⚡ electrical engineering

Transistors to transmons: large-scale cryogenic CMOS control of superconducting qubits

This paper demonstrates a scalable hybrid quantum control architecture that integrates cryogenic CMOS ASICs for two-qubit flux control with room-temperature electronics for single-qubit operations, achieving high-fidelity gate performance on a 156-qubit processor while significantly reducing system footprint and overcoming wiring constraints for large-scale superconducting quantum computing.

Original authors: Devin Underwood

Published 2026-07-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Devin Underwood

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to conduct a massive orchestra where every musician (a quantum bit, or "qubit") needs a very specific, tiny nudge to play the right note. In a superconducting quantum computer, these nudges come in two flavors:

The "Solo" Nudges: High-speed radio waves to tell a single qubit what to do.
The "Duet" Nudges: Precise magnetic currents to make two qubits talk to each other and become entangled.

The problem is that as you add more musicians to the orchestra (scaling up to millions of qubits), the "Duet" nudges become the bottleneck. You end up needing thousands of wires running from a hot control room down into a freezing cold machine. This creates a tangled mess of cables, takes up too much space, and costs a fortune.

This paper presents a solution: Moving the "Duet" controllers inside the freezer.

The Old Way: The Long Commute

Currently, most quantum computers use a "room-temperature" approach. Think of the control electronics as a giant server rack sitting in a warm office building. To reach the qubits, which live in a super-cold fridge (near absolute zero), the control signals have to travel down a long, winding tunnel.

The Problem: For every single qubit, you need a dedicated wire. If you have a million qubits, you need a million wires. This is like trying to feed a million people through a single, narrow straw. It's expensive, bulky, and the long wires distort the signals, making the music sound out of tune.

The New Way: The Cryo-CMOS "Local Manager"

IBM researchers built a new system that splits the job based on where it makes the most sense to do the work:

The "Solo" Work (Room Temperature): The high-power radio waves for single qubits stay in the warm office. They are too power-hungry to run inside the fridge.
The "Duet" Work (Inside the Fridge): The researchers created a special microchip called Cryo-CMOS that can survive inside the cold fridge. They placed these chips on a shelf inside the fridge, much closer to the qubits.

The Analogy: Imagine a large office building.

Old Way: Every employee has to walk all the way to the basement to get a specific stamp on their paperwork, then walk all the way back up. The hallways are clogged with people.
New Way: The researchers put a "stamping station" on every floor. Now, employees only walk a few steps to get their paperwork stamped. The hallways are clear, the process is faster, and the paperwork is less likely to get crumpled on a long walk.

What They Actually Achieved

The team tested this new "local manager" system on a 156-qubit quantum processor (a machine called Heron R2). Here is what they found:

It Works Just as Well: Even though the new chips are running inside the cold fridge, they didn't make the "duet" notes any worse. In fact, they were just as accurate as the old room-temperature system. The error rate was incredibly low (about 8 errors in 10,000 attempts).
It's Flexible: They tested the chips at different temperatures inside the fridge (from 7 degrees to 20 degrees above absolute zero). The chips worked perfectly across this whole range. This is great news because the "warmer" part of the fridge (around 20 degrees) has much more cooling power available, meaning you can pack many more chips into one fridge without overheating it.
It Saves Space: By moving the controllers inside, they reduced the number of wires needed by a huge amount. They showed that this approach could shrink the physical footprint of the control system by ten times.
It's Stable: They ran the system continuously for over 40 hours, and the performance didn't drift or get worse.

The Bottom Line

The paper claims that by moving the specific controllers needed for qubit-to-qubit interactions (the "flux control") from the warm room into the cold fridge, we can solve the wiring and space problems that currently stop quantum computers from getting bigger.

They proved that these "cold" chips are reliable, don't introduce extra noise, and can operate efficiently at a slightly warmer temperature inside the fridge. This creates a practical path forward to building the massive, fault-tolerant quantum computers needed for the future, without needing a warehouse full of control racks for every single qubit.

Technical Summary: Transistors to Transmons – Large-Scale Cryogenic CMOS Control of Superconducting Qubits

Problem Statement
The realization of practical, fault-tolerant superconducting quantum computers requires scaling physical qubit counts from thousands to millions. Current control architectures rely on rack-based, room-temperature electronics (RTE) connected to the quantum processor via extensive wiring. As qubit counts grow, this approach creates severe bottlenecks regarding wiring density, system footprint, cost, and thermal load. Specifically, the number of two-qubit control channels (flux-bias lines) vastly exceeds single-qubit RF drive channels, making the proliferation of room-temperature racks and interconnects a primary barrier to large-scale integration.

