AI Cloud
From rapid prototyping to foundation training to scalable inference
Compute
GB300 NVL72
New1x tray to 2+ racks · NVLink v5
Private, protected AI storage, alongside your compute.
Built for the unstructured data behind training and serving.
Create buckets, manage access keys, and see each bucket's size and approximate object count, plus what your storage has cost so far this month, alongside your instances and clusters.
Move large files without starting over. If a connection drops, run the same command again and the Verda CLI picks up where it left off.
Point the S3 tools and SDKs you already use at Verda, and get straight to work.
Share a single file with a presigned link that expires when you choose, while the bucket stays private.
Share datasets across your project, and stage what each job needs onto fast storage.
Save checkpoints as training progresses, so an interrupted job picks up where it left off.
Keep every model version in one library and load the one you need when an instance starts.
Store generated files, evaluation results and logs, ready to share or feed into the next round of training.
Pay for the data you store. Egress, ingress and API requests are free, with no subscription or minimum commitment.
| Storage type | per GiB per month | per TiB per month | Egress, ingress and requests |
|---|---|---|---|
| Object storage | $0.07 | $71.68 | Free |
During beta, Object Storage is available in FIN-03, with default quotas of 1 TiB, 100,000 objects and 100 buckets. Request more anytime from Account settings / Quotas.
Copy your data across with rclone, straight from any S3-compatible storage into your Verda buckets.
They do different jobs. Object storage is where your data lives: datasets, checkpoint history, model versions and backups, reachable with S3 tools. The shared filesystem is where data sits while your jobs use it, mounted across instances. Many teams use both, and stage what a job needs from object storage onto the shared filesystem or NVMe.
Because your jobs run on Verda. With data and compute in the same cloud, jobs don't wait on transfers from another provider, and you manage buckets, instances and clusters in one console. If you ever want to move your data elsewhere, egress is free.
Usually just your connection settings. Verda Object Storage uses the standard S3 API, so the AWS CLI, boto3, rclone, s5cmd and other S3 tools work once they point at Verda. A few S3 features, such as creating buckets through the S3 API, aren't supported yet. The quickstart shows exactly what to set.
No. Egress is free and the API is the S3 standard, so you can move your data whenever you want.
Object Storage is ready to use, with default quotas that will rise as the beta progresses. You can request higher limits anytime, and we'd love your feedback through the chat in the console.
Built in Europe, trusted globally
From rapid prototyping to foundation training and scalable inference — on a single full-stack AI cloud