Today, we’re really excited to be introducing you to UltiHash Serverless - high-performance storage for AI that hosts your data exclusively on servers in the EU.

It’s the same high-throughput engine behind our self-hosted version, now available serverless. No hardware to manage. No YAML to get lost in. No setup getting in the way of your quick turnaround. Just scalable and fast storage that is ready for you and plugs directly into your AI stack.

✅ Your data is located in the EU, in datacenters in Frankfurt am Main, Germany
✅ Up to 10x higher throughput for data access
✅ Single-digit millisecond latency
✅ Preconfigured sizes for every purpose, from 200GB - 10 TB

UltiHash Serverless is an object storage, designed to store text, images, audio, embeddings, logs, and built to handle the throughput modern AI workloads demand. It also supports open table formats like Iceberg, Delta, and Hudi, so you can query, version, and manage structured data without changing your stack. Whether you're training models, running vector search, or scaling GenAI systems, storage performance is no longer the bottleneck. And all of that, while keeping your data safely located in the EU.

Why storage matters more than you think

Storage is usually the most boring part of your stack, until it becomes the thing slowing everything down: your GPUs are waiting, retrieval lags, pipelines stall. You debug your code for hours before realizing it’s your storage I/O.

UltiHash Serverless is built to make that go away. It’s fast, serverless object storage designed for AI. No infra, no setup, just spin it up and go. Performance is up to 10x faster than standard object storage solutions from hyperscale cloud providers, and it runs in Frankfurt am Main - Germany, keeps your data in the EU.

It offers an S3-compatible API, so you can plug it into whatever you’re already using: PySpark, Iceberg, Airflow, AWS Glue, Kafka, Neo4j, you name it. UltiHash Serverless can stream multi-terabyte datasets into training jobs, fetch documents on demand during retrieval, and store outputs and logs without slowing anything down. And with built-in observability via OpenTelemetry, you can track performance in tools like Prometheus or Datadog, so you always know how storage is impacting your pipeline.

This isn’t just fast storage: it’s storage that fits into your stack, scales with your workload, and gives you visibility as you build.

UltiHash Serverless for GenAI

In GenAI systems, object storage sits behind almost everything: retrieval pipelines, vector databases, fine-tuning workflows, and post-processing outputs. Whether you're building with text, images, audio, or structured tabular data, the speed at which that data can be read and written directly shapes latency, cost, and overall system performance.

UltiHash Serverless provides the high-throughput storage layer needed to keep those systems running smoothly. It’s designed to store everything from raw documents and embeddings to generated outputs and training artifacts—supporting fast access without forcing everything into memory or introducing lag across the pipeline.

With storage located in Frankfurt am Main and optimized for high-throughput reads, UltiHash Serverless supports low-latency pipelines across modalities. For teams running close to our datacenter, we’re happy to help fine-tune infrastructure for even better performance. No special provisioning. No tuning. Just clean integration into their existing stack, and peace of mind around data locality and compliance.

In a GenAI stack, storage is more than a place to put files, it’s part of the performance path. UltiHash Serverless makes sure it doesn’t fall behind.

Get started with a 7-day free trial

Now, all you’ve got to do is get started - and we make it super easy with a 7-day free trial. All you need to do is create an account, tell us to spin up a cluster, and you can start building your AI pipelines in minutes.