# Verda: The Frontier European AI cloud > Verda is a European GPU cloud provider for AI workloads. It offers on-demand GPU instances, multi-node GPU clusters with InfiniBand, serverless containers, and managed inference endpoints. Verda operates some of the most advanced GPUs. All infrastructure runs on 100% renewable energy in EU data centers and is fully GDPR-compliant. Verda is purpose-built for GPU compute. Important notes: - Verda was previously called "DataCrunch." Some SDK packages and older documentation may still reference the DataCrunch name. They refer to the same platform and API. - Verda is not subject to the US Cloud Act. All data stays under European jurisdiction. - There is no sales call required. Instances, clusters, and containers can be launched immediately via the console, API, or SDK. ## Available Hardware Examples of modern GPU stack oferred by Verda: - **NVIDIA Blackwell (B200/B300):** Next-gen AI training clusters. - **NVIDIA H200/H100:** High-availability instances for large-scale inference. - **GB300 NVL72:** One of the first GB300 NVL72 providers in Europe [Current GPU availability and pricing](https://verda.com/products) ## Docs - [Overview](https://docs.verda.com/): Getting started with Verda, navigation of the platform - [Set Up a GPU Instance](https://docs.verda.com/cpu-and-gpu-instances/set-up-a-gpu-instance): Launch a CPU or GPU instance with your chosen OS image and SSH key - [Managing SSH Keys](https://docs.verda.com/cpu-and-gpu-instances/creating-an-ssh-key): Create and manage SSH keys for instance access - [Connecting to Your Server](https://docs.verda.com/cpu-and-gpu-instances/connecting-to-your-datacrunch.io-server): SSH, JupyterLab, and VS Code remote connection methods - [Securing Your Instance](https://docs.verda.com/cpu-and-gpu-instances/securing-your-instance): Security best practices for GPU instances - [Shutdown and Delete](https://docs.verda.com/cpu-and-gpu-instances/shutdown-hibernate-and-delete): Instance lifecycle — shutdown, hibernate, and delete operations ## Clusters - [Instant Clusters](https://docs.verda.com/clusters/instant-clusters): Deploy multi-node GPU clusters with InfiniBand interconnect for distributed training and inference - [Customized GPU Clusters](https://docs.verda.com/clusters/set-up-a-customized-gpu-cluster): Request tailored cluster configurations for large-scale workloads ## Storage - [Block Volumes](https://docs.verda.com/storage/block-volumes): Persistent block storage that can be attached to instances - [Shared Filesystems (SFS)](https://docs.verda.com/storage/shared-filesystems-sfs): High-throughput parallel shared storage for clusters (virtiofs for clusters/NFS for instances) - [Deleting Storage](https://docs.verda.com/storage/deleting-storage): How to remove block volumes and shared filesystems ## Serverless Containers - [Containers Overview](https://docs.verda.com/containers/overview): Scale-to-zero GPU containers for inference workloads - [Container Registries](https://docs.verda.com/containers/container-registries): Connect private container registries - [Scaling and Health Checks](https://docs.verda.com/containers/scaling): Autoscaling configuration, health check endpoints, and cold start behavior - [Batching and Streaming](https://docs.verda.com/containers/batching-and-streaming): Request batching and streaming response support - [Async Inference](https://docs.verda.com/containers/synchronous-and-asynchronous-inference): Synchronous vs. asynchronous inference patterns - [Container Storage](https://docs.verda.com/containers/storage): Persistent and ephemeral storage options for containers - [Batch Jobs](https://docs.verda.com/containers/batch-jobs): Run batch processing jobs on GPU containers ## Managed Inference - [Inference Overview](https://docs.verda.com/inference/overview): OpenAI-compatible API endpoints for hosted models (Llama, Mistral, DeepSeek, and others) - [Getting Started](https://docs.verda.com/inference/getting-started): Quick start guide for managed inference - [Authorization](https://docs.verda.com/inference/authorization): API key authentication for inference endpoints - [Image Models](https://docs.verda.com/inference/image-models): Image generation model endpoints - [Audio Models](https://docs.verda.com/inference/audio-models): Audio model endpoints - [Inference Pricing](https://docs.verda.com/inference/pricing-and-billing): Token-based pricing for managed inference ## Infrastructure as Code - [Terraform Provider](https://docs.verda.com/infrastructure-as-code/terraform): Manage Verda resources with Terraform ## Platform - [Locations and Sustainability](https://docs.verda.com/welcome-to-verda/locations-and-sustainability): Data center locations and renewable energy details - [Pricing and Billing](https://docs.verda.com/welcome-to-verda/pricing-and-billing): Billing model, payment methods, and free egress policy - [Dynamic Pricing](https://docs.verda.com/welcome-to-verda/dynamic-pricing-update): How dynamic pricing works for GPU instances - [Team Projects](https://docs.verda.com/welcome-to-verda/team-projects): Multi-user project management and access control - [Support](https://docs.verda.com/welcome-to-verda/contact-support): How to reach Verda support - [Shared Responsibility Model](https://docs.verda.com/resources/shared-responsibility-model): Security responsibilities between Verda and the customer ## API and SDKs - [API schema](https://api.verda.com/v1/openapi.json): OpenAPI 3.1 compatible JSON schema - [API Reference](https://api.verda.com/v1/docs): Full interactive API documentation (OpenAPI/Scalar) - [Python SDK](https://github.com/verda-cloud/sdk-python): Official Python SDK for the Verda API - [Go SDK](https://github.com/verda-cloud/verdacloud-sdk-go): Official Go SDK for the Verda API ## Console - [Verda Console](https://console.verda.com/): Web UI to manage instances, clusters, storage, containers, and billing ## FAQ **Q: What hardware is available on Verda?** A: We offer the latest NVIDIA B300 SXM6 and B200 as well as H200, H100, and many more. The B200 node delivers 15x faster inference and 3x faster training than the H100 generation. **Q: What is the average latency for Verda Serverless Containers?** A: We achieve an average latency of <300ms. Optimized storage management allows for 30-50% faster startup times compared to legacy providers. **Q: Can I automate my infrastructure?** A: Yes. Verda provides a native Terraform provider, a Python SDK, and direct API access. **Q: What are your data transfer fees?** A: Verda offers free and unlimited egress beyond the standard allowance unlike many general-purpose clouds.