The HCI platform that runs as firmware, not a license you rent by the core.
Storage, compute, networking, and disaster recovery ship as one signed image on commodity hardware. Bare metal to a self-healing HA cluster in minutes, one flat price per server, every feature included.
Fabric is in closed beta today. General availability is planned for Q1 2027.
▶ operation firmware.rollout → 0.9.1 4 nodes · staged in parallel · verified ✓ node-201 parked · 6 VMs live-migrated · flashed · rebooted · healthy 2m41s ✓ node-202 parked · 4 VMs live-migrated · flashed · rebooted · healthy 2m38s ✓ node-203 parked · 5 VMs live-migrated · flashed · rebooted · healthy 2m52s ✓ node-204 parked · 3 VMs live-migrated · flashed · rebooted · healthy 2m35s ✓ fleet on 0.9.1 · 0 s guest downtime · automatic rollback armed on every node
Broadcom broke the model. The market is moving.
Broadcom's acquisition of VMware ended perpetual licensing, collapsed ~168 products into ~4 per-core bundles, and drove renewal increases that sent buyers looking for the exit.
Renewal price increases reported after the Broadcom transition.
Products collapsed into a handful of forced per-core bundles.
Gartner-projected enterprises piloting VMware alternatives, 2025 → 2028.
The market splits in two: cheap but operationally demanding (Proxmox, OpenStack) or turnkey but premium and locked-in (Nutanix, VMware). Almost nobody delivers turnkey and enterprise-grade without the lock-in. Fabric is built for that gap.
Four things, no one else does all four.
Priced per node, not per core
Meter your virtualization per CPU core and denser servers cost you more every year. Fabric prices a physical server by its class — five wide bands, set once — and inside a band the cores, VMs and terabytes are free. Fill the box inside your class. The price doesn't move.
Firmware, not a suite
One signed, A/B-updated image contains the hypervisor, distributed storage, networking, and control plane. One OTA update rolls the entire stack atomically, with automatic rollback, no vCenter, no lifecycle manager, no separate SDDC servers to version and reconcile.
Multitenancy is the product, not an add-on
Tenants, quotas, isolated per-tenant networks with their own DHCP and DNS, and a self-service portal with browser consoles ship in every class, including the free one. No cloud-director product to buy, no second licence family. Bill your tenants, not your cores.
Sovereign by design
Bring your own hardware and run with no internet connection at all. Sign-in is local, the control plane is your own appliance, and updates arrive as signed images you import and verify on site. Nothing in the platform ever needs to reach us.
One node, or five hundred. Same firmware.
Start with a single box at the edge and grow to a multi-rack datacenter without re-architecting. Watch a cluster scale from 1 to 500 nodes, and fail over from edge to datacenter, live.
The whole stack, one signed image.
Platform
Compute, storage, networking, DR, ops, and security, all in the same firmware. No add-on bundles.
See capabilities →Use cases
Edge sites, sovereign on-prem, AI/GPU, VMware migration, and service providers, built for the gap.
See use cases →Pricing
Per node, every feature included, from $500/yr. Free to start. No per-core meter, no feature matrix, no renewal shock.
See pricing →Bare metal to HA cluster in minutes.
Appliance simplicity. No Broadcom tax. Flash a node, run cluster create, and you're
live.
