IBM, UC Berkeley Scientists Map Path to Larger Quantum Processors

Summarize this article with:
Xanthe Croot of the University of Sydney, Kasra Nowrouzi and Christopher Spitzer of Lawrence Berkeley National Laboratory, alongside colleagues from numerous institutions including IBM Quantum and the University of California, Berkeley, suggest critical areas of development to accelerate the realization of large-scale fault-tolerant quantum computers. This article focuses on superconducting qubits and technologies enabling efficient signal and information transmission interfacing with a quantum processing unit (QPU). Advancing beyond proof-of-concept experiments, the authors explore pathways toward systems exceeding 100-1000 physical qubits, essential for robustly generating and manipulating exponentially large computational spaces and achieving the full potential of quantum computation. Enabling Technologies for Scalable Quantum Computing To achieve large scale fault-tolerant quantum computation (FTQC), scaling beyond current qubit numbers is critical, with goals exceeding 100-1000 physical qubits. This necessitates advances in technologies that efficiently shuttle signals on and off quantum processors. Modular architectures are advantageous, ranging from connecting multiple qubit dies on a substrate to linking packaged qubits with cabling. Maintaining quantum-coherent communication between modules, ideally at cryogenic temperatures via microwave signals, is a key consideration for scaling. Different levels of modularity are being explored, with link lengths dictating communication methods. Short links—less than a few wavelengths—allow direct qubit-qubit swaps. Longer links—up to 64 meters have been demonstrated—require resonant cable modes or shaped envelopes for signal transmission. “Bright mode” swaps or itinerant signals demand high-quality cables, while “dark modes” can mitigate losses but are typically slower. Future systems will also benefit from “hot swap” cable configurations for maintenance. For connections exceeding 1000 cm or extending outside a cryostat, optical fibers or free-space transmission are appealing. However, quantum-coherent communication necessitates transduction between microwave signals (from superconducting qubits) and optical signals. Researchers are actively pursuing various methods for this conversion, aiming to enable communication between distant quantum processing units and facilitate larger, more powerful quantum computers. Motivations for Modular Qubit Architectures Modular qubit architectures are being considered to address challenges related to scaling superconducting quantum computers. Increasing qubit density on a single substrate creates issues with qubit yield, performance uniformity, and wiring complexity. Approaches range from attaching multiple qubit dies (“chiplets”) to an interconnect substrate, to using connectorized packages and cabling for inter-die connections. Maintaining low cryogenic temperatures for these linkages enables quantum-coherent communication, though room temperature linkages would require microwave to optical frequency transduction. Different levels of modularity are already being explored, demonstrating varied link lengths and communication modalities. Examples include on-chip waveguides (less than 1 cm), linked chips (~1 cm), and packaged qubits connected via cabling within a single cryostat (~10 cm). Even longer connections, exceeding 1000 cm, involve either native microwave frequency links or conversion to optical photons in separate cryostats. Communication methods shift with link length, from direct qubit-qubit swaps for short links to resonant cable modes or shaped envelopes for longer ones. Link length significantly impacts communication strategies. Shorter links allow direct qubit swaps, while longer cables favor swaps via resonant modes or itinerant photon signals. While “dark mode” swaps can mitigate losses, they are slower than direct “bright mode” swaps or itinerant signals, which require high-quality cables and connections. Future systems will also benefit from “hot swap” cable configurations for maintenance, enabling isolated fridge modules to be serviced without halting the entire system. Tour de gross: A modular quantum computer based on bivariate bicycle codes Modular Approaches to QPU Design Modular architectures are being considered to address challenges associated with scaling superconducting qubits, specifically issues with qubit yield, frequency collisions, and wiring complexity. Approaches range from connecting multiple qubit dies via an interconnect substrate to using connectorized packages and cabling for inter-die connections. Maintaining low cryogenic temperatures for these linkages allows for quantum-coherent communication via microwave signals, though this is conceptually similar to simply building a larger cryostat. Different levels of modularity are being explored, with implementations varying from on-chip waveguides (less than 1 cm) to connections using cabling within a single cryostat (approximately 10 cm), and even linking modules with cables exceeding 64 meters. Communication modality shifts with link length; short links allow direct qubit-qubit swaps, while longer links favor swaps via resonant cable modes or shaped envelopes for mobile photon signals. Maintaining high-quality cable connections is crucial for longer-range coupling. For connections exceeding 1000 cm, or linking modules in separate cryostats, transduction between microwave and optical frequencies becomes necessary. Approaches are being pursued to convert signals for transmission via optical fibers or free-space, though this requires efficient methods of generating microwave-optical conversion. Developing ‘hot swap’ cable configurations at cryogenic temperatures is also important for maintenance without system downtime. On-Chip Waveguide Interconnections On-chip waveguide interconnections represent one approach to modular quantum processing unit (QPU) architecture. The source details that interconnected processing units can be linked by these on-chip waveguides, specifically for microwave transmission lines, enabling communication between qubits on the same die. These links are characterized by very short lengths – less than 1 cm – and allow for direct qubit-qubit swaps, facilitating quantum coherent connectivity within the chip itself. This approach is visualized in Figure 1(a) of the source material. Different levels of modularity are being explored, and on-chip waveguides represent a short-range communication method. Compared to longer links – such as those using cabling or optical fibers – these waveguides enable faster communication modalities, like direct “bright mode” swaps. However, maintaining high quality factor connections remains crucial, even at these short distances, to minimize signal loss and preserve quantum coherence, as detailed in the source’s discussion of communication modalities. The source also notes that while on-chip waveguides allow direct qubit-qubit swaps, longer cable links require alternate methods like resonant cable modes or shaped envelopes for itinerant photon signals. These methods become necessary as link lengths increase beyond a few wavelengths at the qubit communication frequency. The source highlights that, despite these alternative methods, maintaining high-quality connections remains a key challenge for longer-range coupling, even surpassing the needs of on-chip waveguide implementations.
