While a quantum computer could factor a 2048-bit RSA key in less than a week with fewer than a million qubits, there are problems so intractable that even the most advanced quantum hardware would fail forever.
Recent work led by Thomas Schuster at the California Institute of Technology has mathematically proved that certain tasks, including identifying exotic quantum phases of matter, are beyond the reach of any efficient quantum algorithm. These phases-which include topological states with unusual electrical currents-are the quantum analogues of everyday states such as solid or liquid water-but far more complex. In some cases, a quantum computer would have to run for billions of trillions of years to reach an answer. “They’re like a nightmare scenario that would be very bad if it appears,” Schuster told New Scientist. It probably doesn’t appear, but we should understand it better.
The implications go deeper than engineering challenges of decoherence or scalability. According to Schuster’s group, these limits may reflect boundaries in physical observation itself. Their earlier work, which was published in Science, demonstrated more efficient randomness generation in quantum systems, a capability crucial in cryptography and simulation. Yet even with such algorithmic improvements, they found that “several fundamental physical properties evolution time, phases of matter, and causal structure are probably hard to learn through conventional quantum experiments.”
The physical origin of these challenges can be understood by considering the computational complexity of quantum phase transitions. In classical systems, the determination of a phase is usually easy: one measures temperature, pressure, and other macroscopic variables. Quantum systems, however, have a state space which is growing exponentially with the number of particles, and interactions may create non-local correlations such that a simple characterization is no longer possible. For the exotic new phases of matter, the problem becomes similar to trying to reconstruct a huge multidimensional puzzle from incomplete noisy measurements-a task which, from a perspective of complexity theory, may well be inherently uncomputable in any reasonable time.
This is in marked contrast to the rapid developments concerning actual quantum computing performance: Google’s “quantum echoes” algorithm, running on the Willow superconducting qubit chip, demonstrates 13,000× speedups over classical approaches to some molecular simulations. Perturbing and reversing qubit operations allows the algorithm to pick up very weak long-range quantum correlations that aren’t accessible even with NMR. Such breakthroughs have led many to tout real-world applications in five years that range from materials science to enhancing AI models.
Yet the hardware is still far from the scale that’s needed for the most ambitious goals. Estimates for breaking classical cryptography have fallen dramatically-from 20 million qubits to less than one million noisy qubits-thanks to algorithmic optimizations such as compressed error-correction layouts and approximate arithmetic. But no current system can sustain the low-error, high-speed operation needed for such feats. Superconducting qubits, including those in Willow, offer a clear engineering path but must be cooled to near absolute zero and shielded from electromagnetic interference. Trapped-ion systems provide lower noise but face scaling challenges.
The quest to scale quantum hardware is inextricably linked to quantum error correction. Techniques like surface codes can convert millions of noisy physical qubits into thousands of high-fidelity logical qubits, but at great cost. Logical gates for sophisticated operations, such as Toffoli or CCZ, depend on “magic state distillation,” a resource-intensive process that remains one of the bottlenecks. These are tough engineering problems, but Schuster’s results emphasize that even if they get solved, some questions would still be unreachable.
From a physics perspective, the “nightmare scenario” is not so much about computational load; it is more about the nature of measurement. Quantum experiments rely on repeated sampling to reconstruct the state of a system, but some properties may be so deeply embedded in the system’s causal structure that no conceivable experiment can extract them. This fits in with general theoretical insights from quantum information science, which say that speed-ups are task-specific: whatever the most efficient quantum strategy, there are always going to be problems that resist it.
Pragmatically, these limits coexist with the urgency of preparing for Q-day: when quantum computers can break today’s encryption. While postquantum cryptography is being deployed at an ever-quickening pace, with hybrid key agreements like X25519+ML-KEM-768 now protecting over half of human-initiated internet traffic, the field must also grapple with the philosophical reality that quantum computing is not omnipotent. For science and technology enthusiasts, this duality unprecedented capability alongside immutable boundaries offers a rare glimpse into both the power and the humility of physics.
