Every neocloud in the market right now is winning or losing on the same battlefield: GPUs. Availability, performance, price per hour. That’s the game, and it’s the right game to play.
But there’s a problem compounding underneath it, one that doesn’t show up in benchmark reports or investor decks, and one that customers will eventually force platforms to confront—neoclouds need highly available, performant storage so that GPUs are never waiting on data.
Neocloud customers need more than compute
Cloud storage isn’t adjacent to what neoclouds sell—for buyers, it’s the difference between a complete platform and a partial one. Neocloud customers need:
- Somewhere to put a 10PB training dataset before it touches WEKA, VAST, or DDN boxes
- Persistent storage for model checkpoints between training runs
- A place for production inference pipelines to pull weights from at scale, fast enough that GPU nodes aren’t sitting idle waiting on data
Most platforms know this. The ones that are honest about it will also tell what happened when they tried to solve it: Engineering sprints planned around storage tooling that competed directly with GPU roadmap work. Ops overhead for infrastructure that customers take for granted when it’s working and blame the platform for when it’s not. Capital allocated to storage hardware and the teams to manage it.
The platforms that figured this out first didn’t hire their way through the problem. They stopped building storage and started shipping it.
Introducing B2 Neo
That’s the insight behind Backblaze B2 Neo. We built it in direct collaboration with the neocloud operators already utilizing Backlaze in production. The result is a white label cloud storage backbone that neoclouds can launch as a native extension of their platform. Here’s what that means in practice:
- Up to 1Tbps throughput: Storage that keeps GPU clusters fed and AI workflows moving without becoming the bottleneck.
- Launch under your own brand: Branded endpoints, partner-controlled pricing, and a native customer experience that keeps the platform front and center.
- API-driven provisioning: Provision accounts, manage permissions, and handle billing through existing platform tools without a separate console or manual setup.
- 17 years of operational maturity and exabyte scale expertise: Enterprise durability and reliability that customers expect, operated by Backblaze so internal teams don’t have to.
It doesn’t require a storage team. It doesn’t compete with the GPU roadmap. And it doesn’t route customers away from neocloud platforms where their workflows slowly migrate away.
Already in production with leading platforms
Multiple neoclouds are already using B2 Neo. It’s infrastructure that’s already handling AI training workloads, high performance compute (HPC) pipelines, and media delivery at scale and the platforms already running it are direct about why they made the call.
One global edge services platform, after a rigorous technical and business evaluation, put it this way: Their customers were demanding cost-effective yet performant storage as their AI business scaled and Backblaze gave them the ability to deliver cloud object storage as a native extension of their own platform without taking focus away from their roadmap.
Rob Strechay, Principal Analyst at Smuget & theCUBE Research, framed it simply:
With B2 Neo as a first-party service offering to neoclouds, I see the advantage for those organizations of being able to turn on cloud storage without the toil and expense of building it themselves. It is a near-instant value-add offering, helping their customers control costs and achieve the ROI of AI faster.
The question for the rest of the market is how many more quarters of DIY storage ops—or hyperscaler dependency—they can absorb while competitors ship that outcome instead.
B2 Neo is available now. If you’re building a neocloud and storage is either a distraction or a gap, I’d encourage you to talk to us.
