Project Q-Core
Stabilized qubits at higher temperatures
Why this project exists
The dominant cost of running quantum systems at scale is not computation — it is cryogenics. Maintaining millikelvin temperatures across hundreds of thousands of qubits requires infrastructure that scales worse than linearly. Reducing the cryogenic budget by even modest factors changes the economics of quantum coordination.
Q-Core combines topological encoding with learned decoders to operate at higher temperatures than current surface-code stacks would tolerate. We are not claiming room-temperature qubits. We are claiming reduced cryogenic overhead at meaningful coherence-time targets.
How we work on it
Topological encoding under thermal noise
Topological codes localize errors in a geometry-dependent way that gives logical qubits stronger noise tolerance per physical qubit than surface codes in some regimes. We are characterizing the trade frontier between code distance, lattice geometry, and effective temperature.
Learned decoders within tight latency budgets
Off-the-shelf MWPM decoders are too slow for our target operating point. We use neural decoders trained on synthetic syndrome distributions, with formal latency caps and confidence-aware fallback. The decoder runs at the cryostat boundary on dedicated hardware.
Hybrid classical-quantum scheduling
Q-Core is designed to be the substrate for quantum-enhanced sampling and optimization workloads where the classical side dominates orchestration. The scheduling layer treats quantum coherence as a perishable resource and schedules accordingly.
Where the work stands
- ShippedQ3 2025
Decoder evaluation harness
Internal benchmarking suite for neural decoders against synthetic noise.
- In progressQ1 2026
First topological prototype
Bench-top demonstration of the encoding stack at intermediate temperatures.
- PlannedQ4 2026
Sustained logical-qubit operation
Multi-hour logical-qubit operation at the target cryogenic budget.
What we are still figuring out
- 01
Decoder latency vs. accuracy trade
How tight can the decoder latency budget go before logical fidelity collapses? Our current models are characterizing the cliff but not crossing it.
- 02
Code transitions under temperature drift
Real cryostats drift. The encoding choice that is optimal at one temperature may not be at another. Adaptive encoding is an open problem.
- 03
Hybrid scheduling under uncertain coherence
When coherence time itself is a learned estimate, scheduling becomes a planning-under-uncertainty problem. We are still exploring what the right abstraction is.