Which is you
Quantum computing has been plagued with problems relating to reliability since the inception of the technology. The quantum bits, or qubits, used by quantum computers have a reputation for being highly fragile and tend to collapse at even the slightest disturbance. However, Microsoft believes it has solved the problem through the development of Majorana 2, its advanced quantum chip, and the company was not working alone.
Also read: Microsoft has built its own AI reasoning and coding models: No OpenAI, no shortcuts
A 1,000x jump in reliability
This number alone is amazing; Majorana 2's qubits are 1000 times more reliable than those in previous generations. Typically, the lifetime of a qubit, i.e. the time for which the qubit maintains its quantum state, is measured in microseconds. However, Majorana 2 provides a mean qubit lifetime of 20 seconds, with some cases reaching up to a minute. It might be best explained by an analogy made by Microsoft itself: while the average qubit lifespan is comparable to a single day of life of your cell phone battery, the qubit from Majorana 2 will last almost three years.
This achievement is due to changes in material design; while the first version uses aluminium as the superconducting material, Majorana 2 utilizes lead, which is normally used as a means of protection from cosmic rays in hospital equipment.
Where AI came in
Also read: Snowflake-Anthropic deepen AI handshake as India emerges as key market
The switch between materials was a human discovery. However, turning the Majorana 2 into reality entailed dealing with the unfathomable number of variables – fabrication parameters, measurement iterations, years of cross-disciplinary data, international research teams. This is when Microsoft Discovery came in – the AI-based platform developed by Microsoft to accelerate scientific R&D.
The AI-driven agents were able to automate the measurement process which used to take weeks per iteration, reducing the turnaround time exponentially. Some agents analyzed huge volumes of fragmented experimental data from years ago to find links that could not be discovered by humans. One particular agent noticed an uncalibrated temperature sensor which was secretly undermining all the fabrication efforts.
"Agentic AI has permeated almost everything we do," said Chetan Nayak, Microsoft Technical Fellow. The human researchers set the direction; the AI ran the legwork.
2029 is the new target
Having achieved the development of Majorana 2, Microsoft has moved up the year for their scalable quantum computing from 2039 to 2029, which is 10 years sooner than expected. The new chip's impressive attributes include qubits that last 20 seconds, operate at one microsecond and have dimensions 1/100th of a millimeter