PEAKAIO

AI-optimized data platform, purpose-built for GPU workloads. Used by @KingsCollegeLon, Los Alamos National Lab, @ZSLLondon & NHS AI trials.

[PEAKAIO @GETPEAKAIO]

PEAK:AIO has agreed to be acquired by NetApp.

Meanwhile the AI work continues. We're at SNIA SDC next week presenting pNFS Lattice with Los Alamos National Laboratory, and hosting the pNFS Birds of a Feather on Monday evening.

We'd love to see you there.

Full announcement: https://t.co/8G6IKLEzgm

[PEAKAIO @GETPEAKAIO]

Today we're announcing Lattice: the first open-source pNFS metadata server.

AI has a problem nobody likes to talk about. Across 23,000 production clusters, average GPU utilization is 5%, not because the hardware is weak, but because the software feeding it can't keep up.

Lattice fixes that.

Built with Los Alamos National Laboratory, launched under the Linux Foundation, Open source and built on the pNFS standard that ships in every Linux kernel.

We're presenting Lattice this week at MSST 2026 in Santa Clara.

Free to use. Free to build on. Free from lock-in.

Read about it: https://t.co/oQCeNRVLr9
See it in action: https://t.co/9dAh3KIdiu
Build on it: https://t.co/TJGZWRU76H

#AI #HPC #OpenSource #Storage #pNFS #LinuxFoundation #PEAKAIO

[PEAKAIO @GETPEAKAIO]

We wrote about why playing it safe in public sector storage often isn't safe at all.

End-of-life systems. Extended contracts. Known vulnerabilities with no patch.

And why the AI capabilities public sector will need in 2027 depend on decisions being made right now.

https://t.co/fT7J3JCKnw

#PublicSector #NHS #StorageInfrastructure

[PEAKAIO @GETPEAKAIO]

https://t.co/UcteUMjTBZ

[PEAKAIO @GETPEAKAIO]

We took a server at Los Alamos from 70 GB/s to 400 GB/s. Same hardware. Just software.

At a time of memory crunches and SSD shortages, that matters.

Full story in @BlocksAndFiles today. Link below.

#AIStorage #PEAKAIO

[PEAKAIO @GETPEAKAIO]

🏆 PEAK:AIO named to @CRN's 2026 Storage 100

Built for the 80% of AI workloads that can't justify $400-650K minimums:
→ Start at 1 server, scale to thousands
→ Deploy in hours (not weeks)
→ 90-95% GPU utilization from day one

Inference is 60-70% of compute. The majority deserves infrastructure too.

Thank you @CRN

#Storage100

[PEAKAIO @GETPEAKAIO]

Jensen named storage as the AI factory pillar nobody had solved.

Then announced a solution built for gigawatt data centres.

For the hospital running AI on 6 DGX systems, the research lab on 2 GPU servers — the gap is still open.

That's exactly who we built PEAK:AIO for.

Full piece 👇
https://t.co/n8TSXjGMBw

[PEAKAIO @GETPEAKAIO]

Yesterday, Jensen named the five pillars of the AI factory: compute, memory, storage, networking and security.

Most organizations building AI infrastructure have a plan for four of them.

Storage is the one that gets specified last, bought cheapest, and blamed first when the GPU cluster doesn't perform as expected.

The AI factory narrative is right.

The storage conversation that needs to follow it mostly isn't happening yet.

That's the gap PEAK:AIO exists to close.

#GTC2026 #AIFactory #AIInfrastructure #PEAKAIO