Opinion
A quantum computer can work and still be a dead end
According to Q-Factor CEO Dr. Guy Raz, "Useful quantum computing will depend on whether control, calibration, readout and error correction can remain practical as machines grow by orders of magnitude."
Quantum computing has already shown that very different physical systems can potentially serve as working qubits. Much of the field’s development has centered on increasing the number of qubits that can be built and controlled. However, for the useful, fault-tolerant systems we are trying to build, reaching hundreds of thousands or millions of physical qubits also means that control, calibration, readout and error correction have to scale with the qubits themselves.
The next quantum hardware race will be decided by scaling overhead. An architecture can fail even when all of its individual components work, simply because operating them together becomes too complex. A control scheme that looks elegant at 1,000 qubits may become impractical at 100,000; readout can remain accurate yet become too slow. Error correction may suppress noise, yet require so many physical qubits and operations that the system never reaches useful capacity. Even modularity carries a cost: stitching together many smaller quantum cores only helps if the interconnect does not become the next scaling bottleneck in performance or system complexity.
Neutral atoms emerged as a quantum computing modality later than superconducting circuits, trapped ions and photons. Atoms can be seen as a natural realization of quantum bits, or qubits. They are intrinsically identical, can be trapped and moved with light, and can be arranged in flexible geometries that allow any qubit to interact with any other. A neutral-atom quantum computer is, in practical engineering terms, a very large electro-optical system. Those properties create a path to large arrays and optical control without a separate complex wiring stack for every qubit. In my view, these properties make neutral atoms the most promising platform for the scaling phase now beginning.
Caltech researchers have already assembled an array of 6,100 neutral-atom qubits with record quantum coherence times. A Harvard-led team has used reconfigurable arrays of up to 448 neutral-atom qubits to demonstrate repeated quantum error correction and logical operations. These large arrays and practically all key fault-tolerant operations have been demonstrated separately; combining them into a single machine at orders-of-magnitude larger scale remains the challenge.
Our work at Q-Factor builds on decades of research in ultracold neutral atoms and quantum optics at the Weizmann Institute of Science and the Technion. Most neutral-atom systems are built around a fairly conventional microscope-like design, which naturally limits how large a single processor can become. Q-Factor is developing a different approach, designed from the start around a much larger atomic core. Our architecture is designed to let a single machine scale much further without having to split it into separate modules and cores. Working on a core at the million-qubit scale also means solving a multitude of engineering challenges involved in operating such a system.
One such challenge is that two-qubit gates gradually heat the atoms as their traps are switched off and back on, which eventually limits circuit depth. To address this, we developed a refocusing method that uses existing trap controls to reverse this motional effect, avoiding the heating without adding hardware that would itself have to scale.
Measurement creates a different scaling challenge. At large scale, measurement and feedback also have to stay fast and accurate enough for useful computation. In separate work, Q-Factor researchers and collaborators propose using auxiliary atoms to enhance the readout signal in neutral-atom systems; the analysis is supported by simulations.
Scaling costs tend to pile up. Faster measurement may require more hardware, added connectivity makes control harder, and error correction increases the total resource load. A design that works well at one scale can become impractical when the machine grows by 10x or 100x. That is why Q-Factor is designing the full system around those constraints from the outset. In the race to one million qubits, the challenge is not only getting there, but making sure the rest of the machine can scale with them.
Dr. Guy Raz is co-founder, CEO and CTO at Q-Factor.
