Quantum Computation:

Quantum computing represents a transformative leap in information processing. Unlike classical computing—which processes data in discrete bits (0s or 1s)—quantum computing leverages qubits that exploit the principles of superposition and entanglement to exist in multiple states simultaneously. This remarkable capability enables quantum systems to process vast amounts of information in parallel, opening new avenues for breakthroughs in complex problem-solving, optimization, and simulation tasks that are unattainable with conventional computing.

At the hardware level, physical qubits (such as superconducting circuits or trapped ions) are the tangible elements that embody quantum states. However, due to their sensitivity to environmental disturbances and operational imperfections, these physical qubits are inherently prone to errors. To ensure that quantum computations remain accurate and stable as systems scale up, advanced quantum error correction (QEC) techniques are employed. These techniques involve encoding quantum information redundantly across multiple physical qubits, thereby creating more robust logical qubits. This layered approach not only mitigates individual errors but also underpins the scalability and long-term reliability of quantum technologies.

Qubit counts have grown exponentially, which on a logarithmic axis shows up as a straight line. The solid line is a least-squares fit to the largest system built at each point in time, and its slope works out to a doubling roughly every 14 months. It is fitted to the frontier rather than to every release because the series moves in both directions: vendors ship smaller, better-connected chips on purpose, so Google's 105-qubit Willow and IBM's 120-qubit Nighthawk both land well below machines that came before them. Those are real and they are on the chart, but they are not the frontier, and a line drawn through them would say the field went backwards.

What the dashed lines are

The solid line stops at the most recent machine, because that is where the evidence stops. The two dashed lines carry it forward and are projections, not data. Both start from where the fit ends and grow at the lower bound of the fitted slope's confidence interval, which is a deliberately cautious reading of the trend rather than the trend itself. The optimistic line applies that rate in full; the conservative line halves it, which is roughly a doubling every two and a half years instead of every fourteen months.

Neither is a forecast of what any company will ship. They assume the last thirteen years of scaling continue, which is the one assumption a thirteen-year trend line cannot check. The field has already changed direction once on this chart. Nothing built since September 2025 has come close to the largest arrays, and the notable machines since then have been small: a 448-atom processor running 96 error-corrected logical qubits below threshold, and a 108-qubit chiplet system. Both are more capable than anything above them on this axis. Physical qubit count is no longer the number the field is competing on, which is worth holding in mind before reading a line that extrapolates it to 2050.

Below is every system on the chart, with a source on each builder. It counts physical qubits that have been built and held in a real machine, not vendor roadmap targets.