Methodology
The authors present a hybrid quantum control architecture that partitions functionality across temperature stages to address these scaling challenges.

Hybrid Architecture: Single-qubit control, readout, and classical processing remain at room temperature, while high-density two-qubit flux control is moved into the cryogenic environment using cryogenic complementary metal-oxide semiconductor (cryo-CMOS) application-specific integrated circuits (ASICs).
Cryo-CMOS Subsystem: The system utilizes custom-designed Flux-Control ASICs (FCA) fabricated in 14nm FinFET technology. Each 16-channel FCA integrates a digital microprocessor and a triple-DAC analog front end (static-offset, main, and fine DACs) to generate high-precision flux-bias waveforms.
Thermal Regime: The FCAs are mounted on a dedicated intermediate temperature stage ("CTRL-stage") located between the 50 K and 4 K plates of a dilution refrigerator, operating in the 7 K to 20 K range. This placement leverages higher available cooling power compared to millikelvin stages while maintaining sufficient noise performance.
Integration: The FCAs are packaged into compact Cryogenic Control Modules (CCMs) and connected to a 156-qubit Heron R2 quantum processor via high-density flex interconnects. The system employs a unified software stack where cryogenic and room-temperature controllers appear as coordinated back-ends, allowing standard tools like Qiskit to be used without modification.
Benchmarking: Performance was evaluated using Randomized Benchmarking (RB), Purity Benchmarking (PB), and Layer Fidelity (LF) protocols. Comparisons were made directly against a subset of the same processor controlled by traditional room-temperature electronics (RTE) during the same cooldown.

Key Results

Gate Fidelity: The cryo-CMOS subsystem achieved two-qubit Controlled-Z (CZ) gate errors at the $10^{-3}$ level. The median error per gate (EPG) across 95 controlled couplers was $2.2 \times 10^{-3}$ , with a best individual RB result of $8.42 \times 10^{-4}$ .
Performance Equivalence: Under matched gate settings and pulse equalization, the cryo-CMOS subsystem demonstrated statistically indistinguishable two-qubit gate fidelities compared to the room-temperature electronics.
Thermal Stability: Gate performance remained stable as the cryo-CMOS ASIC temperature was swept from 7 K to 20 K. While recalibration was required to maintain optimal fidelity at different temperatures, the system showed no measurable degradation in coherence or gate fidelity due to the elevated operating temperature.
Noise Characterization: Noise analysis revealed that electronics-induced flux noise is not the limiting factor for current gate fidelities. The integrated noise-to-signal ratio of the main DAC was measured to be $\le 100$ ppm, meeting the projected requirements for future fault-tolerant operation (targeting error rates below $10^{-5}$ ).
Long-Term Stability: Continuous two-qubit RB over 40 hours showed a drift of approximately $1 \times 10^{-4}$ EPG per day, consistent with drifts observed in the single-qubit RF subsystem, confirming the robustness of the cryo-CMOS electronics.
Scalability: Thermal modeling indicates that the intermediate temperature regime (near 20 K) allows for the support of several hundred flux-control channels within a single cryostat before exceeding thermal limits, offering a tenfold improvement in system spatial footprint compared to rack-based flux control.

Significance and Claims
The paper establishes cryo-CMOS flux control as a practical and scalable foundation for large-scale superconducting quantum systems. The authors claim that this hybrid architecture successfully decouples the scaling of control channel density from the constraints of room-temperature wiring and rack space. By demonstrating that high-fidelity gate operations can be maintained at intermediate cryogenic temperatures (7–20 K) without attenuation, the work validates a path toward denser, more compact quantum processors. The results suggest that future superconducting quantum systems can benefit from assigning control functions across temperature stages based on wiring requirements and thermal costs, rather than assuming all electronics must operate at millikelvin temperatures. This approach addresses the infrastructure bottlenecks that would otherwise prohibit the scaling of qubit counts to the millions required for fault tolerance.

The Old Way: The Long Commute

The New Way: The Cryo-CMOS "Local Manager"

What They Actually Achieved

The Bottom Line

Technical Summary: Transistors to Transmons – Large-Scale Cryogenic CMOS Control of Superconducting Qubits

More like this