The challenge of scaling quantum computing may lie not in building bigger processors, but in connecting the ones already in existence. According to a new analysis from memQ, every qubit technology, superconducting, trapped ion, and others, is hampered by the escalating complexity of control wiring, cooling and crosstalk as qubit counts rise.
“The path to a commercially useful quantum computer runs through the network, not through a bigger chip,” the company asserts in a recently published reference architecture. memQ is detailing how linking quantum processors with light, specifically through a quantum network interface controller at each processor and a central Entanglement Hub, offers a viable path toward achieving the millions of qubits needed for practical applications.
Scaling Quantum Computing Through Networked Processors
Photon-mediated entanglement between quantum processors located in separate buildings has already been demonstrated, offering a potentially faster route to scalability than solely increasing qubit counts within a single processor. This milestone, achieved in 2025 by a group, involved running a quantum algorithm across two trapped-ion processors connected only by an optical link, marking the first demonstration of distributed quantum computation over a photonic link. The success hinges on the ability to reliably establish and maintain entanglement, a quantum connection, between distant qubits, a feat previously limited by signal loss and decoherence over long distances.
Every qubit technology currently in production, including superconducting, trapped ion, neutral atom and electron spin systems, faces a common limitation: the scaling of control wiring, cooling requirements and crosstalk within a single processor. These factors hinder the ability to increase the number of qubits that can perform useful computations.
Classical computing overcame a similar bottleneck in the 1990s by connecting multiple machines together, and quantum computing appears to be following the same trajectory. The key to this interconnection is light, specifically the transmission of quantum information via photons, and each processor requires a QNIC to prepare qubits for transmission and to interpret incoming quantum signals. The rate at which entanglement can be established is a key performance metric for a quantum network.
MemQ is addressing this need by building QNICs and quantum memory on commercial foundry silicon photonics, enabling connections over standard telecom fiber, the company says. This approach avoids the cost and time-to-market penalties associated with bespoke fabrication.
The practical consequence of successful entanglement is the ability to perform operations on qubits located in different processors. Researchers have demonstrated gate teleportation, where a quantum gate is executed between separate processors over an optical link, and circuit cutting, where a quantum circuit is divided and distributed across multiple processors. MemQ’s work extends to the development of a distributed quantum compiler (DQC), released as an open-source tool on GitHub in September 2026.
In April 2026, the company received a DARPA contract to develop a hardware-aware DQC, aiming to reduce resource demands by up to 1,000 times. The company’s founders, Sean Sullivan, Manish Singh and Supratik Guha, established memQ in 2022, and it is currently led by chief executive Charles Foley.
Co-founder and CTO Sean Sullivan’s postdoctoral work at Argonne National Laboratory, focused on telecom-compatible qubit devices and the Chicago metropolitan-scale quantum communication testbed, provided a foundation for the company’s current efforts. With $12.5 million in funding, memQ is positioned to play a significant role in the development of a networked quantum computing ecosystem.
Photon-Mediated Entanglement for Quantum Links
Photon-mediated entanglement has already been demonstrated between quantum processors housed in separate buildings, a feat detailed in memQ’s recent reference architecture document, MRA-QNRA-091726.1. The core principle involves using photons to create links between qubits without physically moving the delicate quantum states themselves, a necessity given that “a single quantum cannot be cloned,” as Wootters and Zurek established in 1982. Each processor retains its matter-based qubit and transmits a photon entangled with it.
When these photons converge and are measured, the entanglement transfers to the original qubits, effectively linking the processors for computation. The success of this entanglement is heralded by a specific pattern of detector clicks, relaying information back to the quantum processing units to enable further operations.
The company’s work builds on prior demonstrations, including the entanglement of single atoms over 33 kilometers of telecom fiber, as reported by van Leent et al. in 2022, and the entanglement of nanophotonic quantum memory nodes in a telecom network, detailed by Knaut et al. in 2024. The architecture also details an Entanglement Hub positioned as the center of the network, responsible for managing and distributing entanglement resources.
Precision control is paramount, as the success rate of entanglement directly impacts the overall network performance. MemQ’s approach, as described in MRA-QNRA-091726.1, extends beyond hardware to encompass software, specifically a distributed quantum compiler designed to optimize circuits for execution across the network.
QNIC Role: Extracting and Transmitting Qubit Information
The quantum network interface controller, or QNIC, addresses a fundamental challenge in linking quantum processors: translating qubit states onto photons for transmission, a task requiring precise wavelength and polarization control. memQ’s architecture details how the QNIC ensures photons originating from diverse quantum processing units, or QPUs, are indistinguishable, a necessity for successful entanglement. A rubidium atom, for example, emits photons near 780 nanometers with a bandwidth of approximately 6 megahertz, creating photons lasting roughly 26 nanoseconds; the QNIC must standardize these characteristics across technologies.
The need for such standardization stems from the inherent limitations of quantum information transfer. Unlike classical data, a qubit’s state cannot be copied due to the no-cloning theorem, and reading its state destroys it. This prevents the classical approach of amplification and retransmission. Instead, quantum networks rely on entanglement, a shared state between qubits in separate processors that allows for distributed computation.
Establishing this entanglement requires photons to interfere, a process only possible when a detector cannot discern their origin, demanding identical photon properties, according to memQ. The company’s work, detailed in MRA-QNRA-091726. 1, focuses on achieving this uniformity at the network edge. Founded in 2022 and headquartered in Chicago, memQ has secured $12.
This background, combined with the expertise of co-founder Manish Singh, a PhD in quantum engineering from the University of Chicago, positions memQ to deliver integrated quantum photonics over standard telecom fiber, the firm reports. “ARQUIN: Architectures for Multinode Superconducting Quantum Computers,” details the company’s vision for scalable quantum networking.
