Notice
This article was written based on IBM’s publicly available quantum computing announcements and materials as of August 21, 2026. DANA NOTES analysis is also included.
Recommended Reading
What This Article Covers
Key Terms to Know
- Cryogenic System — A system that cools devices such as quantum processors that operate at very low temperatures and maintains a stable environment.
- Fidelity — A measure of how closely the transfer of a quantum state or a quantum operation matches the intended state.
- Logical Qubit — A qubit used as a single stable computational unit by grouping multiple physical qubits through quantum error correction.
- L-coupler — A microwave-based interconnect technology IBM is developing to connect quantum processors located apart from one another.
IBM’s quantum computers start with cooling
IBM develops quantum processors based on superconducting qubits.
Superconducting qubits are highly affected by heat and their surrounding environment, so they operate in cryogenic environments close to absolute zero. IBM’s quantum processors are cooled inside a dilution refrigerator to around 15 mK, or approximately -273.135°C.
Quantum chips are located in the lowest-temperature region of the cooling system. Several temperature stages are arranged around them, while a significant portion of the control equipment operates in higher-temperature regions or at room temperature.
On August 19, 2026, IBM announced that it had successfully connected two cryogenic modules and cooled them as a single ultracold environment.
The two modules were cooled together to 4 K within five days and then brought down to below 15 mK. The new modules are designed to provide up to 12 times the wiring space of existing IBM quantum systems. IBM plans to expand this structure into a system capable of connecting hundreds of quantum chips.
The important part of this technology is how to connect and control many quantum chips inside a cryogenic environment.
Quantum chips inside the refrigerator remain connected to external equipment
For a quantum chip to perform computations, it must receive control signals from external equipment. Measurement signals used to read the results of quantum computations must also be sent back to external equipment.
In IBM’s superconducting quantum computers, microwave signals travel through wiring into the cryogenic environment. Measurement signals from the quantum chip also travel back outside through the wiring.
The structure can be expressed simply as follows.
Control Equipment → Wiring → Quantum Chip → Wiring → Measurement and Control Equipment
When multiple quantum chips perform computations together, connections between the quantum chips are also required.
Improving the performance of quantum processors alone makes it difficult to translate that performance into the performance of the entire system. The paths that carry control signals and connect multiple processors must also support that performance.
This structure can also be understood through the USB devices we use.
A USB 3.0 device is limited by a 2.0 connection environment
Suppose you have a fast storage device that supports USB 3.0.
If the computer port is USB 2.0 and the USB hub in between is also USB 2.0, the actual data transfer is affected by that connection environment.
To utilize the performance of a device, the path used to exchange data must also support that performance.
Quantum computers also depend on both the quantum processor and its connectivity environment.
Control signals travel through wiring, computation results come back out, and when multiple quantum chips are used, information also moves between processors.
In IBM’s superconducting quantum computers, these connections extend into cooling systems operating near absolute zero.
With 1,121 qubits, the wiring inside the refrigerator exceeded one mile
IBM’s Condor, unveiled in 2023, is a quantum processor integrating 1,121 superconducting qubits.
A single dilution refrigerator built for Condor contained more than one mile of high-density cryogenic input/output wiring. That means more than approximately 1.6 km of wiring was configured inside a single cooling system.
This wiring sends signals to the quantum chip and reads out the results.
At the same time, because it connects room-temperature and cryogenic regions, it can become a path through which heat travels. When microwave signals are transmitted, signal attenuation, electromagnetic noise, and interference from nearby signals must also be managed.
As the number of qubits increases, the infrastructure required to cool, control, and connect quantum chips also grows in scale.
Condor demonstrated IBM’s implementation of more than 1,000 qubits on a single chip, while also showing how large the cryogenic wiring and control environment required at that scale can become.
IBM chose modularization because it is difficult to keep scaling a single quantum chip
IBM is developing a structure that connects multiple quantum processors for large-scale quantum computers.
As the number of qubits increases, the resources required for chip area, manufacturing yield, packaging, signal input/output, and wiring also increase. The size that semiconductor manufacturing equipment can handle and the yield that must be secured during production also affect system scalability.
IBM Research identifies semiconductor manufacturing equipment, manufacturing yield, and signal delivery as major technical conditions in scaling large superconducting quantum systems.
IBM is therefore pursuing an approach in which multiple quantum processors of a certain scale are configured and connected to expand the overall computing system.
The structure can be expressed simply as follows.
Quantum Chip + Quantum Chip + Quantum Chip → One Larger Quantum System
When multiple quantum chips are configured together, the cryogenic environment that accommodates them must also be able to scale in the same way.
This is why IBM has modularized the cryogenic system as well.
IBM is building a structure that can scale the overall system by adding quantum processors and cryogenic modules according to the required scale.
Modularization makes ‘connectivity’ a larger technical challenge
Connecting multiple quantum chips and cryogenic modules creates sections where quantum information must move between processors.
In these sections, signal loss, thermal noise, electromagnetic noise, and interference from nearby signals can affect quantum information.
Signal attenuation and noise also occur in conventional digital communications. Digital systems have evolved to transmit data reliably by distinguishing between 0 and 1 and using technologies such as error detection, error correction, and retransmission.
In quantum computers, quantum states themselves, including superposition and phase, are used as computational information.
