Why the Control Plane Is Separate
Teams that depend on a single hardware provider are exposed when that provider changes. QCOS is designed with a control plane that is separate from the hardware backend: you describe the circuit once and name the backend for each job. The backend proven today is a CPU state-vector simulator.
The Vendor Lock-In Problem
Today's quantum landscape is fragmented across multiple competing technologies:
Superconducting
Superconducting QPUs
Fast gates, but requires extreme cooling
Trapped Ions
Trapped-ion QPUs
High fidelity, but slower operations
Neutral Atoms
Neutral-atom QPUs
Scalable arrays, native multi-qubit gates
Each technology has different strengths for different use cases. Committing to a single vendor means missing out on the optimal platform for each workload — and risking obsolescence as the technology evolves.
Key Findings
Backend-Independent Circuits
A circuit is described once and submitted to the backend you name for each job. The backend proven today is a CPU state-vector simulator (up to 24 qubits).
Informed Backend Choice
QCOS publishes each backend's capabilities and estimates a job's cost before it runs, so teams choose the processor with the facts in front of them. Automatic selection is not offered yet.
Unified Governance
Manage access, plan limits and credit balance from one interface.
Market Flexibility
Keeping the control plane separate from the hardware backend is intended to limit lock-in to a single provider.
QCOS Architecture
QCOS implements a three-layer abstraction that separates concerns:
Intent Layer
High-level algorithm specification, independent of hardware
Orchestration Layer
Compilation and optimization (the operator selects the backend)
Execution Layer
Submission to the backend you name and retrieval of results
Choosing a Backend
QCOS does not route jobs automatically: you name the backend for every job. What it gives you to make that choice:
- •Capability contracts — Each backend's capabilities in one machine-readable format
- •Cost estimates — A dry run returns the estimated cost before anything executes
- •Data sovereignty — Where a job runs stays your decision, because you name the backend for every job
Automatic backend selection is not offered yet.
Practical Takeaways
For CTOs
Keep the control plane separate from the hardware backend, so that a change of hardware provider does not force a rewrite of your software layer.
For IT Managers
Use QCOS to manage access, plan limits and credit balance for your quantum projects from one interface.