Caltech Breakthrough Slashes Quantum Computer Requirements to Just 10,000 Qubits

Researchers at the California Institute of Technology have published a breakthrough finding that slashes the qubit requirements for a useful, fault-tolerant quantum computer by nearly two orders of magnitude — from the millions previously thought necessary to as few as 10,000 to 20,000 neutral atom qubits. The research, carried out in collaboration with Oratomic, a Caltech-linked quantum computing startup, was covered extensively by Time magazine on April 7 and is already reshaping timelines across the quantum computing industry.

The key advance lies in a new error-correction architecture that dramatically reduces overhead. Earlier fault-tolerant quantum computing designs required roughly 1,000 physical qubits for every one reliable logical qubit — a crushing inefficiency that put practical machines decades out of reach. The Caltech-Oratomic team demonstrated that their approach can achieve fault tolerance with as few as five physical qubits per logical qubit, a 200-fold improvement. "Our results reduce qubit counts by up to two orders of magnitude," said Madelyn Cain, one of the lead researchers on the project. The team also credited artificial intelligence with accelerating the development of the underlying algorithm. "There is no question that we used AI to accelerate this development," one researcher said.

The platform relies on neutral atom quantum computing, which uses optical tweezers — highly focused laser beams — to trap and manipulate individual atoms in programmable arrays. Unlike superconducting qubits, which must remain millidegrees above absolute zero and interact only with their immediate neighbors, neutral atom systems allow qubits to connect across large physical distances within the array. "Unlike other platforms, neutral atom qubits connect over large distances," said Manuel Endres, Professor of Physics at Caltech. The largest neutral atom array built to date contains 6,100 atoms, suggesting the 10,000-qubit threshold is within near-term experimental reach. John Preskill, the Caltech physicist who coined the term "quantum supremacy," said flatly: "Fault-tolerant quantum computing is now within reach."

The implications extend far beyond academic physics. A fault-tolerant quantum computer of this scale could execute Shor's algorithm — a quantum procedure that efficiently factors large numbers and threatens the RSA and elliptic curve encryption that secures virtually all internet communications, financial transactions, and government communications. The threat is serious enough that Google announced in late March a plan to quantum-proof its own systems by 2029, six years ahead of the 2035 deadline set by the U.S. National Institute of Standards and Technology. Oratomic, launched as a startup to commercialize the Caltech research, is now seeking to build the first utility-scale neutral atom machines. CEO Dolev Bluvstein said the company aims to demonstrate a commercially relevant device within the next three years.

The publication arrives at a moment of intense competition in the quantum computing sector, with Google, IBM, Microsoft, and a constellation of startups all racing toward quantum advantage. Microsoft has made a separate bet on Majorana qubits — topologically protected quantum bits announced in March 2025 — while IBM has pursued incremental scaling of its superconducting qubit systems. The Caltech finding suggests neutral atoms may have leapfrogged both architectures in the race to fault tolerance. Whether a 10,000-qubit neutral atom machine can be built before a competing technology reaches the finish line remains the defining question in quantum hardware — but for the first time, the finish line looks achievable on a human, rather than a generational, timescale.