As quantum information moves through a connection, thermal noise or electromagnetic noise can affect it, while signal loss and interference can alter the original quantum state. These changes lead to computational errors.
Arbitrary quantum states can also change during measurement, and the same state cannot be copied exactly.
For connections between quantum processors, transmission speed and the ability to reduce information degradation caused by noise, signal loss, and interference are therefore important technical conditions.
The USB structure provides another easy way to understand this.
To utilize the performance of a USB 3.0 device, the port and hub must support sufficient speed. In a quantum computer, the system must also manage the possibility that noise and signal loss can damage the information itself as it passes through cables and interconnects.
IBM is also developing technology to reduce information loss between modules
IBM is developing the L-coupler to connect quantum processors located apart from one another.
In 2025, IBM Research published experimental results in which quantum states were transferred between separate modules using a microwave connection approximately 60 cm long.
The state-transfer fidelity, which indicates the accuracy of the quantum-state transfer, was 98.8%.
The remote CNOT gate fidelity for a single CNOT operation performed across the two modules was 93.3%.
Even across a 60 cm connection, IBM measures not only whether the signal arrived but also how accurately the original quantum information was transferred.
In an earlier experiment, IBM also implemented a CNOT operation between two quantum chips using a connection up to approximately 1 meter long. The experimental setup measured an operation time of approximately 235 ns and a gate error rate of 3.5%.
As more quantum chips and modules are connected, the number of these connection sections also increases.
Technology that keeps errors at each connection at a low level becomes an important factor in determining the scalability of modular quantum computers.
IBM is modularizing quantum chips and cooling systems together
The new cryogenic system IBM unveiled in August 2026 was designed with a structure that allows multiple modules to be connected.
IBM connected two cryogenic modules into one shared ultracold environment and plans to place quantum processors inside multiple modules to configure one larger quantum computer.
IBM plans to install Nighthawk processors in the new cryogenic modules in the second half of 2026 and expand operational testing.
By 2027, IBM is targeting a system with at least 1,000 programmable qubits by connecting multiple processors through L-couplers.
At the Starling stage, IBM plans a structure in which each cryogenic module accommodates thousands of qubits.
This test confirmed that multiple cryogenic modules can be connected as a single environment and cooled to the target temperature.
The next stage will install actual quantum processors and test whether cooling and computational accuracy can remain stable even with connections between modules.
This challenge provides the foundation for fault-tolerant quantum computing
IBM aims to build IBM Quantum Starling in 2029.
Starling is designed as a large-scale fault-tolerant quantum computer capable of running circuits consisting of 100 million quantum gates across 200 logical qubits.
Logical qubits operate through multiple physical qubits working together while continuously detecting and correcting errors.
Therefore, advances in error-correction technology alone are not enough to complete Starling.
This process also requires a cooling system that keeps many quantum chips at cryogenic temperatures, a control environment that sends signals to each processor, and connectivity technology that transfers quantum information between processors.
IBM’s current cryogenic module experiment is part of the process of scaling this physical foundation.
Modular quantum processors, modular cryogenic systems, and inter-processor connectivity technology must operate together as one structure to support large-scale fault-tolerant quantum computing.
Competition in quantum computing is expanding beyond the chip
In June 2026, IBM announced plans to invest more than $10 billion in quantum computing over the next five years.
The investment covers research and development, capital expenditure, manufacturing expansion, ecosystem partnerships, and strategic acquisitions.
IBM’s quantum computing roadmap also shows an expanding range of technologies required to build large-scale systems.
Cryogenic equipment, high-density wiring, microwave and RF components, control electronics, packaging, and processor interconnects all operate together within a quantum computer.
IBM Quantum System Two also integrates scalable cryogenic infrastructure, classical-computing-based runtime servers, and modular control electronics into a single system.
As quantum computers become larger, the ability to design and integrate quantum chips and the surrounding infrastructure as one system is also becoming a larger component of industry competitiveness.
DANA NOTES Commentary
The core of IBM’s modularization strategy is creating scalable units.
By configuring quantum chips and cryogenic environments as modules of a certain scale, the system can be designed to expand by adding modules according to the required computing scale.
For this structure to translate into actual performance gains, adding one module must also increase available computing resources. At the same time, noise, signal loss, and operational errors arising at connection points must remain within manageable levels.
In IBM’s modular quantum computer, the computing capacity gained by adding modules and the errors introduced through connectivity therefore become important indicators together.
If this technology becomes stable, quantum-computer scaling can develop toward a system architecture that combines multiple processors and cryogenic modules.
IBM’s latest cryogenic module test is significant because it marks the beginning of validating this scalable architecture in actual hardware.
Variables to Watch
The nearest point to watch is the operational results after Nighthawk processors are installed in the new cryogenic modules in the second half of 2026.
The first question is whether quantum processors can operate stably while the cryogenic modules are connected and whether the target temperature can be maintained as the wiring and control structure expand.
The next variables are L-coupler fidelity and the operational error rate between processors. An important question is whether connectivity technology currently being tested across distances ranging from tens of centimeters to around one meter can maintain low error rates as it expands to multiple processors and modules.
The development of a multi-processor system with at least 1,000 programmable qubits in 2027 and Starling with 200 logical qubits and 100 million quantum gates in 2029 will provide clearer evidence of how reliably IBM’s modular architecture translates into actual computing performance.
