choose you can use all of it: no overage charges and no egress bills, ever. Moving datasets, checkpoints and model weights in and out costs nothing extra.
- 1 tenant
Bare Metal, Full Control
Single-tenant hardware with root and IPMI access. You choose the OS, drivers, CUDA or ROCm version, and the NVLink or MIG layout, and a private VLAN can link your RedSwitches servers.
- Up to 8x
Multi-GPU, Built to Order
Single and multi-GPU nodes, up to 8 GPUs per server with NVLink where the card supports it. Volume discounts on multi-GPU and multi-server orders, plus 6 and 12-month term savings.
- 24/7
Engineers, Not Bots
Live chat, Telegram and email answered by engineers around the clock. Pay by card, PayPal, bank wire or crypto with no KYC, in 20+ Tier III data centers across the EU, US and Asia.
// compared with cloud
RedSwitches RTX PRO 6000 Blackwell Server vs a Typical Cloud GPU Instance
Same accelerator, different economics: what changes when the GPU sits in a dedicated server you control instead of a metered instance.
| RedSwitches RTX PRO 6000 Blackwell server | Typical cloud GPU instance | |
|---|---|---|
| Billing | Flat monthly price per server | Per-hour or per-second metering |
| Bandwidth | Unmetered 1/10/25 Gbps, use the full port, no overage or egress fees | Egress billed per GB |
| Private networking | Private VLAN between your servers on request | Paid VPC and peering constructs |
| Setup fee | $0 | Varies by instance and region |
| Tenancy | Single-tenant bare metal | Shared, virtualised hosts |
| Access | Root and IPMI, your OS and drivers | Hypervisor-managed images |
| Multi-GPU | Up to 8 per node, NVLink where supported, built to order | Fixed instance shapes |
| Payments | Card, PayPal, bank wire, crypto | Card or invoice |
| Support | 24/7 engineers on chat, Telegram and email | Ticket tiers, paid support plans |
NVIDIA RTX PRO 6000 Key Specifications
Blackwell GB202 silicon with 96 GB GDDR7 at 1,597 GB/s, 4 PFLOPS FP4 Tensor performance, up to 4 MIG instances, and PCIe Gen 5 at up to 600 W.
- Architecture
- NVIDIA Blackwell (GB202), 24,064 CUDA cores, fifth-generation Tensor Cores, 188 fourth-generation RT Cores
- Memory
- 96 GB GDDR7 with ECC, 512-bit interface, 1,597 GB/s bandwidth
- FP4 Tensor Core
- 4 PFLOPS
- FP8 Tensor Core
- 2 PFLOPS
- FP16/BF16 Tensor Core
- 1 PFLOPS
- TF32 Tensor Core
- 234 TFLOPS
- FP32
- 120 TFLOPS
- RT Core
- 355 TFLOPS
- Form Factor & TDP
- Air-cooled passive dual-slot FHFL or liquid-cooled single-slot FHXL, up to 600 W (configurable)
- PCIe
- Gen 5 x16 to the host
- NVLink
- None, multi-GPU scaling over PCIe
- Multi-Instance GPU
- Up to 4 fully isolated instances per GPU
- Media & Display
- Ninth-generation NVENC, sixth-generation NVDEC, AV1 encode and decode, 4x DisplayPort 2.1
- Virtualization
- NVIDIA vGPU software support for virtual workstations and VDI
- Software
- CUDA, NVIDIA AI Enterprise, NIM microservices, TensorRT-LLM, NVIDIA Omniverse, PyTorch, JAX, Triton Inference Server
Computeper GPU, Tensor figures with sparsity
Interconnect
// why rtx pro 6000
Why Choose RTX PRO 6000
Blackwell for standard PCIe servers: double the memory of L40S and RTX 6000 Ada, FP4 inference, MIG isolation, and full graphics, RT and video engines on one card.
Blackwell FP4 inference
Fifth-generation Tensor Cores deliver 4 PFLOPS of FP4 and 2 PFLOPS of FP8 performance per card, and NVIDIA rates LLM inference at up to 5x the previous-generation Ada Tensor Cores: more tokens per second per card and per watt.
Multi-Instance GPU for multi-tenant serving
Partition each card into up to 4 fully isolated MIG instances, each with its own memory and compute, so several models, tenants or agents share one GPU with hardware-level isolation and predictable latency.
