Enakta Labs builds enterprise storage on DAOS — the open-source engine that dominates every IO500 benchmark. From media & entertainment post-production to GPU-accelerated AI inference, from HPC simulations to general enterprise file services — we make DAOS usable, manageable, and production-ready for enterprise workloads.
DAOS (Distributed Asynchronous Object Storage) dominates the IO500 global benchmark. The Enakta Storage Platform makes it enterprise-ready — with end-to-end orchestration, familiar protocols like SMB and NFS, and production-hardened operations on commodity hardware. Whether you're editing 8K timelines, training models, or running general file services at scale.
Deploy, monitor, and recover storage clusters from bare metal. PXE boot with immutable OS images, automated node replacement, and cluster-wide namespace management.
270+ GB/s from a single client over RDMA. User-space I/O bypasses the kernel entirely. Proven scale to thousands of nodes and multiple 10s of TB/s aggregate throughput.
Failed node rebuilt in under 10 minutes — before customers notice the alert. No 18-hour fsck. No over-provisioning to hit performance targets.
Full SMB support for Windows and macOS clients, with NFSv4.x coming soon. S3 interface in tech preview. Developed in partnership with Tuxera — enterprise file access without compromise.
Direct PyTorch integration for datasets and checkpointing — developed for Google, open-sourced for everyone. Native support for AI training, inference, and HPC simulation workloads.
NVMe SSDs, Ethernet, standard Linux. No exotic memory fabrics or proprietary hardware. Runs on industry-standard servers and commodity infrastructure.
Volumetric capture, non-linear editing, 8K post-production. SMB access for Windows and macOS editing suites. 1PB+ reference architecture with Kioxia ready for production use.
GPU checkpoint storage, dataset serving, and model weight distribution at speeds that keep NVIDIA and AMD clusters fully utilised instead of waiting on I/O.
The same DAOS engine powering the world's top supercomputers at Argonne and LRZ. Simulation output, scientific datasets, and parallel workloads at exascale.
General-purpose high-performance file sharing via SMB and NFSv4.x (coming soon). Replace ageing NAS infrastructure with a platform that scales linearly.
Your hardware. Your data centre. Your jurisdiction. Full data residency control for government, financial services, healthcare, and sovereign cloud deployments.
White-label and OEM options for cloud providers and neoclouds. Simple node-based licensing. Perpetual or subscription models for CSPs and MSPs.
SMB and NFS protocol integration for enterprise file access.
NVMe SSD validation and reference architecture collaboration.
Hardware reference architecture for production deployments.
Independent global benchmark — top 3 in 10-Node Production.
The Enakta Labs AI Platform extends our storage foundation with bare-metal GPU provisioning and sub-second model activation — purpose-built for NVIDIA and AMD inference clusters with zero cloud dependency.
The AI Platform provisions hundreds of GPU nodes in under an hour via PXE boot with immutable OS images. Stateless design means any node can be replaced without cluster reconfiguration.
FlashActivate enables sub-second model swaps through atomic pointer operations in the Enakta Storage layer — no file copies, no engine restarts, no downtime. 20x+ faster first-token latency versus conventional model reloads.
Canary rollouts, blue/green deploys with automatic rollback, cross-pool failover, and session affinity preservation. Compatible with vLLM, TGI, and TRT-LLM out of the box.
Your hardware. Your data. Your rules. Full data residency control for regulated industries, government, and sovereign cloud deployments. Zero cloud lock-in.
Enakta Labs was founded in 2023 with a clear mission: make the highest-performance open source storage technology usable for production workloads. We are a founding member of the DAOS Foundation alongside Argonne National Laboratory, Google Cloud, Hewlett Packard Enterprise, and Intel.
Our team has been working on DAOS since 2020 and brings deep expertise in distributed systems, high-performance computing, and enterprise storage — with years of hands-on experience deploying and scaling storage platforms for some of the world's largest neoclouds and supercomputing centres.
We're active contributors to the DAOS core and work extensively on enabling DAOS outside of Aurora, the world's largest DAOS installation. We don't do 400-page contracts, dishonest roadmaps, or 'professional services' tax. If you have a support question at 3am, we pick up the phone.
Every design decision optimises for speed. User-space I/O. Zero unnecessary abstraction. NVMe as fast as possible.
Built on open-source DAOS. No vendor lock-in. No proprietary traps. Hardware vendor neutral.
Proven from 10 nodes to 1,000+. The same engine that runs exascale supercomputers at Argonne.
Engineer-level support from the people who build the platform. Simple node-based licensing. No games.
How the Enakta Storage Platform achieved IO500 10-Node Production #3 with Core42 on TCP — and what it means for AI infrastructure.
We've submitted our first IO500 result with Core42. Two years of building, one week to prove it.
New reference architecture whitepaper combining KIOXIA CD8P-R NVMe with the Enakta Storage Platform for record-shattering performance.
Significant improvements in I/O performance, memory utilisation, error detection, and simplified configuration.
The DAOS Foundation launches under the Linux Foundation to accelerate open source storage innovation and community collaboration.
Insights from the IT Press Tour on orchestration, offline recovery, and GPU idle time reduction at scale.
In-depth coverage of the platform v1.3, SMB/S3 integration with Tuxera, and network-speed node recovery.
DAOS holds top IO500 positions. Enakta Labs' mission to bring supercomputer-grade storage to enterprise AI.
DAOS dominates across all IO500 categories. Core42 enters at #3 in 10-Node Production running DAOS.
Enakta commercialising DAOS for neo-clouds and sovereign AI, benchmarking on a half-billion dollar GPU cluster.
Argonne, Enakta Labs, Google Cloud, HPE, and Intel come together to advance next-generation HPC and AI storage.
Whether you're editing 8K timelines, training AI models, running HPC simulations, or replacing ageing NAS — if performance and simplicity matter, we should talk.