Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

The architecture wars

Why quantum's "winner-take-all" narrative is flawed.

[Photo: Getty Images]

Profile photo of Yuval Boger

Chief Commercial Officer · QuEra Computing

Every major computing shift eventually sparks an argument about architecture. The arguments tend to start early, get emotional fast, and usually miss the point. Quantum computing follows the same script.

Today, the debate over quantum hardware is usually framed as a contest between rival approaches. Quantum computers are still-experimental machines that tackle certain problems very differently from ordinary computers. But there’s more than one way to build one, and the field is arguing over which way is best.

Some designs use superconducting circuits chilled to near absolute zero, and others trap individual atoms or ions and steer them with lasers or electric fields. Others use particles of light. It’s an old argument wearing new clothes: Which approach will win? Which one will scale? Which one should customers bet on now?

These questions feel practical, but they’re premature. Architecture debates are about understanding the tradeoffs, the limits, and the paths to scaling up before reality forces the answer.

Never just one winner

History is instructive here. Classical computing developed specialized categories to answer specific technical requirements. Today, CPUs and GPUs coexist alongside specialized chips built for narrower jobs, like running AI models efficiently or handling specific types of data in real time. Memory hierarchies and programming models emerged because they best answered the requirements of modern hardware and data.

Quantum computing is likely to evolve the same way. The challenge is resisting the urge to turn architecture into ideology along the way.

The most common mistake in today’s debate is treating qubit count as the dominant metric. A qubit, the basic unit of information in a quantum computer, is the counterpart to a classical bit, but far more fragile.

Individual qubits are noisy and error-prone, so a raw count says nothing about the reliability or amount of computation they can actually deliver. A quantum system’s usefulness depends far more on how those qubits behave together, how reliably they can be controlled, and how errors scale as the system grows.

This is where architecture becomes real. Different quantum modalities embody different assumptions about scaling. Some designs are built for raw speed but sacrifice flexibility. Others are more adaptable but slower. Each pushes its complexity somewhere else: extreme cooling, complex wiring, or the software layered on top. None of these choices are free.

What actually matters as systems grow

What matters is which tradeoffs can be managed in practice. Can the system scale without introducing exponential operational complexity? Does performance degrade predictably as the system grows, or does it fail in abrupt and opaque ways? Are its problems predictable enough to plan around, or do new ones keep appearing as they grow?

These questions tend to be underrepresented in public discussions because they are hard to explain and harder to demonstrate. But they are exactly what customers and developers eventually care about.

From a user perspective, architecture shows up in workflow. How much extra work does it take to translate a business question into something the machine can compute? Is the system’s tuning managed well enough so that the system behaves consistently day to day, or does it need constant upkeep to stay reliable?

From an operational perspective, architecture shows up in reliability. Can the system run continuously for extended periods? Can it be maintained and serviced without heroic intervention? Can trained engineers operate it, or does it still require inventors and a small group of specialists who understand its quirks?

These concerns are the dividing line between experimental systems and production infrastructure.

At QuEra, we work with neutral-atom technology, which comes with its own architectural profile. Individual atoms are held in place with light and can be moved and rearranged on demand, more like rearranging seats in a room than being stuck with a fixed layout. That flexibility suits certain kinds of problems well, and certain approaches to catching and fixing errors as the system grows.

It also brings its own engineering challenges: precisely controlling individual atoms, keeping that control accurate as the system scales up, and building it all into one dependable machine. We’re not done yet, but the center of gravity is shifting as the remaining work is increasingly engineering, even as important science questions remain.

The point is that every modality must translate its physics advantages into operational reality. That translation is where most platforms succeed or fail.

What will happen next

For cloud providers, high-performance computing (HPC) centers, and enterprise users evaluating quantum platforms, the right posture today is structured experimentation. Supporting multiple architectures while demanding rigor forces the field to learn faster. It also reduces the risk of premature lock-in to a platform that looks impressive in isolation but struggles at scale.

Governments and funding agencies face a similar challenge. Architecture diversity at this stage offers insurance and exploration, because with multiple hard bottlenecks, there’s no guarantee that a single architecture will resolve all in the near term. Funding multiple approaches makes sense, as long as funders keep demanding real evidence of progress in stability, usability, and reproducibility.

Over time, the architecture wars will narrow. Certain approaches will prove easier to industrialize. Others will hit scaling walls that are hard to overcome. Software abstractions will mature, hiding some hardware differences while exposing others. Hybrid classical-quantum workflows will emerge, treating quantum systems less like standalone machines and more like specialized accelerators.

That is how real computing platforms evolve, through years of disciplined engineering and user feedback.

The architecture wars will end when quantum systems stop being arguments and start being dependable tools. That is a much harder outcome to achieve, and a far more meaningful one.

Yuval Boger is chief commercial officer at QuEra Computing.