Standard PCIe servers, up to 8 cards
PCIe Gen 5 x16 in a passive dual-slot air-cooled or single-slot liquid-cooled card means dense builds in off-the-shelf chassis: 1x to 8x RTX PRO 6000 per node, no SXM baseboards and no NVLink fabric to plan.
Graphics, ray tracing and video on the same card
188 fourth-generation RT Cores at 355 TFLOPS, 4x DisplayPort 2.1, ninth-generation NVENC and sixth-generation NVDEC with AV1 make one card the platform for Omniverse digital twins, rendering, video pipelines and vGPU workstations alongside AI.
Bare metal, not shared cloud
Single-tenant servers with full root and IPMI access, dual AMD EPYC or Intel Xeon hosts, NVMe storage, unmetered 1/10/25 Gbps uplinks, and no noisy neighbours, deployed in 20+ Tier III data centers across the EU, US and Asia.
// use cases
Ideal Use Cases
From 70B-class LLM inference on one card to agentic AI, digital twins, rendering, video and VDI, where up to 8 RTX PRO 6000 cards per node pay for themselves.
LLM Inference on One Card
Serve Llama, Mistral, Qwen and DeepSeek-class models up to 70B parameters in FP4 from a single 96 GB card with TensorRT-LLM or NIM, then scale out across up to 8 cards per node.
Agentic AI and Multi-Tenant Inference
Run many concurrent model instances and agent workers on MIG partitions with hardware isolation, built for AI platforms that serve several customers or teams from one server.
Digital Twins and Omniverse
Build and stream physically accurate factory, facility and product twins and large USD scenes with fourth-generation RT Cores and 96 GB of scene memory.
Rendering and Visual Effects
Ray-traced final-frame rendering in Arnold, V-Ray, Blender Cycles, Octane and Redshift, with 96 GB for large scenes and 4x DisplayPort 2.1 for review stations.
Video Processing and Streaming
Transcode, encode and decode at scale with ninth-generation NVENC and sixth-generation NVDEC, including AV1, for live streaming, media pipelines and generative video.
Virtual Workstations and VDI
NVIDIA vGPU support lets one card host multiple virtual workstations for CAD, media and data science teams, with hardware-accelerated graphics from a server you control.
Trusted by Enterprise Teams Worldwide
// customer reviews
Trusted in Production
Real reviews from Google, HostAdvice, and Cryptwerk.
- 25Gbps Bare Metal
Port speeds on cloud were capped and shady. With RS I actually get the 25Gbps they say. No throttle bs.
- Validators
Set up multiple Solana + Avalanche validators through them. Hardware was clean, latency was low (especially in Europe), and uptime’s been 100% so far.
- Full Root Access
As a sysadmin, I care more about control than flashy dashboards. RedSwitches gives me root access, IPMI, and actual hardware specs I can configure. Good for serious users.
- Cloud Exit
RedSwitches gave us full control over our infrastructure without the vendor lock-in we kept running into with cloud hosts. Customizable builds, fast provisioning, and actual humans handling support tickets. It’s refreshing.
- 99.99% Uptime
My website works smoothly, thanks to their 99.99% uptime guarantee. The support team is another plus for me, as I always have been able to get help whenever I needed it in just 5 minutes on average.
- ETH & BTC Nodes
Deployed a few Ethereum and Bitcoin nodes here. Uptime’s been flawless, and sync speed was great thanks to their storage config.
- Game Servers
Been using RedSwitches for 6 months now for my small game server biz. Uptime has been great, and I haven’t run into any hidden charges.
- AI Storage
Using the storage servers to archive logs and snapshots from our AI pipeline… I also love that I could pick the datacenter closest to our team.
- Support
Support is super responsive. I had an issue with an OS reinstall and they jumped in within 10 minutes… Transparent pricing = win.
- Instant Delivery
They have an instant delivery section… they delivered it within 120 mins with all my requirements fulfilled (OS/RAID/Software configured etc).
- Global Locations
With servers available in numerous strategic locations, RedSwitches offers exceptional versatility and performance for our company’s diverse hosting needs. Plus, their no setup fee policy really helps keep costs down.
- Bare Metal Cloud
The dedicated server I got from RedSwitches has been incredibly reliable and fast. Their bare metal cloud solutions offer excellent performance, and the cloud VPS options are perfect for scaling. Highly recommended!
- Easy Onboarding
Absolutely delighted with RedSwitches! The setup was quick and free, and the fact that they accept all major payment gateways made the process seamless.
- Bare Metal
Very good experience using their bare metal servers. Their customer service is one of the finest I have experienced - always prompt at resolving troubles. Highly recommend.
