Microsoft Unveils Next-Gen Quantum Chip Majorana 2, Boosting Reliability 1000-Fold

(Central News Agency, Taipei, June 3) Microsoft today unveiled its latest topological quantum chip, Majorana 2, at the Microsoft Build 2026 developer conference in San Francisco. The chip was developed with the assistance of Agentic AI on the advanced R&D platform Microsoft Discovery, utilizing a new material stack design. It improves the reliability of qubits by 1000 times compared to the previous generation.

With this progress, Microsoft expects to achieve a commercially viable, scalable quantum computer by the end of 2029, halving the original timeline. Such a computer would be able to solve complex problems currently considered insurmountable in areas such as global health, food supply, sustainability, and energy production.

According to Microsoft's press release, through recent advances in Agentic AI designed to accelerate scientific research processes and collaboration, the Microsoft quantum team is overcoming long-standing limitations, including reliability, speed, and scale—key challenges for applying quantum computing to real-world scenarios. For example, the new chip's qubits can maintain their quantum state 1000 times longer than the first generation, enabling more reliable computation.

While other common methods measure qubit 'lifespan' in microseconds, Majorana 2 offers an average qubit lifetime of 20 seconds, and in some cases up to one minute. Microsoft says this breakthrough is roughly equivalent to a phone's battery life going from lasting one day to lasting nearly three years on a single charge.

Microsoft also announced the general availability of its advanced R&D platform, Microsoft Discovery. The platform integrates dedicated AI agents designed for scientific research and development, a Discovery Engine that drives research and reasoning workflows, and enterprise-grade security, governance, and transparency. This allows customers to deploy teams of AI agents to accelerate scientific discovery under the guidance of human expertise. (Editor: Chang Liang-chih) 1150603