Compute where you want. Our cloud runs on every cloud.
Your work runs on ZB Cloud. ZB Cloud puts each job on the cloud that should run it, chosen by price, capacity and location. When one of those changes, the work moves.
Work from running the store passes through ZB Cloud, which weighs price, capacity and location and routes each job along a dotted line to one of four clouds: Hyperscaler, everywhere at list price; Sovereign cloud, data stays inside the border; Neo Centers, GPU capacity by the hour; Decentralized Clouds, thousands of independent machines at a fraction of the price. Each job runs where it lands. Patient records always go to the Sovereign cloud. The nightly batch first runs on the Hyperscaler; when the price changes, ZB Cloud moves it to the Decentralized Clouds.
Every cloud is best at something. ZB Cloud uses all of them.
Each kind of cloud wins at a different job. ZB Cloud treats all of them as one pool and picks from it for every job.
Hyperscaler
The big global clouds. Everywhere, at list price.
Every region and every service, on demand. Where the store, the database and the frontier model calls run when nothing says otherwise.
Sovereign cloud
Data stays inside the border.
Compute inside the country the data belongs to. Patient records, government work, and anything the law keeps at home.
Neo Centers
GPU capacity, by the hour.
Specialist GPU clouds. Where open models, image generation and long builds get their compute at the hour's best price.
Decentralized Clouds
Thousands of independent machines, a fraction of the price.
Backups, nightly batches and any job that can run anywhere, when the price says so.
Price. Capacity. Location. ZB Cloud decides every time.
Every workload on ZB Cloud passes through the Cloud Router. It weighs three things, sends the work to the cloud that wins, and keeps watching.
Input 01
Price
What each cloud charges for this job right now. The cheapest cloud that meets the other two wins.
Input 02
Capacity
Who has the machines free today. GPUs, memory and bandwidth where the job needs them, the moment it needs them.
Input 03
Location
Where the data may live and where the users are. A workload that must stay inside a border stays there.
When one of them changes, the work moves.
A clinic's records stay on the sovereign cloud of the country they belong to. A nightly batch runs wherever it is cheapest tonight. The store runs wherever is cheapest and keeps its data where it belongs. You never learn a cloud's name or region unless you want to.
Everything a business runs on, and everything an AI runs on.
From the website to the model behind it. ZB Cloud runs each kind of work with the services that kind needs.
Hosting
The store, the app, the database and the queues, live and backed up.
- Web and API hosting
- Databases and queues
- Object storage
Inference (running models)
Model calls served fast, at any volume, on whichever cloud has the capacity.
- High-throughput serving
- Low latency
- Auto-scaling
Training and fine-tuning
Teach a model your language, your data or your trade, and keep teaching it.
- Model adaptation
- Fine-tuning on your data
- Continuous learning
GPU scheduling
Many jobs, many tenants, one fair queue for the GPUs.
- Scheduling
- Multi-tenant workloads
- Fair allocation
Batch and background jobs
Nightly runs, reports, backups and long builds, on the cheapest machines open tonight.
- Nightly batch
- Backups
- Long builds
Build, release and monitoring
Every change built, tested, released and watched, with the way back ready on every release.
- Build and release pipelines
- Logs and metrics
- Monitoring and rollback
The system operates itself.
Deployed, monitored, backed up. ZB Cloud carries the scaling, the networking, the backups and the rollback. Your AI team builds; ZB Cloud runs.
Releases
Every change goes live through a pipeline. A release that fails its check rolls back on its own.
Scaling
More customers, more machines. Fewer customers, fewer machines. The bill follows the traffic.
Resilience
Backups by default and a second region ready. When a cloud has trouble, the work moves to one that does not.
The record
Every workload in the record with its cloud and what it cost, beside the tasks and the models it ran for.
Sovereign where sovereignty matters. Global where it adds value.
ZB Cloud is the compute layer of a national AI stack. Sensitive work stays on sovereign infrastructure inside the country. Everything else runs on the best cloud in the world for that job, under one policy.
Identity and access
One login, clear roles, each person sees only what they need.
Security and encryption
End to end, with key management.
Data residency
Data kept in the country, per policy.
Governance and policy
Rules for models, usage and data access.
Budget and token allocation
Quotas and cost limits, charged back to each ministry or team.
Audit and compliance
Full audit trails.
Model safety
Safety filters and risk detection.
Logging and reporting
Central logs, monitoring, incidents.
One of three parts. The one everything runs on.
AI teams decide what to build and build it. The Model Router sends every task to the best model. ZB Cloud runs all of it, on the cloud that should run it. Build, operate, and scale AI your way.
- Who does the workAI teamsPut AI teammates to work.
- Which modelModel RouterUse the best model for every task.
- Where it runsZB CloudCompute where you want.
A few things people ask.
What is ZB Cloud?
Our cloud runs on every cloud: the big global clouds, sovereign clouds, specialist GPU clouds and decentralized clouds. The work runs on ZB Cloud, and ZB Cloud sends each workload to the cloud that should run it, by price, capacity and location. When one of them changes, the work moves.
Which clouds?
Hyperscalers, sovereign clouds, neo centers and decentralized clouds. A workload that must stay inside a border stays there; everything else runs where price, capacity and location say it should.
Can I choose the cloud myself?
Yes, on any workload. Pin it to one cloud or one region and the Cloud Router keeps it there. Most people never do.
Where does my data live?
Where you say it may live. Set a location rule once and every workload that touches that data follows it. Everything else is free to move.
What if a cloud goes down?
The Cloud Router moves the work to the next cloud that meets the same rules. Backups and a second region are there by default.
Who runs it?
ZB Cloud. Releases, scaling, backups, monitoring and rollback are part of it. Zero DevOps.
What does it cost?
Hosting and compute are charged at provider cost and shown in your record beside every task, its model and its cloud. No subscription.
More questions? The full FAQ