Linked Chip Implementations Modular architectures are being explored to overcome challenges associated with scaling superconducting qubits, such as qubit yield, frequency collisions, and wiring complexity. Approaches range from attaching multiple qubit dies (“chiplets”) to an interconnect substrate, to using connectorized packages and cabling for inter-die connections. Maintaining cryogenic temperatures for these linkages allows quantum-coherent communication via microwave signals; however, room temperature linkages require transduction between microwave and optical frequencies. Different levels of modularity have been demonstrated with varying link lengths, influencing communication modalities. Very short links—less than a few wavelengths at the qubit communication frequency—allow direct qubit-qubit swaps. Longer links utilize resonant cable modes or shaped envelopes for mobile photon signals, though achieving this requires high quality factor cables and connections. Future development includes ‘hot swap’ cable configurations for maintenance without system downtime. The source details several demonstrations of linked QPUs, ranging from on-chip waveguides (1000cm). These varying lengths necessitate different communication strategies; while short links support direct swaps, longer distances rely on resonant modes or itinerant photons, all requiring careful consideration of cable quality and connection integrity to maintain quantum coherence.
Linked Cryopackage Connections The source details several modular approaches to building quantum processing units (QPUs), including linking cryopackages. Specifically, independently packaged QPUs can be connected via cabling within a single cryostat. This falls under the “Linked Cryopackages” approach, characterized by connection distances around 10 centimeters. Maintaining cryogenic temperatures for these linkages allows for quantum-coherent communication using microwave signals, though this is conceptually similar to simply building a larger single cryostat. Different link lengths dictate communication methods. Very short links – less than a few wavelengths at the qubit communication frequency – enable direct qubit-qubit swaps. However, for longer cable links like those connecting cryopackages, swaps rely on resonant cable modes or shaped envelopes for itinerant photon signals. Maintaining high quality factor cables and connections is crucial for effective communication over these distances, presenting a significant engineering challenge. For connections exceeding 1000 centimeters, or linking modules in separate cryostats, transduction between microwave and optical frequencies becomes necessary. This allows for signal transmission via optical fibers or free-space transmission. Developing effective methods for this transduction is an active area of research, focusing on converting signals between superconducting qubits and optical signals for long-distance, quantum-coherent communication.
