IBM plans to connect two 520-qubit superconducting processors into one quantum system and open it to outside users in 2027. The aim is to grow quantum computers by linking chips together instead of making single chips ever larger.
A two-chip quantum computer
A quantum computer stores information in qubits rather than ordinary bits. A classical bit is always either 0 or 1. A qubit can be held in a superposition of both, and groups of qubits can be entangled, so their states are linked in ways no classical system can copy. For certain problems, such as simulating molecules and materials or some kinds of optimisation, this could in principle beat conventional machines.
IBM builds its quantum processors from superconducting circuits. These are tiny electrical components that act as qubits when cooled to temperatures close to absolute zero. Each processor is a chip holding a fixed number of these qubits.
Ars Technica reports that IBM intends to join two such chips, each with 520 qubits, into a single system before the end of next year, and to let users access it in 2027. A dual-processor system means the two chips work together on a task instead of running separate jobs. That gives users more qubits than either chip could offer alone.
Origins of the modular approach
For years, progress in quantum hardware was mostly measured by how many qubits fit on one chip. That approach runs into practical limits. Bigger chips are harder to make without defects. They need more control wiring running into the refrigerator. Larger arrays of qubits also tend to pick up more noise and interference, which damages the fragile quantum states the machine relies on.
IBM, like several other companies and research groups, has therefore turned to modularity. The idea is familiar from classical computing. Supercomputers are not single giant processors but many processors connected by fast links. In quantum computing, the equivalent is building smaller, more reliable chips and connecting them so they behave as one larger machine.
IBM has published a public roadmap for its quantum hardware, and its plans have increasingly centred on systems built from several connected processors. The company’s modular platform, known as IBM Quantum System Two, was designed to house more than one processor alongside the cryogenic and control equipment they need. The dual-processor plan reported by Ars Technica fits this broader direction. Exactly how it slots into IBM’s published roadmap cannot be confirmed from the summary of the report available here.
How the system is expected to work
Every superconducting quantum processor sits inside a dilution refrigerator, which keeps it cold enough for the qubits to work. Microwave signals sent through cables control the qubits and read out their states. Classical computers then process the results.
Connecting two processors can be done in broadly two ways. In the simpler case, the chips are linked through classical communication. They exchange ordinary data, and software splits a computation across them and stitches the results together. This works, but it adds overhead and limits the kinds of calculations that can be shared. In the more ambitious case, the chips are linked by quantum connections, such as couplers or channels that carry quantum states, so that qubits on one chip can interact directly with qubits on the other. That lets the two processors act more like one large chip.
The reporting summarised here does not say which method IBM will use for the 520-qubit pair, how fast or reliable the link will be, or what error rates the combined system is expected to reach. These details matter a great deal and are not known from the available material.
Access will almost certainly be remote. IBM has long offered its quantum hardware through the cloud. Researchers and companies submit programs, often written with IBM’s open-source Qiskit software, and receive results without ever being near the machine.
Common misconceptions
More qubits do not automatically mean more power. Qubit count is the easiest number to report, but it says little on its own. Error rates, how long qubits keep their quantum state, and how well they connect to one another often matter more. Two linked 520-qubit chips are not necessarily equal to one flawless 1,040-qubit chip. How useful the system is depends heavily on how well the link performs.
This is not yet a fault-tolerant machine. Today’s quantum computers are noisy. Running long, reliable calculations will need quantum error correction, which spreads one “logical” qubit across many physical qubits so that errors can be detected and fixed. The dual-processor system is better seen as a step towards the scale error correction will require, not as the finished product.
Quantum computers will not replace ordinary computers. They are specialised devices suited to particular kinds of problems. Most everyday computing, from email to video, gains nothing from them.
Access does not mean an immediate breakthrough. Making a system available lets researchers test algorithms and look for applications. It does not mean the machine will solve commercially important problems faster than classical supercomputers. Whether and when that happens for practical tasks is still an open question.
Where to follow developments
Readers who want to follow this story should start with IBM’s own quantum research publications and its published hardware roadmap. These set out the company’s stated goals and the names of its processor generations. Comparing announcements with later technical papers shows which plans were delivered as described.
Peer-reviewed journals such as Nature and Physical Review publish detailed results on qubit performance, error correction and links between processors. They are the most reliable place to check claims about error rates and benchmarks. Preprint servers such as arXiv carry research before formal review, though those results should be treated with more caution.
Specialist science and technology outlets, including Ars Technica, which reported this plan, often explain hardware announcements in more depth than general news. To learn the basics, the Qiskit documentation and IBM’s free cloud access let anyone run small programs on real quantum hardware.
Frequently asked questions
What is IBM’s dual-processor quantum system?
According to Ars Technica, it is a planned IBM quantum computer made by linking two superconducting chips with 520 qubits each, so they can work together on one computation. IBM intends to build the link before the end of next year and offer access to users in 2027. It is part of a wider move towards modular quantum computers built from several connected processors.
Why link two quantum chips instead of building one bigger chip?
Very large single quantum chips are hard to make without defects. They need large amounts of control wiring and tend to suffer more noise. Linking smaller, more reliable chips is a way to increase the total number of usable qubits while avoiding some of these manufacturing and engineering limits. Classical supercomputers grew in a similar way, by connecting many processors rather than relying on a single enormous one.
How will people access IBM’s 2027 quantum system?
The access details have not been described in the material available. However, IBM has long offered its quantum computers remotely through a cloud service, where users submit programs and receive results over the internet. Programs are typically written with Qiskit, IBM’s open-source quantum software toolkit. It is reasonable to expect a similar model, though the exact terms and eligibility are not known.
Is a 1,040-qubit linked system better than a single large chip?
Not necessarily. Two 520-qubit chips add up to 1,040 qubits, but performance depends on error rates, qubit stability and especially how well the two chips communicate. If the link is slow or error-prone, the combined system may behave less like one large processor and more like two separate ones working together. Published benchmarks will be needed to judge its real capability.
Can IBM’s new quantum computer break encryption?
There is no indication that it can. Breaking widely used public-key encryption would need large, fault-tolerant quantum computers running error-corrected algorithms over long periods. Current machines, including systems of this size, are still noisy and lack full error correction. Security bodies nonetheless encourage organisations to prepare for post-quantum cryptography, because data captured today could be decrypted by more capable machines in the future.
Sources and further reading
- Ars Technica: the original report on IBM’s plan to link two 520-qubit chips and offer access in 2027
- IBM Research: official publications and the public quantum hardware roadmap describing processor generations and modular systems
- Nature and Physical Review: peer-reviewed research on superconducting qubits, error correction and links between processors
- Qiskit documentation: IBM’s open-source software guides explaining how users program and access quantum hardware
Surfaced from the rss:arstechnica signal “modular quantum computer plan”. AI-assisted draft, editorially reviewed.
