Physics Just Set a Hard Limit on How Many Qubits One Quantum Computer Can Have

The laws of physics have set a ceiling on how many qubits can operate in a single quantum computing module: between 100,000 and 1,000,000, depending on the architecture. Above that range, according to a Peking University paper posted to arXiv last week, a quantum processor must be split into smaller modules that coordinate with each other. Not because engineers ran out of money or patience, but because the classical systems required to control a quantum computer cannot keep pace with the quantum hardware itself.

The paper calls this the "control light cone" problem. Classical coordination — the ordinary communication needed to synchronize operations across a chip — scales with physical size. Quantum coherence, the fragile property that makes quantum computing useful, does not. The mismatch becomes impossible to engineer around once a processor grows past roughly 100,000 to 1,000,000 physical qubits, the researchers write in arXiv:2604.24059. They formalize the constraint with a scaling law: when the ratio of classical coordination time to quantum coherence time exceeds a critical threshold, the system must become modular. Physics does not negotiate.

The practical consequences are already here. Helium-3, the coolant that dilution refrigerators need to operate, is on seven-month backorder for repair services. The FPGA chips at the heart of error correction decoders — the hardware that catches quantum computing mistakes before they cascade — have been allocated to a defense contractor with higher priority, according to PostQuantum.com's quantum infrastructure survey. Cryo-CMOS, the technology that would move classical control electronics from room temperature into the cryostat's 4 kelvin stage, is the most credible proposed solution, but existing semiconductor designs were not built to function near absolute zero.

The bottleneck is not a future concern. SDxCentral reported that quantum computing's main constraint is increasingly the classical systems surrounding the qubits, not the qubits themselves. The market for quantum control electronics tells the rest of the story: roughly $84 million for superconducting systems in 2025, projected to exceed $200 million by the early 2030s, growing at 17% per year, according to PostQuantum.com. Zurich Instruments launched a new system in March 2026 supporting over 1,000 control channels per rack with microsecond-scale feedback for quantum error correction.

The quantum computing industry has spent years competing on qubit count. The Peking paper argues the number that actually matters is the number of modules, and the bandwidth between them.

This work has been posted as a preprint and has not yet undergone peer review.