Entanglement Hub: Enabling Photon Interaction and Measurement
in Nature. This advance hinges on a shared facility, termed an Entanglement Hub by memQ, which receives photons from multiple quantum processing units (QPUs) and performs the important measurement that links them, the company states. memQ builds quantum network interface controllers and quantum memory on commercial foundry silicon photonics, connecting quantum processors over standard telecom fibre.
This approach circumvents the limitations of direct qubit connection, which suffers from signal degradation over distance and the inability to copy quantum states, a fundamental principle known as the no-cloning theorem. Successful entanglement isn’t guaranteed, requiring repeated attempts and careful detection of photons, but recent demonstrations show increasing reliability.
Two rubidium atoms were entangled across 33 kilometers of telecom fiber, and solid-state memory nodes linked over 35 kilometers within the Boston metropolitan area, demonstrating the potential for long-distance quantum communication. The Entanglement Hub’s design prioritizes efficient photon reception, routing and measurement, a complex task when multiple nodes attempt entanglement simultaneously. The hub must reserve entanglement time in advance, a function managed by a network scheduler, to coordinate teleported gates and optimize circuit execution.
This capability is particularly significant because it allows for a truly heterogeneous network, where different qubit technologies can contribute their strengths to a larger problem, the company’s account states. A vertically integrated system, limited to a single qubit type, cannot offer this flexibility. MemQ’s implementation uses commercial foundry silicon photonics for the fabrication of these integrated photonic circuits, a deliberate strategy to reduce costs and accelerate time to market.
The company’s approach centers on a chiplet-based design, integrating multiple photonic components onto a single substrate. This contrasts with bespoke fabrication methods, which are often expensive and time-consuming. The company reports that its system is designed to handle photons arriving from several nodes simultaneously, directing them toward a common measurement point without signal loss. “Distributed quantum computing across an optical network link,” explains the need for precise control over photon arrival and measurement within the hub.
The success of this architecture depends on the ability to efficiently extract quantum information from qubits and encode it onto photons, a process managed by the QNIC. The ultimate goal is to create a scalable quantum internet, where distributed QPUs can collaborate to solve problems beyond the reach of any single machine.
Precision Control Within Nanosecond Timeframes
The orchestration of quantum information exchange demands synchronization to a timescale of tens of nanoseconds, a constraint that extends beyond hardware limitations into the realm of software design. memQ’s architecture addresses this by separating timing infrastructure from the data network, ensuring precision is not compromised by network traffic. A dedicated optical fiber distributes a common timing reference, establishing a “fast layer” of control distinct from conventional datacenter orchestration.
This separation acknowledges that existing software systems are ill-equipped to manage operations within such narrow time windows. This precise control is critical because photon interference, the basis of entanglement-based communication, requires overlapping emissions and routing decisions occurring within the 26-nanosecond duration of a rubidium atom’s photon.
Achieving this synchronization across machines potentially meters apart necessitates a dedicated infrastructure, a point the company emphasizes for those planning quantum network deployments, memQ claims. “Quantum networking cannot be fully bolted onto conventional datacenter orchestration,” the documentation states, highlighting the need for a specialized control system. The “slow layer”, data processing and higher-level functions, can integrate with existing systems, but the fast layer requires a separate architectural approach.
The company’s approach to this challenge is about managing the complexity of coordinating multiple quantum processors. This hub is the central point where photons from different processors converge, are temporarily stored if needed, and are then measured together to establish entanglement. The entire process, from photon emission to measurement, must occur within that critical nanosecond timeframe. The need for this level of precision is further highlighted by the company’s development of a distributed quantum compiler.
This compiler determines how to divide a computational problem across multiple networked quantum processing units (QPUs), optimizing for the limitations of the network link, memQ says. The compiler’s efficiency is directly tied to the ability to accurately predict and control the timing of photon transmission and reception.
As Dr. Sean Sullivan, Dr. Manish Singh, and Dr. Shobhit Gupta explain in their documentation, the system relies on “precision quantum control, because the entire operation happens inside a time window a few tens of nanoseconds wide.” This focus on timing and control is a deliberate strategy, positioning memQ to provide essential components for a broader quantum ecosystem. By building these specialized interfaces in commercial photonic foundries, the company aims to offer solutions compatible with diverse qubit technologies, rather than being locked into a single platform. This approach, detailed in the company’s reference architecture, allows for a heterogeneous network where different types of quantum processors can collaborate, ultimately accelerating the path toward scalable quantum computing.
Distributed Quantum Compilation Across Processors
This photon-mediated entanglement, where quantum information is transferred via linked photons, relies on four key elements for commercial viability, as detailed in recent research. This hub’s function is critical, as the entire operation must occur within a time window measured in tens of nanoseconds. Precision quantum control is the third necessity, ensuring the fidelity of the entanglement process despite the inherent challenges of manipulating quantum states.
MemQ focuses on building the hardware and software components, QNICs, entanglement hubs and the distributed compiler, using commercial photonic foundries. This deliberate strategy avoids the cost and time associated with bespoke fabrication, offering a potentially universal solution applicable to any qubit technology. Two primary methods exist for dividing a quantum computation across networked processors.
One, termed circuit partitioning, involves dividing a circuit into independent pieces that run on separate processors, with classical post-processing used to stitch the results together. “Circuit knitting with classical communication,” is an alternative approach, though the optimal method depends on the specific problem and network characteristics. The fundamental principle underpinning this network approach is that a single quantum state cannot be copied or transmitted reliably, a concept formalized by Wootters and Zurek in 1982.
Instead, entanglement is used to create a correlated state between qubits on different processors, enabling distributed computation without physically moving quantum information. This method, detailed in publications by Main et al., Knaut et al., and van Leent et al., is rapidly moving from theoretical possibility to practical demonstration.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
