Nature News · Feb 11, 2026 · Collected from RSS
MainLarge-scale quantum communication networks promise secure information transfer among numerous clients over long distances2,3. Over the past few decades, notable milestones have been achieved in long-distance QKD in satellite and fibre-optic systems. Progress has been made towards various network models, for example, trusted-node networks5,6,7,8,9, point-to-multipoint networks10,11 and wavelength-division-multiplexing (WDM) entanglement networks12,13. Simultaneously scaling the number of quantum clients N and the communication distance L poses substantial challenges, hence necessitating advancements in protocols, architectures and hardware. Recently, advanced protocols, such as TF-QKD14, have been proposed to extend communication distances, enhancing key rates by the square root of channel transmittance, and experimental results have surpassed the repeaterless bound by distances of up to 1,000 km (refs. 15,16,17,18,19,20,21). Furthermore, TF-QKD operates under a measurement-device-independent (MDI) framework22, enabling untrusted central nodes and sharing of expensive resources among users on the shared central node. This architecture therefore holds the potential for large-scale cost-effective networks with large N and long L. TF-QKD implementation hinges on achieving ultrastable single-photon interference in large optical interferometers, requiring low-noise lasers and high-precision optical phase tracking. Despite notable advancements in protocols, the present implementation of TF-QKD remains primarily limited to point-to-point communications15,16,17,18,19,20,21. Practical implementations of TF-QKD networks face substantial technical challenges. These include development of many ultralow-noise coherent laser sources, massive manufacture of identical, high-performance QKD chips that maintain uniformity across all client nodes and system-level integration that enables scalable quantum node operation and quantum channel transmission, while remaining compatible with existing phase tracking. They collectively represent the fundamental requirements for TF-QKD network infrastructure, demanding simultaneous advances in photonic integration, manufacturing scalability and system control.Here we present a proof-of-principle laboratory demonstration of a user-massively scalable and long-haul TF-QKD network enabled by integrated photonics, named the ‘Weiming Quantum Chip-Network’. Our integrated-photonics QKD network comprises a central server chip that integrates an optical microresonator frequency comb (microcomb) on a silicon nitride (Si3N4) platform and 20 copies of independent client-side QKD transmitter chips that monolithically integrate all essential components for key preparation based on indium phosphide (InP) photonic circuits. These client-side QKD chips fabricated on a 3-inch InP wafer and server-side microcomb chips fabricated on a 4-inch Si3N4 wafer both demonstrated high performance, reproducibility and operational yield, thus providing scalable quantum devices for network implementations. We use integrated microcomb technology, which has revolutionized classical photonics23,24,25, to substantially enhance the scalability and reliability of the TF-QKD network by delivering a large number of ultralow-noise, phase-coherent comb lasers for WDM-based networks across distributed client chips. We implemented pairwise TF-QKD between 20 client chips through ten wavelength-multiplexed channels aligned with microcomb lines, realizing a total network spanning capability NL/2 = 3,700 km (in which N = 20 and L = 370) without any trusted relays.Architecture of integrated-photonics TF-QKD networkThe TF-QKD14 network retains MDI properties22, effectively eliminating detector vulnerabilities using untrusted resources and maintaining the star topology around a centralized server node, and greatly enhances the communication distance15,16,17,18,19,20,21. In typical TF-QKD, frequency references are distributed among user nodes to generate coherent light pulses and secure keys are then derived from single-photon interference measurements at the untrusted server node. Figure 1a shows a WDM-based TF-QKD network enabling parallel key distribution between paired clients with identical wavelengths. Clients in one metropolitan zone connect to intermediate nodes, transmitting WDM weak-coherent pulses through backbone fibres for long-haul transmission between different zones. At the server, light pulses are demultiplexed, interfering at beamsplitters and secure keys are extracted through single-photon detections.Fig. 1: An integrated-photonics TF-QKD network system.a, Schematic diagram for a many-user long-haul TF-QKD WDM-based star network system. This network could support a large number of QKD clients (N) and long client-to-client communication distances (L). It can mitigate detector vulnerabilities at the central (untrusted) server node, while enabling cost-efficient resource sharing among many