Crosstalk between qubits scales catastrophically: EM simulations validated for only 6 qubits while IBM plans 4,158-qubit systems

When qubits are packed closely together on a chip, they interfere with each other through electromagnetic crosstalk -- unwanted quantum interactions that introduce errors. This problem grows worse with density and connectivity. State-of-the-art electromagnetic simulations used to predict and mitigate crosstalk have been validated on systems of only about 6 qubits, yet IBM's 2026 Kookaburra processor targets 1,386 qubits per chip and 4,158 qubits across three linked chips. The gap between simulation capability and hardware scale is roughly three orders of magnitude. Why it matters: Because crosstalk cannot be accurately modeled at scale, chip designers are essentially flying blind when laying out qubit architectures above a few dozen qubits, so crosstalk errors compound unpredictably as systems scale, so error correction overhead increases far beyond theoretical projections, so the number of physical qubits needed per logical qubit balloons, so the '1,000 logical qubit' milestone that would enable useful quantum chemistry and optimization problems gets pushed from the early 2030s to an unknown future date. The structural root cause is that simulating electromagnetic interactions between qubits is a classically hard problem -- the computational cost scales exponentially with the number of interacting elements. This means the very tool needed to design large quantum processors (classical EM simulation) hits a computational wall long before the quantum processors reach their target size. It is an ironic bootstrapping problem: we need quantum computers to simulate the physics required to build quantum computers.