UltiHash offers high‑throughput, S3‑compatible object storage designed for AI and analytics workloads, delivering up to 1 GB/s read per node to keep GPUs and TPUs fed. It provides both a fully managed Serverless service hosted in EU Tier 1 data centers and a self‑hosted, Kubernetes‑native deployment with binary‑level deduplication, erasure coding, and enterprise security controls.
Funding
$2.5M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
4OFounders
Product
Problem
AI and analytics workloads often stall because traditional object storage cannot deliver the high throughput and low latency required for large model training, inference, and retrieval‑augmented generation, leading to idle GPUs and inflated compute costs.
Solution
UltiHash provides high‑performance, S3‑compatible object storage optimized for AI pipelines. It offers a flash‑optimized architecture delivering up to 1 GB/s read per machine, binary‑level deduplication to cut storage costs, and built‑in erasure coding for data resiliency. Customers can choose a fully managed Serverless service hosted in EU Tier 1 data centers or a self‑hosted Kubernetes‑native deployment on any cloud or on‑prem environment, both with policy‑based access control, versioning, and encryption. The storage integrates via a standard S3 API, allowing seamless connection to frameworks such as PyTorch, Spark, vector databases, and data lakehouse engines, while maintaining consistent throughput for bursty reads and writes.
Target Audience
Primary customers are AI/ML engineering teams, data science platforms, and analytics groups that require fast, scalable object storage for model training, inference, RAG pipelines, and data lakehouse workloads, whether in cloud, hybrid, or on‑prem environments.
Features
- Flash‑optimized, high‑throughput architecture (up to 1 GB/s read per node) for GPU/TPU‑bound AI workloads
- Binary‑level deduplication that removes duplicate byte strings across versions and datasets
- Reed‑Solomon erasure coding and object locking for robust data resiliency and compliance
- Fully S3‑compatible REST API with native support for PyTorch, Spark, Trino, Neo4j, vector databases, and other AI tools
- Kubernetes‑native deployment model with CSI driver for self‑hosted clusters on any cloud or on‑premise
- EU‑based, SOC 2‑certified, GDPR‑compliant data residency and security controls (policy‑based access, encryption, versioning)
- Pay‑as‑you‑go and subscription pricing, with instant scaling from 100 TB to petabyte‑scale storage