legitimate QKD users. b, Conceptual diagram of a large-scale integrated-photonics TF-QKD network, ‘Weiming Quantum Chip-Network’, with both large N and long L. QKD client chips integrate local lasers, PMs, IMs and VOAs, delivering high-speed operation, full integration, cost-effectiveness, miniaturization and mass-manufacturability for the TF-QKD network. The server chips encompass an integrated microcomb, as well as linear-optic circuit (LOC; including optical switches, wavelength multiplexers and beamsplitters) modules, single-photon electro-optic frequency shifters (EOFSs), SNSPD modules and electronic logics. The integrated microcomb provides a broad range of ultralow-noise coherent Hz-level comb lines for TF-QKD, enabling WDM-based networks with inherent phase coherence between QKD client chips. Microcomb and client chips are clock synchronized within the network, with the microcomb driving many chip-based transmitters to reliably share quantum-secure keys across long-haul fibre channels. c, Photograph of integrated-photonics quantum chips used in our network, including an integrated Si3N4 microcomb chip (unpackaged) at the centre and 20 copies of identical InP QKD client chips (two InP chips are packaged on one printed circuit board). Both server-side microcomb chips and client-side QKD transmitter chips can be massively manufactured with wafer-scale reproducibility. These 20 InP QKD chips used in this experiment were randomly selected from a set of 24 copies fabricated on a 3-inch wafer, without any preselection.Full size imageWe introduce an integrated-photonics TF-QKD network, as illustrated in Fig. 1b. At the central server node, an integrated optical microcomb serves as the heart, generating a broad range of ultralow-noise coherent dark-pulse frequency combs with linewidths at the Hz level in the telecommunication band. The microcomb can be generated through the self-injection locking of a semiconductor laser to an ultrahigh-quality (Q) Si3N4 microresonator26,27, eliminating the necessity for complex electronic controls and bulky table-top lasers and hence leading to cost-effectiveness, compact footprint and high scalability. Microcomb lines are dispersed across the network by means of the backbone fibres and wavelength demultiplexed at intermediate nodes before reaching the client-side chips. Our InP-based TF-QKD client transmitter chips monolithically integrate all essential components, encompassing lasers, electro-optic (EO) phase modulators (PMs), intensity modulators (IMs) and variable optical attenuators (VOAs). Using the injection-locking technique21,28,29, the original lasers with MHz-level linewidths on QKD chips can be locked to the Hz-linewidth comb lines to align their frequencies and phases, locally regenerating low-noise light fields. Following this, QKD chips implement the encoding and transmit wavelength-multiplexed pulses back to the server node. As shown in Fig. 1c, 20 copies of InP QKD chips are synergistically used in this network. At the server’s detection module, a full integration of quantum devices is possible, encompassing linear-optic devices (for example, switches, wavelength multiplexers and beamsplitters), quantum frequency shifters (see Supplementary Information Section 5.2 for details), superconducting nanowire single-photon detectors (SNSPDs) and fast electronic logics, further bolstering scalability and cost efficiency. In our experiment, the integration of devices other than microcombs on the server is still underway. A Supplementary Video is provided to illustrate the entire procedure of the integrated-photonics TF-QKD network.We emphasize several unique capabilities of integrated-photonics QKD networks. (1) Ultralow-noise coherent dark-pulse microcombs provide ultralow-noise lasers required for TF-QKD and establish a globally stable phase reference for the entire network. (2) Broadband and equally spaced comb lines (with intervals adjustable to International Telecommunication Union standards) can facilitate massive wavelength-multiplexed QKD between a large number of client chips (at present, we use a 30-GHz spacing for the proof-of-concept demonstration). (3) High-yield fabrication and operation of fully functional QKD transmitter chips ensuring high performance, along with the wafer-scale reproducibility of microcomb chips, critically advance the scalability and reliability of quantum networks.Integrated microcombs and QKD client chipsThe advancement of integrated quantum photonic technologies has sparked notable interest in high-speed, cost-effective, miniature and readily manufacturable devices for quantum communications, with the specific emphasis on practical QKD applications30. Previously, protocols of prepare-and-measure QKD31,32,33,34,35,36,37, MDI-QKD38,39,40 and entanglement-based QKD13,41,42,43 have been implemented with integrated-photonics devices but primarily in the point-to-point scenario. Notably, chip-based point-to-point TF-QKD has been demonstrated using partially inte