IBM has demonstrated high-fidelity quantum connections between two Nighthawk processors, advancing an approach that could allow future quantum computers to scale beyond individual chips. The achievement highlights a changing engineering challenge: as processors become more capable, the connections, cooling systems and supporting infrastructure needed to turn them into larger machines are becoming increasingly important. It also offers fresh evidence for the industrial transformation examined in The Quantum Space’s State of Quantum 2026 report.

Quantum computing has spent much of its development history pursuing more capable processors. Qubit counts, coherence times, gate fidelities and error-correction performance have become familiar measures of technological progress. But a commercially useful quantum computer will eventually need to do more than operate a sophisticated collection of qubits on a single chip. It must bring together sufficiently large numbers of reliable quantum operations within a system that can be manufactured, maintained and expanded.

This introduces a different engineering problem: how to connect individual quantum processors so that they can operate as parts of a larger machine.

On 27 October, IBM announced an advance towards that objective. Its researchers demonstrated four high-fidelity quantum connections between two Nighthawk processors, using superconducting cables to transfer quantum information between the chips. The experiment achieved an average quantum-state transfer fidelity of 99%, with individual links reaching approximately 99.3%. More significantly, the researchers used the connections to perform quantum operations between the processors and generate entangled states across the two chips.

The result adds another building block to IBM’s roadmap towards modular fault-tolerant quantum computing. It also illustrates why the next stage of development will depend increasingly on infrastructure and systems engineering alongside improvements in quantum processors themselves.

Why quantum computers need to move beyond a single chip

Building larger quantum computers presents challenges that conventional semiconductor manufacturers do not face in quite the same way. Quantum information is exceptionally sensitive to errors, interference and unwanted interactions with its environment. Adding more qubits to a processor increases the complexity of controlling those interactions while maintaining sufficient reliability for useful computation.

Superconducting quantum processors also operate at extremely low temperatures, requiring specialised cryogenic infrastructure, control electronics and extensive connections between the chips and their supporting systems. Simply enlarging a processor does not provide an unlimited route towards more capable machines. IBM’s proposed alternative is a modular architecture in which multiple quantum processors operate together within a shared system.

Instead of requiring every qubit to reside on a single chip, separate processor modules would exchange quantum information through specialised connections. This approach could allow engineers to develop larger systems from smaller, more manageable units. It also creates opportunities to manufacture, test and improve individual modules without redesigning the entire machine.

There is, however, a fundamental requirement. The connections between processors must preserve quantum information reliably enough that they do not undermine the advantages of the architecture. A modular quantum computer with unreliable interconnections would introduce new errors every time information moved between processors. That makes the quality of the connections almost as important as the performance of the processors themselves.

What IBM has demonstrated

IBM’s experiment connected two 120-qubit Nighthawk R1 superconducting processors using four long-distance couplers, known as L-couplers. Each connection incorporated an 80-centimetre superconducting aluminium coaxial cable, together with additional components linking the cable to the qubits on the processors. The researchers achieved an average quantum-state transfer fidelity of 99%, demonstrating that quantum information could be transferred between the chips with relatively low error rates. The experiment then went further.

Using the connections, the team performed controlled-Z, or CZ, gates between qubits located on different processors. These operations achieved an average fidelity of 94.75% across the four links. The researchers also generated entangled states spanning the processors, including a demonstration using two connections simultaneously. These results are important because a useful modular architecture must support quantum operations across processor boundaries, rather than simply transfer information between separate devices. There are nevertheless significant distinctions between the measurements.

The 99% result concerns quantum-state transfer fidelity. It should not be interpreted as a 99% success rate for every quantum operation involving the connected processors. The reported average fidelity for the inter-processor CZ gates was lower, at 94.75%. Both measurements represent progress, but they also show that transferring quantum information and performing high-quality computational operations across a connection are different engineering tasks. The researchers also wired and measured only a subset of the processors’ qubits for the experiment.

The work therefore demonstrates the connections within a large-processor architecture, rather than a fully operational 240-qubit modular computer executing large-scale algorithms. Understanding those limitations does not diminish the achievement. It helps establish what the experiment has demonstrated and what remains to be developed.

The challenge of making connections scalable

Quantum processors have been connected in previous experiments, so the existence of an inter-processor quantum link is not itself new. The particular significance of IBM’s work lies in the attempt to make such connections compatible with larger processor architectures and a greater number of links between modules. That introduces requirements beyond laboratory performance.

Connections must be reproducible, mechanically reliable and compatible with the way processors are assembled and installed. They must also operate within the demanding environment of a cryogenic system. IBM’s design allows the cables to be attached after processor modules have been mounted inside the dilution refrigerator. This is an important manufacturing consideration because a technology that performs well in an experimental assembly may be difficult to reproduce across a larger commercial system.

As the number of processors and connections increases, the infrastructure must accommodate more cables, control systems and thermal constraints without compromising the quantum operations taking place inside it. A future modular computer may need hundreds or thousands of such connections, depending on its architecture. The performance of an individual link will therefore be only one measure of progress.

Engineers will also need to demonstrate that the links can be produced consistently, installed efficiently and operated together without creating unacceptable interference or maintenance requirements. This is where quantum computing begins to resemble the development of a complex industrial platform rather than an isolated scientific instrument.

The growing importance of cryogenic infrastructure

IBM’s interconnect announcement follows another development in August, when the company reported that it had connected and cooled two modular cryogenic units into a shared ultra-low-temperature environment. The architecture is intended to support the expansion of quantum systems through interconnected modules, with room for additional processors, wiring and control infrastructure. Together, the two announcements address complementary requirements.

