Microsoft Announces Majorana 2 and Predicts the Arrival of Useful Quantum Computers in Three Years. But the Focus is on AI
Microsoft has announced Majorana 2, the second generation of its quantum processor based on Majorana fermions, which promises hundredfold improvements in reliability compared to the first model, announced last year. The announcement is striking not just for the (impressive) improvements in the Majorana chip but also for the significant emphasis placed on artificial intelligence (AI). In fact, four-fifths of Microsoft's announcement are dedicated to discussing the contribution that AI can provide to scientific research, materials development, and the calibration of quantum computers.
Microsoft Majorana 2 and Quantum Computing Predictions
Microsoft announced the first Majorana chip last year, promising to build a quantum computer capable of performing useful calculations (inaccessible to classical computers) by 2035. The uniqueness of the chip lies in the use of so-called “topological qubits.” The device is constructed using superconductors that can create what are referred to as “Majorana zero modes”, or Majorana fermions: particles that are their own antiparticles. The advantage of this approach is that the quantum information is contained within the topology of the particles, making it more resilient to environmental influences that lead to the loss of quantum properties and errors in calculations.
The announcement of Majorana 2 states that Microsoft has achieved a significant improvement in chip performance, with an average qubit lifespan of 20 seconds and up to a minute in some cases. This result is significant, considering that alternatives using superconductors with Josephson junctions, such as those used by IBM and Google, last only a few hundred microseconds (meaning 100,000 times less). Microsoft compares this advancement to phone batteries: instead of lasting a day on a charge, it is as if they have invented a battery that lasts nearly three years.
This leap forward has also been made possible thanks to a change in the materials used in manufacturing the chip. Microsoft has employed a lead-based superconductor, a material known for its ability to shield against radiation. The company claims that “the critical parts of the Majorana quantum devices are designed atom by atom,” adding impurities to the chips to keep the correct atoms in the right positions.
All of this leads Microsoft to make a highly relevant statement: the first quantum computers with practical applications will arrive in 2029. This is a significant acceleration from the roadmap the company shared last year, which projected such devices to arrive by 2035. “We need to make improvements every year to bring us closer to creating a computer that we believe will have enormous commercial and social value,” said Chetan Nayak, a technical fellow at Microsoft. “We need to continue marching according to this schedule to get there, but where are we compared to last year? We are at a point that is a thousand times better.”
However, there are some criticisms. The scientific and academic world remains skeptical that Microsoft has actually discovered Majorana fermions as claimed. Critics argue that the phenomena observed by Microsoft could be explained otherwise, and thus there is no real evidence that the chips exploit such phenomena. It remains to be seen whether the new chip will bring about changes in this regard.
Artificial Intelligence at the Center, Even in Quantum Chips
As previously mentioned, perhaps the most interesting aspect of Microsoft’s announcement is the strong emphasis placed on artificial intelligence. The announcement itself contains extremely important and relevant information: the fact that Microsoft sees the arrival of the first quantum computers with sufficient power and reliability to perform practical tasks in three years instead of ten is, in itself, almost revolutionary. But all the emphasis in the announcement is on artificial intelligence.
Microsoft claims that a substantial part of the development of the new chip has occurred thanks to AI agents used by researchers, and the announcement almost seems like an advertisement for Microsoft Discovery, the AI service designed to accelerate scientific discoveries. In fact, the company appears to treat the announcement as an opportunity to say “our system works, to the point that we’ve used it ourselves.”
And it is with this lens that the entire discussion is approached: it is agentic AI that has helped manage the discovery of materials for the fabrication of Majorana 2; it is agentic AI that has found a way to use such materials; it is agentic AI that has enabled various teams of researchers scattered across the globe to communicate and share data; it is agentic AI that has accelerated experiments; it is agentic AI that has aided in the calibration of quantum devices; it is agentic AI that has filtered experimental data to allow researchers to focus on important data. All the section titles in the announcement begin with “Agentic AI.”
This approach by Microsoft, which could almost be described as obsessive, can be read in two ways: on one hand, there is the positive contribution that AI is indeed making to research and is already reshaping how many studies are conducted in the academic world; on the other hand, there is the fact that Microsoft has been repeatedly invoking the mantra “AI, AI, AI” for years, trying to place it wherever possible. With the risk, as in this case, of eclipsing other information that is at least equally relevant, if not more so.