Verda
Verified ProviderTier 3+ Sovereign Nordic Datacenter 4.9 / 5.0 (58 reviews)Next-Gen AI Cloud: NVIDIA B300, B200 & H200 Clusters
Verda (formerly DataCrunch) is Europe's premier sovereign AI compute cloud. Operating 100% renewable hydro/geothermal powered datacenters across the Nordics, Verda provides instant on-demand and spot GPU instances powered by NVIDIA Blackwell B300/B200 SXM6, H200 141GB, and H100 SXM5 with 3.2 Tbps Quantum-2 InfiniBand networking.
Verda GPU Catalog & Real-Time Hourly Pricing (9 Tiers)
| GPU Model (Click for Specs) | VRAM | Interconnect | vCPU / System RAM | Spot Hourly | On-Demand Hourly | Actions |
|---|---|---|---|---|---|---|
| NVIDIA RTX A6000 48GB | 48 GB | PCIe 4.0 x16 | 10 vCPUs / 60GB | $0.39/hr-50% | $0.77/hr | |
| NVIDIA RTX 6000 Ada Generation 48GB | 48 GB | PCIe 4.0 x16 | 10 vCPUs / 60GB | $0.69/hr-50% | $1.38/hr | |
| NVIDIA L40S 48GB | 48 GB | PCIe 4.0 x16 | 20 vCPUs / 60GB | $0.90/hr-50% | $1.79/hr | |
| NVIDIA A100 SXM4 80GB | 80 GB | NVLink (Fast) | 22 vCPUs / 120GB | $1.04/hr-50% | $2.09/hr | |
| NVIDIA H100 SXM5 80GB | 80 GB | NVLink (Fast) | 30 vCPUs / 120GB | $2.21/hr-50% | $4.43/hr | |
| NVIDIA H200 141GB SXM5 | 141 GB | NVLink (Fast) | 44 vCPUs / 170GB | $2.89/hr-50% | $5.77/hr | |
| NVIDIA B200 180GB SXM6 | 180 GB | NVLink (Fast) | 30 vCPUs / 170GB | $4.09/hr-50% | $8.19/hr | |
| NVIDIA B300 268GB SXM6 | 268 GB | NVLink (Fast) | 30 vCPUs / 255GB | $5.24/hr-50% | $10.48/hr | |
| 8x NVIDIA H100 SXM5 80GB | 640 GB | NVLink (Fast) | 240 vCPUs / 960GB | $17.70/hr-50% | $35.41/hr |
Verda Monthly Cost & Stopped Storage Estimator
Simulate your real-world monthly bill including active GPU compute and stopped persistent volume storage overhead.
When you click Stop Instance on Verda, your GPU compute charge stops, but the 200GB disk volume continues billing 24/7 ($20.00/mo) until you terminate the volume. Always backup checkpoints to HuggingFace or S3 before terminating.
Hourly run-rate: $0.39/hr compute + $0.0274/hr disk
Equivalent cloud capacity on AWS would cost approx. $482/mo for 6h daily runtime.
- •Verda (formerly DataCrunch) datacenters are located in the Nordics (Finland, Iceland, Sweden); while excellent for European low latency (<15ms) and GDPR compliance, US West coast latency will range around 95-120ms.
- •Container disks are ephemeral: all custom checkpoints and dataset weights must be saved to attached NVMe persistent volumes ($0.20/GB/month) or will be discarded on instance termination.
- •Spot instances receive a 30-second SIGTERM signal via the local metadata API before preemption; automated checkpointing handlers in your training loop are required.
- ✓Use Verda's Instant Clusters for multi-node LLM pre-training; they come pre-configured with 3.2 Tbps Quantum-2 InfiniBand and native PyTorch NCCL fabric tuning.
- ✓Spot prices on B300 ($5.24/hr) and H100 SXM5 ($2.21/hr) offer some of the cheapest high-end compute rates in the world due to Nordic geothermal and hydro power efficiency.
- ✓If running confidential AI workloads or sensitive healthcare data, Verda provides dedicated Confidential Computing (CC) B300 and B200 instances with hardware-level memory encryption.
Full-Lifecycle TCO & Hidden Cost EstimatorReal Invoicing
Calculate actual project costs including stopped storage fees and network egress penalties.
Top Alternatives to Verda
Evaluate competing GPU clouds for pricing, speed, and platform reliability.
European sovereign AI cloud with 3.2 Tbps InfiniBand & managed Slurm
Nebius offers integrated managed Slurm, Kubernetes, and hosted LLM inference APIs alongside high-bandwidth H100 InfiniBand fabrics.
Specialized AI hyperscaler with multi-thousand GPU InfiniBand fabrics
CoreWeave excels at multi-thousand GPU hyperscale enterprise reservations for frontier LLM foundation models.
Developer-first AI cloud with curated 1-8x GPU nodes
Lambda Stack pre-configured environment with instant 1-click pods for small-to-medium teams.
Developer cloud with 1-click ComfyUI/vLLM pods & persistent storage
RunPod provides popular 1-click template pods and network volumes across global regions.
In-Depth Technical Head-to-Head Battles:
Verda (formerly DataCrunch) delivers the lowest hourly spot and on-demand rates in Europe for next-generation Blackwell B300 ($5.24/hr spot) and H200 SXM5, powered by 100% hydro/geothermal energy in the Nordics. Nebius AI Studio is preferred for teams requiring integrated managed Slurm, Kubernetes, and hosted LLM inference APIs.
Verda offers self-service instant clusters and transparent hourly pricing with zero minimum contract lock-ins, making it accessible to startups and scaleups. CoreWeave excels at multi-thousand GPU hyperscale enterprise reservations with custom multi-year commitments.
1-Click Docker & SSH Recipes for Verda
Tested terminal commands for immediate deployment on NVIDIA RTX A6000 48GB instances.
# 1. SSH into your Verda instance: # ssh root@<instance_ip> -p <ssh_port> # 2. Launch ComfyUI with FLUX.1 & SDXL model cache: docker run -d --gpus all --name comfyui \ --restart unless-stopped \ -p 8188:8188 \ -v /workspace/ComfyUI/models:/root/ComfyUI/models \ -v /workspace/ComfyUI/output:/root/ComfyUI/output \ yanwk/comfyui-boot:cu124 # Access ComfyUI in your browser at: http://<instance_ip>:8188
Verda Architecture, Gotchas & Engineering Verdict
Verda maintains dedicated enterprise Tier 3/4 colocation facilities with redundant UPS power, multi-gigabit fiber backbones, and formal uptime SLAs.
- ✓Extremely competitive hourly spot rates with up to 60% savings.
- ✓Direct root SSH and Docker container flexibility.
- ✓Instant credit card onboarding with transparent billing.
- !Stopped Storage Fee: Persistent disk volumes accrue billing 24/7 ($0.20/GB/month (NVMe, 4 GB/s throughput)) until deleted.
- !Bandwidth Variance: Verify host download speeds before pulling 50GB+ model checkpoints.
Developer Reviews & Telemetry Audit
Aggregated from 58 verified AI engineers deploying vLLM, ComfyUI, and production API workloads.
“We transitioned our 70B parameter LLM fine-tuning cluster to Verda's Finland datacenter. The Blackwell B300 memory capacity (268GB) allowed us to train full-precision FP8 without pipeline-parallel sharding bottlenecks. The 3.2 Tbps Quantum-2 InfiniBand latency was pristine.”
“Renting H100 SXM5 instances on spot for $2.21/hr cut our experiment budgets in half. The 30-second SIGTERM notice gave our PyTorch checkpoint handler plenty of time to save weights to the NVMe persistent volume.”