The quantum computing industry has spent years measuring progress in qubit counts.

Physical qubits are the fundamental building blocks of a quantum computer, but they are fragile and error-prone. What determines the practical utility of a fault-tolerant quantum computer is the number of reliable and error-corrected operations it can perform before errors compromise the calculation.

John Preskill has helped popularize “quop” as shorthand for a quantum operation. The industry is increasingly using the terms megaquop and gigaquop to describe systems capable of approximately one million and one billion reliable logical operations, respectively.

However, quop is not a complete measure of performance, as gate speed, connectivity, parallelism, and the nature of the workload remain important. Nevertheless, it provides a more useful complement to raw physical qubit counts because it relates more directly to computational capability.

This distinction also has implications beyond the quantum processor itself. As quantum systems become more capable, they will increasingly need to be considered as components of a broader computing infrastructure. Their practical value will depend not only on the quantum processor, but also on the classical computing, networking, power, cooling and control infrastructure required to operate it.

A megaquop system represents an important threshold: approximately one million reliable operations may enable meaningful simulations of selected materials systems and quantum dynamics, as well as the testing of quantum algorithms at a scale beyond that available on today's hardware. Some chemistry applications may also become accessible, although many commercially significant molecular calculations are likely to require substantially deeper circuits.

QuEra's Libra system, on our public roadmap for 2028, is designed to provide 256 logical qubits built from more than 10,000 physical qubits at a 10⁻⁶ (e.g., one in a million) logical error rate, placing it in the megaquop class. One million reliable operations is unlikely to be sufficient for large industrial molecules, complete reaction pathways, or the deeper circuits required by many optimization and machine-learning workloads.

A billion operations changes the conversation

The next threshold is a thousandfold increase. A gigaquop system, capable of approximately one billion reliable operations, could substantially expand the range of problems that can be addressed. Deeper circuits can support more complex simulations, including systems with more electrons and interactions that are difficult to represent accurately using classical methods.

However, the transition from megaquop to gigaquop is potentially more significant than a conventional performance increase. It could bring a broader range of scientifically and commercially relevant workloads within practical consideration, including problems that may ultimately become impractical to solve classically at comparable accuracy and scale.

For data center operators, this raises a different set of questions. Quantum computers are unlikely to operate as standalone replacements for conventional computing infrastructure. Most useful quantum workloads will involve substantial classical computation before, between, and after quantum operations. Conventional CPUs and GPUs may prepare data and problem instances, manage algorithms and error correction, and process the results generated by the quantum processor.

The quantum processor therefore becomes a specialized accelerator within a heterogeneous computing environment.

This has direct implications for data center planning. Operators will need to consider the total infrastructure required to deliver quantum compute rather than the physical footprint of the quantum processor alone. Depending on the underlying architecture, this may include specialized cooling or environmental systems, control electronics, classical processing, networking, and significant supporting infrastructure.

The requirements will vary considerably between quantum technologies. Some architectures require cryogenic environments, while others operate at or closer to room temperature but rely on sophisticated optical, electronic or environmental systems. There is therefore unlikely to be a single model for the quantum data center.

The QPU doesn’t live alone

Instead, quantum systems are likely to become integrated into broader high-performance computing and cloud environments. In such an environment, the relevant measure of capacity may increasingly be useful quantum throughput: how many workloads a system can execute, how efficiently the quantum resource can be utilized, how quickly jobs can be completed, and how effectively quantum resources can be coordinated with conventional computing.

Networking and latency may also become material considerations. Workloads that alternate repeatedly between classical and quantum computation could be sensitive to the time required to move information between the two environments. The location and architecture of the classical resources supporting a quantum processor may therefore influence overall application performance.

Power and cooling present another area where assumptions based on conventional data centers may not apply directly. The quantum processor itself may not be the largest consumer of infrastructure resources. Control systems, error correction, readout, and classical processing can all contribute to the total system requirements. Conversely, improvements in quantum error correction and hardware efficiency could allow computational capacity to grow faster than the associated physical infrastructure.

This creates a new capacity-planning problem. Data center operators must consider a computational technology whose performance is advancing rapidly, whose workloads are still developing, and whose infrastructure requirements differ from those of conventional CPUs and GPUs.

Unpredictable workloads

The applications most likely to benefit first are those for which the underlying problem is inherently quantum. Materials science and chemical simulation are leading candidates because the behavior of electrons in molecules and materials is fundamentally quantum mechanical and can be computationally expensive to model classically. A concrete near-term example is computing the electronic structure of small strongly correlated molecules, such as catalytic reaction intermediates, where the number of interacting electrons puts exact classical methods out of reach. Potential applications include catalysts, batteries, superconducting materials and aspects of drug discovery. The commercial value of these applications, however, will depend on achieving sufficient scale, accuracy and economic advantage.

Optimization and certain machine-learning applications may also benefit, although the evidence for broad quantum advantage in these areas remains less established. A credible industry discussion should distinguish between demonstrated capability, plausible future applications and longer-term possibilities. There is also a fundamental limitation to forecasting the impact of higher-capacity quantum systems: we do not yet know all of the applications they will enable.

Don’t build the quantum data center yet

Historically, new computing platforms have created uses that were difficult to predict when the underlying technology was introduced. The most important applications of gigaquop-scale quantum computing may similarly emerge only once sufficient computational capacity is available.

That uncertainty is an argument for preparation; identifying potentially relevant workloads, establishing classical baselines, developing quantum expertise and testing hybrid workflows all require time. The same applies to infrastructure planning. Data center operators might not need to commit to a particular quantum architecture today, but they can begin evaluating how quantum resources could interact with their existing compute, networking, power and cooling environments.

The progression from megaquop to gigaquop therefore represents more than a milestone in quantum hardware. It is also a transition toward thinking about quantum computing as an element of the broader data center ecosystem. The organizations that prepare for that shift now, by mapping where quantum could fit alongside their existing compute, will be the ones ready to use it when the capacity arrives.