Microsoft unveils Majorana 2 quantum chip with 1,000x gain

Microsoft has unveiled Majorana 2, its latest quantum chip, and says its qubits are 1,000 times more reliable than those in the previous version.

The announcement marks a step in Microsoft's long-running effort to build a scalable quantum computer. It now expects to reach that goal by 2029.

Majorana 2 uses a new materials stack and topological qubits designed to hold their quantum state far longer than earlier versions. Microsoft said the mean qubit lifetime is 20 seconds, with some lasting as long as one minute, compared with conventional approaches that measure qubit lifetimes in microseconds.

That improvement addresses one of the main obstacles in quantum computing: keeping qubits stable long enough to carry out useful calculations. Microsoft also said the chip operates at one microsecond and uses qubits measuring 1/100th of a millimetre.

Chetan Nayak, Technical Fellow at Microsoft, described the scale of the change in the company's own terms. "We need to make improvements each year that will get us closer to delivering a computer that we believe will have massive commercial and societal value," he said. "We've got to keep marching to that roadmap to accomplish that, but where are we relative to last year? We're 1,000 times better."

AI in research

Alongside the chip, Microsoft said its Microsoft Discovery platform is now generally available to organisations using AI in scientific research and development. The platform includes AI agents for research tasks, a reasoning engine for workflows, and controls for governance and transparency.

It also introduced an early preview of a Microsoft Discovery app that users can run locally on their computers with a GitHub Copilot account. Microsoft said the aim is to lower the barrier to advanced AI-led research tools.

Microsoft's quantum team has already been using the same AI tools in the Majorana programme. According to the company, the systems have managed workflows, automated measurements, refined fabrication, identified flaws and suggested possible solutions.

"Agentic AI has permeated almost everything we do-it's just become kind of a very natural part of our workflow," Nayak said. "The agents can really accelerate things as much or as little as you want. It can be as little as pulling information together and summarizing it, or it can go further down the road of synthesizing it more or generating an interesting hypothesis. I think that's extremely powerful right now."

Materials change

A significant shift in Majorana 2 lies in the materials used in the chip. Majorana 1 relied on aluminium in its superconductor, while the new version uses lead.

Microsoft said lead helps shield fragile qubits from disturbances that can make them unstable, though it took years to manage the trade-offs involved in adopting it. The change came after the team revisited the original design following the earlier proof of concept.

"That was actually a fairly large change, and it led to big, big improvements in device quality," Nayak said.

The work also depends on precision at the atomic level. In some cases, impurities are added to a crystalline structure to hold atoms in the right place, but too much can disrupt the material and alter the energy structure in unwanted ways.

Zulfi Alam, Corporate Vice President for Quantum at Microsoft, said AI modelling can narrow the field before laboratory work begins. "Finding the exact recipe, the right amount to put to get the desired energy structure, requires a lot of experimentation in the old world order. In the new world order, through simulations, you can see where the highly probable target is. And then with that knowledge, you ideally only have to experiment once," he said.

Data and automation

The quantum project spans software, architecture, device design, fabrication and measurement, with changes in one area affecting several others. Microsoft said AI agents help researchers track those interdependencies across a programme built over nearly two decades.

Alam said the team had large volumes of data stored in different formats and locations before newer AI tools made it easier to extract patterns across them. He said the systems can identify links that individual researchers may miss because no one person has a complete view across all disciplines and datasets.

Measurement has been one of the most difficult parts of the work. Creating a topological state requires setting hundreds of parameters, and Microsoft said those steps can take weeks when handled manually.

According to Alam, newer AI agents built with Microsoft Discovery reduced that cycle time sharply by automating adjustments and mapping operating conditions in ways a single scientist could not manage alone. "Using agentic AI to automate the measurements was a game changer," he said. "It goes through some math and starts saying, 'Hey, where do I find the lowest point where everything sort of works?' And it can do all these voltage adjustments in parallel, which a human cannot do. The way our minds work, we are more linear."

Microsoft said AI has also helped identify problems in fabrication by filtering noisy raw data and highlighting anomalies. In one case, an AI agent combined physics, device and institutional knowledge to detect an uncalibrated temperature sensor reading that had been affecting results.

Alam said researchers still make the decisions, with AI acting as a guide rather than an authority. "It's always 'scientist in the loop'," he said.