// faqs
Deep Dive & FAQs
Availability and reservations, RTX PRO 6000 vs L40S and RTX 6000 Ada, pricing, MIG, CUDA and FP4, power and cooling, and graphics, vGPU and video.
When will RTX PRO 6000 Blackwell servers be available and how does reservation work?
RTX PRO 6000 Blackwell Server Edition capacity is being installed across our data centers now. Submit the reservation form with your preferred card count and region; our engineers confirm the lead time for that build, hold the hardware for you, and send a quote. Reservations are free and non-binding until you approve the quote.
RTX PRO 6000 vs L40S: which should I choose?
RTX PRO 6000 Blackwell Server Edition doubles the L40S on memory, 96 GB GDDR7 vs 48 GB GDDR6, and nearly doubles bandwidth, 1,597 GB/s vs 864 GB/s. FP16/BF16 Tensor throughput rises from 733 TFLOPS to 1 PFLOPS, and FP4 is new to Blackwell, which is where the largest inference gains come from; max power rises from 350 W to a configurable 600 W. Choose L40S for budget-sensitive inference and graphics where 48 GB is enough; choose RTX PRO 6000 for 70B-class models on one card, FP4 serving and multi-tenant MIG.
RTX PRO 6000 vs RTX 6000 Ada: what changes?
RTX PRO 6000 Blackwell moves from Ada to Blackwell: 96 GB GDDR7 at 1,597 GB/s vs 48 GB GDDR6 at 960 GB/s, fifth-generation Tensor Cores with FP4 and up to 5x the LLM inference performance according to NVIDIA, and up to 4 MIG instances per card. Max power rises from 300 W to a configurable 600 W, and the Server Edition is a passive card built for data center chassis rather than workstations. RTX 6000 Ada remains a strong choice for workstation-class graphics and smaller models; RTX PRO 6000 is the upgrade for memory-bound inference and dense AI plus graphics servers.
How is RTX PRO 6000 priced and are there discounts?
Pricing is quoted per configuration because card count, host CPU, memory, storage and uplink choices vary. Volume discounts apply to multi-GPU and multi-server orders, and every GPU server qualifies for committed-term discounts on 6 and 12-month billing. Crypto, card, PayPal and bank wire are accepted, and there is no setup fee.
Does RTX PRO 6000 support MIG and multi-tenant inference?
Yes. Each RTX PRO 6000 Blackwell Server Edition can be partitioned into up to 4 fully isolated Multi-Instance GPU instances, each with dedicated memory and compute. That lets one card serve several models, customers or agent pools with hardware-level isolation and predictable performance, and an 8-card node can expose up to 32 isolated GPU instances.
Is my CUDA code compatible and how do I use FP4?
Yes. Blackwell runs the same CUDA programming model, so existing CUDA builds, PyTorch, JAX, TensorFlow and TensorRT workloads run unchanged. FP4 is exposed through the fifth-generation Tensor Cores via TensorRT-LLM, NIM microservices and NVIDIA AI Enterprise: quantize a model to FP4 or pull a prebuilt FP4 checkpoint and serve it without application changes. We pre-install the NVIDIA driver, CUDA toolkit and container runtime on request, on Ubuntu, Debian, Rocky/AlmaLinux, Windows Server or your own ISO.
What about power and cooling, air or liquid?
RTX PRO 6000 Blackwell Server Edition draws up to 600 W, configurable lower, and ships as a passive dual-slot FHFL air-cooled card or a single-slot FHXL liquid-cooled card. Our GPU racks are provisioned for high-density accelerators; we size the chassis, power feeds and cooling for the card count you reserve, so you plan the workload, not the facility.
Can I use it for graphics, vGPU and video as well as AI?
Yes. Unlike pure compute accelerators, RTX PRO 6000 includes 188 fourth-generation RT Cores, 4x DisplayPort 2.1 outputs, ninth-generation NVENC and sixth-generation NVDEC with AV1, and NVIDIA vGPU support. The same server can run Omniverse digital twins, ray-traced rendering, video transcoding and virtual workstations alongside AI inference, and you can split those workloads across MIG instances. If your workload is training-only and bandwidth-bound, compare it with H100 NVL, 94 GB HBM3 at 3.9 TB/s.
// need help
Not Sure Exactly What You Need
No problem. Our talented engineers will consult, architect, migrate, manage, and do whatever it takes to help your business grow and succeed.