Heading to SC25 — Two Years of Work, One Week to Prove It

It's the week before Supercomputing 2025 in Atlanta, and honestly, we're nervous.

Not the bad kind of nervous. The kind where you've been building something for two years, you've submitted your first IO500 result with a world-class partner, and now you wait.

What we submitted

We've been working with Core42 — the AI infrastructure arm of Abu Dhabi's G42 group — to validate the Enakta Storage Platform on their Maximus infrastructure. The IO500 submission is in. We can't share numbers yet (the results get announced at the IO500 BoF session during SC25), but we can say this: the entire run was done over standard TCP networking. No RDMA. No special interconnects. Just Ethernet, NVMe, and DAOS.

If the results land where our internal testing suggests, it'll be a statement about what commodity infrastructure can do when the storage software isn't the bottleneck.

How we got here

Two years ago, Enakta Labs was an idea and a bet on DAOS. The bet was simple: the storage engine behind the world's fastest supercomputer — Argonne's Aurora — could be turned into an enterprise product that runs on commodity hardware. Not a science experiment. A product you can actually deploy, manage, and support in production.

That meant building everything around the engine. Bare-metal provisioning that gets a cluster running in under an hour. SMB access so Windows and macOS workstations can use it natively. S3 for object workflows. A management layer that doesn't require a PhD to operate. Failed node recovery in under 10 minutes.

Along the way, we published a reference architecture with Kioxia on enterprise NVMe hardware. We developed the native PyTorch integration for DAOS for Google and open sourced it. We helped establish the DAOS Foundation under the Linux Foundation alongside Argonne, Google Cloud, HPE, and Intel.

It's been a lot. And this week, it all converges.

What we're looking forward to

Beyond the IO500 results, SC25 is where our community comes together. We're looking forward to reconnecting with partners, catching up with the DAOS Foundation members, and meeting the people who are building the next generation of infrastructure — whether that's exascale HPC, sovereign AI, or the media workflows that are quietly generating more data than anyone talks about.

These conferences are also where the real conversations happen. The ones in hallways and over coffee, not on stage. The ones where someone says "we have this problem with our storage" and you realise you've spent two years building exactly the thing they need.

Come find us

We'll be at the show all week. If you're building GPU infrastructure and your storage can't keep up — let's talk. If you're a neocloud looking for a differentiated storage offering — let's talk. If you're running media workflows and 8K timelines are choking your NAS — let's talk. If you just want to geek out about DAOS and user-space I/O — definitely let's talk.

Reach out at contact@enakta.com or find us at the IO500 BoF session where the results get announced.

We'll be the ones trying to look calm.

See you in Atlanta.