How Alice & Bob’s Cat Qubit Architecture Could Accelerate Fault-Tolerant Quantum Computing
For years, the race to build powerful quantum computers looked like a simple numbers game.
More qubits meant more power.
Technology companies proudly announced systems with 50 qubits, then 100, then hundreds. The assumption across the industry was straightforward: scaling up hardware would eventually lead to machines capable of solving real-world problems.
But as researchers pushed deeper into the realities of quantum hardware, an uncomfortable truth emerged.
Brute-force scaling doesn’t work.
To build a useful quantum computer using traditional superconducting architectures, researchers would need millions of physical qubits just to produce a small number of reliable logical qubits. The engineering challenges involved—cooling, stability, control electronics, and energy consumption—would be enormous.
At the APS Global Physics Summit in Denver (March 11–13, 2026), a French-American startup called Alice & Bob presented a breakthrough that challenges this assumption.
Instead of building more qubits, they focused on building smarter qubits.
Their approach, known as cat qubits, could reduce the hardware requirements for fault-tolerant quantum computers by as much as 200 times.
If the results continue to hold, the timeline for useful quantum computing could move forward by nearly a decade.
The Hidden Challenge of Quantum Computing: Errors
To understand why Alice & Bob’s work matters, we need to start with the biggest obstacle in quantum computing: errors.
Quantum systems are incredibly fragile.
Qubits interact with their surroundings in ways that degrade their quantum states. Even tiny disturbances—thermal noise, electromagnetic interference, or hardware imperfections—can introduce errors.
Two types of errors dominate quantum systems:
Bit-flip errors
This occurs when a qubit changes state unexpectedly, similar to a classical bit flipping from 0 to 1.
Phase-flip errors
This is more subtle. Instead of changing the state, the phase of the quantum wavefunction changes, thereby altering the state’s mathematical sign.
Both types of errors disrupt quantum computations.
And because quantum information cannot simply be copied (due to the no-cloning theorem), traditional error-correction strategies from classical computing do not work.
Instead, quantum computers rely on quantum error correction (QEC).
The Cost of Traditional Error Correction
Most superconducting quantum systems today rely on a method called the surface code.
Surface codes protect quantum information by spreading it across many physical qubits.
This approach allows the system to detect and correct errors without directly measuring the quantum information itself.
But surface codes come with a massive cost.
To create one reliable logical qubit, the system may require thousands of physical qubits.
For large-scale applications—such as breaking cryptographic systems or simulating complex molecules—researchers estimate that millions of qubits may be required.
That scale introduces several serious challenges:
• enormous cryogenic infrastructure
• extremely complex control systems
• massive energy consumption
• expensive fabrication and maintenance
In other words, traditional approaches could produce quantum computers so large and costly that practical deployment becomes unrealistic.
This is where Alice & Bob’s design begins to stand out.
The Cat Qubit Idea
Alice & Bob’s technology is based on a concept known as the cat qubit, inspired by the famous Schrödinger’s Cat thought experiment.
In the thought experiment, a cat inside a box is simultaneously alive and dead until the box is opened. The idea illustrates quantum superposition—where multiple states exist at once.
Cat qubits use a similar principle.
Instead of encoding information in a single fragile quantum state, they encode it in a superposition of two stable coherent states inside a microwave resonator.
This design has an important property:
It naturally suppresses bit-flip errors.
In other words, the hardware itself is built to resist one of the two major sources of quantum error.
Traditional qubits treat both error types equally.
Cat qubits eliminate one of them almost entirely.
Turning a 2D Problem Into a 1D Problem
Because bit-flip errors are suppressed at the hardware level, the remaining challenge becomes correcting phase-flip errors.
This dramatically simplifies error correction.
Traditional superconducting systems require a two-dimensional grid of qubits to track both types of errors.
Cat qubits reduce the problem to a one-dimensional chain.
This change may sound simple, but the implications are enormous.
Less complex error correction means:
• fewer qubits required
• simpler hardware layouts
• lower energy consumption
• easier scaling
At the APS summit, Alice & Bob shared experimental data from their latest design, known as the Galvanic Cat.
The results were remarkable.
Their system demonstrated resistance to bit-flip errors for over one hour—millions of times longer than traditional superconducting qubits.
That level of stability allows engineers to focus their error-correction efforts entirely on the remaining phase-flip errors.
Elevator Codes: A New Approach to Error Correction
Alice & Bob didn’t stop at improving qubit design.
At the 2026 APS Summit they also introduced a new error-correction architecture called Elevator Codes, developed in collaboration with researchers at Inria.
Elevator Codes build on the inherent stability of cat qubits using a technique known as concatenated error correction.
In simple terms, this means layering one error-correction method on top of another.
The name “Elevator” comes from the way a logical ancilla qubit moves through different layers of the system, performing parity checks and identifying errors.
This dynamic structure allows the system to detect errors more efficiently without requiring large numbers of additional qubits.
The improvement compared with traditional architectures is dramatic.
Typical surface-code systems might require:
100,000 to 1,000,000 physical qubits to produce 100 reliable logical qubits.
Using cat qubits and Elevator Codes, the same logical capacity could be achieved with roughly:
1,500 physical qubits.
That represents a 200-fold reduction in hardware overhead.
Even more striking, the system can achieve a 10,000-fold reduction in logical error rates while increasing the number of physical qubits by only three times.
In the world of quantum error correction, that kind of trade-off is almost unheard of.
Why This Matters for Real Quantum Computers
For engineers and investors watching the quantum sector, a 200× reduction in hardware requirements could reshape the entire industry.
First, there is the issue of infrastructure.
A quantum computer requiring millions of qubits would need enormous cryogenic facilities and specialized power systems.
By contrast, a system built around 1,500 qubits could fit inside a single dilution refrigerator in a standard data center.
Second, there is time-to-market.
Alice & Bob’s roadmap targets a 100-logical-qubit fault-tolerant machine by 2030.
With traditional architectures, that milestone might remain decades away.
Finally, there is algorithmic feasibility.
Many important quantum algorithms require extremely low error rates to function correctly.
These include:
• Shor’s algorithm for cryptography
• molecular simulation for drug discovery
• materials design for batteries and superconductors
At the APS summit, researchers emphasized that Alice & Bob’s approach could enable high-fidelity magic-state preparation, a critical step required for executing many advanced quantum algorithms.
Without that capability, large-scale quantum computing remains out of reach.
A Shift in the Quantum Race
For years, quantum companies competed by announcing larger qubit counts.
But the conversation in Denver made something clear.
The race is changing.
Instead of asking:
“How many qubits do you have?”
Researchers are starting to ask a much more important question:
“How many physical qubits does it take to create one logical qubit?”
This metric determines whether quantum computers can scale into practical machines.
And right now, Alice & Bob’s cat-qubit architecture is one of the most promising approaches on that front.
The Bigger Picture
The history of computing is filled with moments when smarter architecture mattered more than raw scale.
Transistors replaced vacuum tubes.
Integrated circuits replaced discrete components.
Cloud computing replaced single-machine infrastructure.
Quantum computing may be approaching a similar moment.
Instead of building ever-larger machines, researchers are discovering that better qubit design and smarter error correction may unlock the next stage of progress.
If the results from Alice & Bob continue to scale, the path to universal fault-tolerant quantum computing could arrive much sooner than many expected.
And in the evolving quantum ecosystem, the most powerful machines may not be the ones with the most qubits.
They may be the ones with the smartest qubits.
