Genesis operated a public cloud from 2018 and served more than 20,000 users on it. Our own sites were tenant zero for every release, and we carried the on-call rota. What you license is the software that came out of those eight years — not a design document, and not a pilot.
These are Genesis' own operating figures from the cloud business we ran. In those eight years the stack was rewritten from the ground up twice. You license the second rewrite, and every claim further down this page names the record it comes from.
the year Genesis began operating cloud infrastructure itself
users served on the platform Genesis ran
B2B customers since 2023, around 160 of them recurring
Genesis ran it from 2018 and served more than 20,000 users on it, with 5,000+ B2B customers since 2023 and around 160 of them recurring. Our own sites were tenant zero for every release and we carried the on-call rota. That business is where the software comes from. Genesis today is a software company: what you license is the stack itself — not capacity, not hosting, not services wrapped around it.
.svg){kind=small}
Because the layer that decides whether installed capacity earns anything is the layer we had already written for ourselves. Long-term contracts give you stable operations and something a lender can underwrite, while the short-term on-demand market pays materially higher average prices per hour, and one stack can serve both. On the estate we ran ourselves, a third of installed capacity had no contract against it, and more than half of that went under a paying tenant — measured on GPU-hours over the first half of 2025, our own record rather than an industry figure.
You do. Volta licensed Genesis Grid and runs its own platform on clusters Volta operates itself. Genesis engineers the software and ships releases; we do not hold your keys, your tenants or your on-call phone. You still need an operations team, but a smaller and less specialised one than a platform-engineering organisation — the way VMware let you run virtualisation without building it. A platform upgrade costs your control plane, portal and metering under 15 minutes; running instances are untouched.
Then you are in the range the second rewrite was built for. Multi-tenancy is scoped to the compute cluster: tenants are separated inside one cluster, and if you operate several sites you operate several clusters under the same control-plane software and the same meter. Standing up a further cluster takes 2 days, and from signed contract to your first tenant invoice is two weeks.
Only up to a point, and the point is worth naming. Uptime Institute's Tier I to IV describe topology — redundancy and what can be maintained while the site keeps running — not availability percentages, which Uptime stopped publishing years ago. Software resilience shifts how much redundancy the building has to carry for a given class of tenant; it does not replace it. A deployment is one cluster in one building, and if that building loses power the cluster goes down. Sell availability that depends on maintaining while running, and you need a facility that delivers it.
