Microsoft Quantum has unveiled a family of new four-dimensional (4D) geometric codes, that can reduce the error rates of physical qubits by orders of magnitude to reach the level required for reliable quantum circuits. Available in the Microsoft Quantum compute platform, the error correction codes deliver a 1,000-fold reduction in quantum error rates (from 10⁻³ physical error rate to 10⁻⁶ logical error rate) while requiring 5–6x fewer physical qubits per logical qubit. These codes are designed for quantum systems with all-to-all connectivity such as neutral atoms, ion traps and photonics, directly addressing the two most critical barriers to practical quantum computing. The technology, already implemented with Microsoft’s partner Atom Computing to create 24 reliable logical qubits, could help pave the way for commercially viable quantum computers capable of solving real-world R&D challenges.
Microsoft partner Atom Computing’s neutral-atom platform provides several advantages for implementing these codes, including the ability to tightly pack qubits in arrays owing to their lack of electrical charge. The platform achieves 99.6% two-qubit gate fidelity, the highest fidelity of neutral-atom qubits in a commercial system, along with long coherence times that make them well-suited for creating logical qubits.
The advance also addresses the limitations of traditional error correction, especially 2D surface codes, which require significant overhead, often thousands of physical qubits to create a single stable logical qubit. This approach also thus tends to suffer from slow, multi-round error correction cycles that create computational bottlenecks.
Image courtesy of Microsoft
Conversely, Microsoft’s research team has developed a novel approach using four-dimensional geometry to optimize quantum error correction. By applying geometric “rotation” in 4D space, they’ve created a framework that fundamentally changes the economics and practicality of quantum computing.
In an arXiv preprint, the researchers note that the novel geometry resulted from the rotation of “the four dimensional self-correcting quantum memory,” and yields codes that are “targeted to both near-term and utility-scale quantum computers.” They go on to note that the “design achieves single-shot error correction, significant reductions in required qubits, and low-depth logical operations.”
It results in a dramatic resource reduction on the order of 5–6x fewer physical qubits required per logical qubit compared to standard topological codes. Microsoft’s new approach also enables the creation of more logical qubits with existing hardware while making quantum systems more economically viable for research organizations.
Unlike 2D codes requiring multiple correction rounds, 4D codes extract error information in one constant-time cycle, yielding significantly faster logical operations and higher computational throughput. The approach also cuts the time quantum information must be maintained, thus lowering overall error rates.
Recently, Atom Computing demonstrated improved mid-circuit measurement with qubit reset and reuse on their hardware. This technique allows for the measurement of a subset of qubits while preserving both measured and unmeasured qubits—a capability that improves error correction, increases the complexity of quantum computations that can be performed, and paves the way for fault-tolerant quantum computing.
Earlier this year, Microsoft debuted Majorana 1, which it dubbed as “the world’s first Quantum Processing Unit (QPU) powered by topological qubits.” At that point, the company noted that it had already placed eight topological qubits on the chip designed to house one million, as Technical Fellow, Advanced Quantum Development at Microsoft Krysta Svore noted at the time. At least one physicist, University of Basel’s Henry Legg, however, has questioned the company’s framing in an ArXiv preprint.
As Svore explained in a blog post, Microsoft plans to scale these capabilities significantly: “Incorporation of these codes into our full-stack will enable the creation and entanglement of 50 logical qubits in the near term, with the potential to scale to thousands of logical qubits in the future.” The technology is part of Microsoft’s broader quantum ecosystem that includes integration with their Discovery platform for scientific research, cloud HPC, and advanced AI models.
As Microsoft positions 2025 as “the year to become Quantum-Ready,” the company is emphasizing practical applications where quantum computing enhances existing technologies like HPC and AI, rather than replacing them entirely.