Distant Module Communication Modular quantum processing unit (QPU) architectures are being explored to overcome challenges related to qubit yield, frequency collisions, and wiring complexity as qubit numbers increase. Approaches range from connecting multiple qubit dies on a substrate to using cabling for inter-die connections, even extending to packaged qubits linked within a single cryostat. Maintaining cryogenic temperatures for these linkages allows microwave signals to maintain quantum coherence; however, room temperature linkages require transduction between microwave and optical frequencies. Different link lengths dictate communication modalities. Very short links, under a few wavelengths, enable direct qubit-qubit swaps. Longer cable links favor swaps via resonant modes or shaped envelopes for mobile photon signals, though losses can be sidestepped with “dark modes” at the cost of speed. High-quality cabling and connections are crucial for longer ranges, alongside considerations for “hot swap” configurations enabling maintenance without system downtime at temperatures like 4K or mK. For connections extending beyond a single cryostat, optical fibers or free-space transmission offer appeal. However, quantum-coherent communication necessitates converting between microwave signals from superconducting qubits and optical signals for transmission. Several transduction methods are being pursued to achieve this conversion, utilizing approaches to generate microwave-optical interfaces for long-distance connectivity exceeding 1000 cm, as demonstrated by a 30m cryogenic link used in a Bell inequality violation experiment. Communication Modalities and Link Lengths Modular quantum processing unit (QPU) architectures are being explored to address challenges with scaling superconducting qubits. Increasing qubit numbers on a single substrate creates issues with yield, frequency collisions, and wiring complexity. Approaches range from connecting multiple qubit dies via an interconnect substrate to using packaged qubits connected by cabling. Maintaining quantum-coherent communication linkages between these modules—even at low cryogenic temperatures—is key, enabling microwave signal transmission for connectivity. Different link lengths necessitate varying communication modalities. Very short links—less than a few wavelengths at the qubit communication frequency—allow direct qubit-qubit swaps. However, longer cable links favor swaps via resonant cable modes or shaped envelopes for itinerant photon signals. While “dark modes” can reduce losses, they are slower than direct swaps or itinerant signals, which require high-quality cables and connections—a challenge for extended ranges. For connections exceeding 1000 cm, or extending outside a cryostat, optical fibers or free-space transmission of optical signals are attractive. Establishing quantum-coherent communication over these distances, however, demands transduction between microwave signals (from the qubits) and optical signals for transmission. Examples of demonstrated link lengths range from on-chip waveguides (<1 cm) to 30m cryogenic links, and even 64m cables within a single cryostat. Qubit-Qubit Swaps and Communication Modular quantum processing unit (QPU) architectures are being explored to overcome challenges associated with scaling superconducting qubits, such as yield and wiring complexity. Approaches range from connecting multiple qubit dies on a substrate to using cabling to link packaged qubit circuits, even across separate cryostats. Maintaining sufficiently low cryogenic temperatures allows for quantum-coherent communication between modules using microwave signals; however, room-temperature links require transduction between microwave and optical frequencies. Different link lengths dictate communication methods. Very short links—less than a few wavelengths at the qubit communication frequency—enable direct qubit-qubit swaps. Longer links utilize resonant cable modes or shaped envelopes for itinerant photon signals. While “dark mode” swaps can reduce losses, they are slower than “bright mode” swaps or itinerant signals, which require high-quality cabling and connections. For connections exceeding 1000 cm, or outside a cryostat, optical fibers or free-space transmission become appealing. However, quantum-coherent communication necessitates transduction between microwave signals from superconducting qubits and optical signals. Future development will focus on “hot swap” cable configurations at cryogenic temperatures, enabling maintenance of isolated fridge modules without system downtime. Challenges with Long-Range Coupling Increasing the number of superconducting qubits presents challenges with qubit yield, performance uniformity, frequency collisions, and wiring complexity. This drives consideration of modular qubit architectures, ranging from attaching multiple qubit dies to an interconnect substrate, to using connectorized packages and cabling for inter-die connections. Maintaining quantum-coherent communication linkages between these modules, ideally at low cryogenic temperatures using microwave signals, is a key advantage, though larger designs could simply require a larger cryostat. Different levels of modularity have been explored with varying link lengths influencing communication methods. Very short links—less than a few wavelengths at the qubit communication frequency—allow direct qubit-qubit swaps. Longer cable links favor swaps via resonant cable modes or shaped envelopes for itinerant photon signals. However, achieving high-quality factor cables and connections is challenging for longer-range direct coupling, with “dark modes” offering a slower alternative to mitigate losses. For connections exceeding 1000 cm, or outside a single cryostat, optical fibers or free-space transmission of optical signals become appealing. However, quantum-coherent communication necessitates transduction between microwave signals from superconducting qubits and optical signals. Approaches to this transduction are actively being pursued, aiming to enable communication between distant modules without sacrificing quantum coherence, and considering ‘hot swap’ configurations for maintenance. Hot-Swappable Cable Configurations Modular quantum processing unit (QPU) architectures are being explored to overcome challenges related to qubit yield, frequency collisions, and wiring complexity. Approaches range from connecting multiple qubit dies on a substrate to using cabling for inter-die connections, and even linking modules in separate cryostats. Different link lengths dictate communication methods; very short links allow direct qubit-qubit swaps, while longer links require resonant cable modes or itinerant photon signals. The source highlights a crucial future consideration: the development of ‘hot swap’ cable configurations at 4K or milliKelvin temperatures. These configurations would enable isolated fridge modules to undergo maintenance without interrupting the entire quantum