The cryogenic infrastructure provides the physical environment in which multiple processors could operate. The L-couplers provide a means of transferring quantum information and performing operations between those processors. Neither is sufficient on its own. A modular processor architecture requires supporting infrastructure that can preserve the operating conditions of the hardware while allowing the system to expand. The implications extend into refrigeration, materials, precision manufacturing, electronics, specialised cabling and control software.

For hardware developers, this increases the importance of suppliers that can deliver highly specialised components with consistent performance. For investors and industrial policymakers, it broadens the definition of the quantum computing sector.

Progress towards fault-tolerant quantum computing will be shaped partly by the companies manufacturing quantum processors, but also by the suppliers providing the equipment and systems needed to operate them. The industrial capabilities surrounding the quantum processor are becoming an increasingly important part of the technology’s development.

A building block for fault tolerance

Modularity is central to IBM’s plans for fault-tolerant quantum computing. Its roadmap includes IBM Quantum Starling, a proposed system targeted for 2029 and designed to support 200 logical qubits and 100 million quantum operations. Achieving that objective will require further progress in error correction, quantum processor design, real-time decoding and systems integration. IBM’s latest interconnect work contributes to the architecture required to combine those elements.

The ability to establish quantum operations between separate processors could allow error-correction schemes and computational workloads to operate across a larger collection of physical qubits. However, the interconnect demonstration does not establish that the wider system is fault tolerant. The measured gate fidelities still leave considerable room for improvement, while the researchers must demonstrate greater reproducibility, additional simultaneous connections and reliable operation within larger processor assemblies.

The distinction matters because hardware announcements can easily be interpreted as evidence that the remaining steps towards fault tolerance are becoming routine. They are not.

Each engineering milestone resolves part of a broader problem. The challenge is to combine those advances into a system whose overall reliability supports meaningful computation. IBM’s work provides evidence that one of the necessary architectural components is advancing. It does not eliminate the need for the other components or guarantee that the company’s roadmap targets will be met.

Industry ahead. Market behind.

The development is particularly relevant to the findings of State of Quantum 2026: Industry Ahead. Market Behind., published by The Quantum Space in October. The report examines an industry moving beyond experimental research towards manufacturing, specialised infrastructure, international supply chains and commercially delivered systems.

Its central argument is that quantum computing’s industrial foundations are developing faster than the customer market that must ultimately sustain them. IBM’s latest work reinforces the industrial side of that assessment. The emerging requirements of modular quantum computing create demand for specialised engineering capabilities, advanced manufacturing and infrastructure integration.

They also increase the significance of investment in production facilities, supplier relationships and the ability to construct complete systems rather than individual devices. These are characteristics of an industry developing the capacity to manufacture and deploy more advanced technology. But the commercial implications need to be considered separately.

Demonstrating that two processors can exchange quantum information does not establish that a customer can achieve a useful computational advantage from the resulting architecture. Similarly, developing a larger modular system will not automatically make its operating costs acceptable to enterprises.

The market will eventually need to assess the performance of complete quantum computing services, including the costs of cooling, control, error correction, software and integration with existing computing environments. More sophisticated hardware may enable new applications, but those applications will still need to deliver measurable value. The distinction between industrial capability and customer demand remains as important as the engineering milestones themselves.

A changing question for the quantum industry

The next phase of quantum computing is likely to require a broader understanding of what constitutes technical progress. Processor performance and error-correction results remain essential, but they will increasingly need to be evaluated alongside system-level measures.

Can multiple processors operate together reliably? Can the connections between them be scaled without unacceptable losses? Can the cryogenic infrastructure accommodate larger systems while remaining practical to manufacture and maintain? And can these capabilities be combined into a dependable computing service that customers can access and use? These questions are closely related, but their answers will emerge at different stages.

A successful laboratory experiment may establish that a particular engineering approach is viable. Further development may demonstrate that it can be manufactured reproducibly. Only after complete systems are deployed can the industry properly evaluate operating costs, reliability and practical application performance. IBM’s interconnect work represents progress towards the first of those objectives, with some encouraging indications for the second.

The commercial test lies further ahead. For technology providers and their suppliers, this suggests that the quantum opportunity is expanding beyond the processor itself. Engineering organisations with expertise in cryogenics, advanced electronics, materials and systems integration may become increasingly important participants in the industry’s development.

For enterprises considering quantum computing, the announcement is another indication that the technology’s underlying capabilities are advancing. It is not yet a reason to assume that commercially valuable fault-tolerant computing has arrived.

From connected processors to connected markets

IBM’s latest demonstration provides evidence that modular quantum computing is moving towards more practical system architectures. The achievement is measurable: two large quantum processors, four connections, high-fidelity state transfer and quantum operations performed across processor boundaries. The limitations are also measurable, particularly the lower fidelity of inter-processor gates and the need to demonstrate reliable performance across larger numbers of connected modules. Both are necessary to understand the technology’s progress.

The wider significance is that quantum computing’s development is becoming increasingly dependent on the infrastructure surrounding its processors. As the industry moves towards fault tolerance, the performance of the complete system will matter more than the characteristics of any individual component. That creates new engineering challenges and opportunities for a growing network of suppliers. It also reinforces the need to distinguish industrial progress from commercial adoption.

IBM has demonstrated another important building block for larger quantum computers. The next steps will determine whether that building block can become part of a reliable, manufacturable and economically viable computing platform.

The future of quantum computing may depend as much on how its processors are connected as on how powerful those processors become.

---

Sources and further reading

• IBM Quantum — Modular Quantum Computing Takes Shape With L-Couplers, 27 October 2026
• IBM research paper — Meter-Scale Quantum Interconnects Between Large Quantum Processors
• IBM — Modular Cryogenic Systems, 19 August 2026
• IBM Quantum Roadmap
• The Quantum Space — State of Quantum 2026: Industry Ahead. Market Behind.