PROJECTS APPROACH FOUNDER INSIGHTS CONTACT
Quantum systems

Qontos

A modular quantum computer linked by light.

Project source ↗
DomainQuantum systems
FocusDistributed fault-tolerant quantum architecture
Research stagePre-G1 design and simulation
Research thesis

Fault tolerance at datacentre scale may require better interconnects, not simply larger refrigerators.

QONTOS-1 is specified as a 1,000-physical-qubit architecture formed from two superconducting modules connected through a heralded 1,550-nanometre photonic link.

The machine is an architecture-validation target, not a completed commercial computer. Its public specifications define measurable gates for devices, control, interconnect, software, and error correction; later generations remain conditional scenarios or research studies.

1,000Physical-qubit target

Two modules, five chiplets each

20 mKBase-temperature target

Tantalum-on-silicon transmons

1,550 nmOptical link

Heralded microwave-to-optical interconnect

04Acceptance gates

Architecture through logical-qubit validation

The scaling problem

A monolithic machine eventually collides with wiring, cooling, control, and fabrication yield.

Qontos treats distributed operation as a first-class system problem: quantum modules, an entanglement link, and a real-time classical control loop must behave as one machine.

Superconducting processors operate at millikelvin temperatures while control electronics, decoders, and optical interfaces live across very different thermal and timing domains. Adding qubits without redesigning those interfaces increases integration pressure faster than useful computation.

The proposed alternative is modular. Each cryogenic unit can be engineered and accepted independently, while a probabilistic photonic channel supplies cross-module entanglement. This makes link rate, fidelity, latency, decoding, and scheduling part of the computer’s architecture rather than peripheral networking.

Cryogenic engineers tracing dense microwave wiring inside an open dilution refrigerator during integration testing
Research visualA monolithic architecture concentrates wiring, thermal, control, and serviceability pressure inside one cryogenic boundary.
System architecture

Three engineering domains, coordinated by one runtime.

Superconducting compute

Two modules are each specified with five 100-qubit tantalum-on-silicon transmon chiplets in a heavy-hex topology, with tunable-coupler CZ gates and a target base temperature of 20 mK or below.

Photonic interconnect

An electro-optic lithium-niobate resonator is intended to convert microwave photons toward 1,550 nm; a detector coincidence heralds a Bell pair between modules so failed attempts need not corrupt logical state.

Real-time control

Nanosecond-resolution waveform generation, FPGA sequencing, syndrome ingest, MWPM decoding, and feed-forward share a timing budget with local gates and the slower probabilistic link.

QONTOS-1 specification

Engineering targets, not measured achievements.

Coherence
T1 ≥ 200 µs

Target for the superconducting device layer.

Gate error
1Q 1×10⁻⁴ · 2Q 5×10⁻³

Specified acceptance objectives.

Readout
99% fidelity target

With a one-microsecond stabiliser cycle objective.

Link
η ≥ 0.1% · Fraw ≥ 0.85

Base transduction and raw Bell-pair fidelity targets.

Latency
≤ 25 µs end to end

Photonic-link objective over a five-metre reference path.

Feed-forward
≤ 10 µs

Target after syndrome processing in the real-time loop.

Conceptual cutaway of two superconducting quantum modules connected by a central photonic interconnect and real-time control electronics
Conceptual engineering visualConceptual engineering visualization of the QONTOS-1 target boundary: two independent cryogenic modules, a heralded optical link, and coordinated classical control. Targets shown elsewhere are not measured achievements.
Software platform

Compile once, route across heterogeneous backends, preserve a proof chain.

The runtime is the currently usable part of the programme; native QONTOS hardware is intended to join the same contract at gate G2.

  1. Ingest and normalise

    Accept circuits from OpenQASM 3 and common Python quantum frameworks, then normalise them into a shared intermediate plan.

  2. Partition and schedule

    Separate work by spectral and hardware constraints, score candidate backends for fidelity, queue, and cost, and bind partitions through an ExecutorContract.

  3. Execute concurrently

    Dispatch to simulation, IBM Quantum, or Amazon Braket today; the proposed native QONTOS executor enters after first-module acceptance.

  4. Aggregate and prove

    Reassemble results while retaining a three-layer SHA-256 record intended to support later replay and verification.

Five-generation research arc

Every generation is gated by evidence from the one before it.

  1. Target
    QONTOS-1

    Architecture validation

    Two modules, approximately 10³ physical qubits, and a first distance-five logical-qubit objective.

  2. Scenario
    QONTOS-2

    Logical-qubit advantage

    Four to eight modules and tens of logical qubits, conditional on device and link performance.

  3. Scenario
    QONTOS-3

    First useful FTQC pilot

    Sixteen to thirty-two modules and a proposed 50–100 logical-qubit range.

  4. Research
    QONTOS-4

    Larger fault-tolerant system

    A research vision spanning dozens of modules and hundreds of logical qubits.

  5. Study
    QONTOS-5

    Datacentre-scale architecture

    A long-horizon study rather than a product commitment or dated delivery plan.