system. This capability is essential for scaling and maintaining large, complex quantum computers, as it allows for modular repair and upgrades without complete system downtime. For connections extending beyond a single cryostat, optical fibers or free-space transmission of optical signals become appealing. However, achieving quantum-coherent communication requires transduction between microwave signals from superconducting qubits and optical signals. Various methods are being pursued to facilitate this transduction, enabling communication between distant quantum modules and furthering the development of scalable quantum computing systems. Optical Transduction for Quantum Communication Optical transduction—converting between microwave and optical frequencies—becomes necessary when linking qubit modules located in distant cryostats. The source details that for connections exceeding the scale of a single cryostat, methods of generating microwave-to-optical conversion are being pursued to enable quantum-coherent communication. This is because maintaining quantum coherence requires a way to transmit quantum information across larger distances, and optical signals are well-suited for this purpose beyond cryogenic environments. Different link lengths dictate communication modalities; very short links (under a few wavelengths) allow direct qubit-qubit swaps. However, longer cable links necessitate swaps via resonant cable modes or shaped envelopes for mobile photon signals. While “dark modes” can minimize losses, they are slower than “bright mode” swaps or itinerant signals, which demand high-quality cabling and connections. For connections exceeding 1000cm, optical fibers or free-space transmission are appealing, but still require microwave-to-optical transduction. The source highlights ongoing research into methods of achieving this transduction between microwave signals from superconducting qubits and optical signals for transmission. This is critical for scaling quantum systems beyond the limitations of a single cryostat, allowing for the modular connection of quantum processing units. Different modular approaches, ranging from on-chip waveguides to linked cryopackages, are being explored to facilitate these connections, with optical links becoming essential for distances exceeding 1000cm. Microwave to Optical Signal Conversion To scale superconducting quantum computers, modular architectures are being explored. These range from connecting multiple qubit dies on a single substrate to linking qubit dies in separate cryostats. Maintaining quantum-coherent communication between these modules is key; short links utilize microwave signals directly. However, for longer distances, especially connecting modules in distant cryostats, transduction between microwave and optical frequencies becomes necessary, requiring methods to convert signals for transmission. For connections exceeding 1000 cm, optical fibers or free-space transmission of optical signals are attractive options. Establishing quantum-coherent communication via these methods demands a conversion process—transducing between microwave signals from superconducting qubits and optical signals. Several approaches are being pursued to achieve this conversion, focusing on generating microwave-optical links. This is crucial for extending the reach of quantum communication beyond a single cryostat. Different link lengths dictate communication methods. Very short links allow direct qubit-qubit swaps, while longer links favor swaps via resonant cable modes or shaped envelopes for itinerant photon signals. Although “dark modes” can minimize losses, they are slower than direct “bright mode” swaps or itinerant signals. “Bright mode” and itinerant signals require high-quality cables and connections, posing challenges for longer-range coupling and necessitating advancements in cable technology for maintenance and system operation. Cryogenic Back-Plane Considerations Modular architectures are being explored to overcome challenges related to qubit yield, frequency collisions, and wiring complexity as qubit numbers increase. These range from connecting multiple qubit dies on a substrate to using cabling for inter-die connections. Maintaining low cryogenic temperatures for these linkages allows for quantum-coherent communication using microwave signals; however, room-temperature linkages would require transduction between microwave and optical frequencies. Different link lengths dictate communication modalities, impacting qubit-qubit swaps. The cryogenic back-plane—or the environment where a QPU is placed—is a key consideration. Very short links (under a few wavelengths at qubit communication frequency) enable direct qubit-qubit swaps. Longer cable links rely on resonant cable modes or shaped envelopes for itinerant photon signals. Maintaining high quality factor cables and connections is critical for longer-range coupling, and ‘hot swap’ configurations are envisioned for maintenance without system downtime at 4K or mK temperatures. Various modular approaches are being tested, including interconnected processing units using on-chip waveguides (under 1 cm), linked chips (~1 cm), and cryopackages connected by cabling (~10 cm). For distances exceeding 1000 cm, independent QPUs can be linked with microwave frequency links or via conversion to optical photons—requiring transduction between microwave and optical signals—to facilitate communication between separate cryostats. Need for Cryogenic Electronics Scaling superconducting quantum computing presents challenges including qubit yield, frequency collisions, and wiring complexity. Modular architectures are proposed as a solution, ranging from connecting multiple qubit dies to using connectorized packages and cabling. Maintaining quantum-coherent communication between these modules, ideally at low cryogenic temperatures via microwave signals, is key. However, room temperature linkages would necessitate transduction between microwave and optical frequencies. Different modular approaches are being explored with varying link lengths, impacting communication methods. Short links—less than a few wavelengths—allow direct qubit-qubit swaps. Longer cable links require resonant modes or shaped envelopes for signal transmission. Maintaining high quality factor cables and connections becomes increasingly crucial with distance, and “hot swap” cable configurations are considered for maintenance without system downtime. The need for cryogenic electronics is directly tied to the pursuit of larger, scalable quantum computers. Technologies are required to efficiently shuttle signals on and off the quantum processor (QPU). This is particularly important when considering modular designs where maintaining quantum coherence through cabling—ranging from centimeters to tens of meters—demands high-quality connections and potentially, transduction between microwave and optical frequencies for longer-range communication. Source: https://arxiv.org/pdf/2512.15001 Tags:
