The tech giant is moving past the noisy, intermediate-scale quantum era, focusing on building a fault-tolerant supercomputer as it looks to democratise quantum computing and speed up scientific discovery
Microsoft has declared the end of an era of noisy, error-prone quantum machines as it channels its vast resources towards building a fault-tolerant quantum computer designed to work with artificial intelligence (AI) and classical high-performance computing (HPC).
In an interview with Computer Weekly in Singapore, top executives from Microsoft’s quantum team outlined a strategy that relies on a breakthrough in a new type of qubit, a commitment to being an open platform player, and the power of AI to make quantum programming accessible to everyone.
Zulfi Alam, corporate vice-president of quantum at Microsoft, noted that the industry is at a major inflection point. “More of our colleagues are recognising that the era of NISQ [noisy intermediate-scale quantum] is more or less over, and it’s going to be the era of logical qubits,” he said.
This shift from NISQ systems to stable, error-corrected logical qubits is central to Microsoft’s work in quantum computing. The company’s confidence is bolstered by a 17-year research effort into the technology, a high-risk, high-reward endeavour that it believes has finally paid off. This has culminated in the creation of the Majorana 1 chip, which uses a new state of matter by splitting electrons to form qubits that are inherently protected from environmental noise.
“Quantum machines don’t have good fidelity because their states are not stable,” Alam said. “But when you split electrons into two halves, you’re forcing a protected state which will not be disturbed. We’ve changed the problem to where you get it right once, and just like you would with any other memory circuit, step and repeat,” he explained, highlighting the technology’s potential for scalability.
This progress has been validated by the US government’s Defense Advanced Research Projects Agency (Darpa), which selected Microsoft as one of two companies for the final phase of its programme to deliver a fault-tolerant prototype quantum computer.
Despite its progress, Microsoft does not see quantum computers as a replacement for classical machines. Instead, it envisions a hybrid model where quantum processing units act as powerful accelerators for specific problems, working alongside classical CPUs and graphics processing units.
“The quantum machine is not a standalone quantum entity,” Alam said. “It has a very powerful supercomputer sitting right next to it to control the quantum machine. It also needs an AI engine to orchestrate the workflows.”
This integration is key to democratising the technology. Srinivas Prasad Suga, vice-president for quantum and AI at Microsoft, said the goal is to abstract the complexities of quantum physics by leveraging AI, specifically tools like GitHub Copilot, to translate natural language prompts into complex quantum circuits.
“So, when you say, ‘design a new catalyst that helps increase yield by XYZ’, the AI will then design the workflow,” Alam explained, adding that the system will discern which parts of the problem run best on classical or quantum hardware, execute the computations, and synthesise the results for the user.
From chemistry to a platform for all
This is something that’s going to come to life in the next five years or so. Where AI was five years ago is kind of where we are right now. You need to start getting ready to harness the power of quantum
Zulfi Alam, Microsoft
While the killer application for quantum computing today is largely in chemistry and materials science, Microsoft is building a broad platform to serve multiple industries.
Alam noted that as a platform supplier, Microsoft’s goal is to support different types of quantum technologies, including those from third parties. The Azure Quantum platform already provides access to quantum hardware devices from partners like Quantinuum, IonQ and PASQAL, which use different technologies such as trapped ions and neutral atoms.
Microsoft’s own work has shown immense promise. Suga cited a project that used AI and HPC to slash the discovery time for a new battery material, reducing 32 million potential candidates to just 18 in 80 hours of computing time. This shows how the combination of AI and HPC is already tackling scientific challenges, with quantum poised to accelerate it further.
In pharmaceuticals, quantum machines will enable researchers to simulate interactions between large proteins and small metallic molecules to develop cancer drugs, opening “an entire new area of medicine,” according to Alam. In finance, the focus is on “small data, high compute” optimisation problems rather than sifting through large datasets.
On the security front, Microsoft is doubling down on post-quantum cryptography (PQC), which relies on new mathematical algorithms thought to be resistant to quantum computers, to protect data from future quantum attacks, a direction driven by strong demand from governments.
“The guidance we are giving customers is to make sure you get PQC on client devices, so that the data is encrypted using symmetric keys that even a quantum machine cannot break. We’re quite confident PQC works,” Alam said.
With these advancements, Microsoft’s message to enterprises is clear: the time to prepare is now. “This is not going to be 15 years from now,” Alam said. “This is something that’s going to come to life in the next five years or so. Where AI was five years ago is kind of where we are right now. You need to start getting ready to harness the power of quantum.”
Singapore is doubling down on quantum technology with two new programmes aimed at growing quantum-related talent and manufacturing capabilities in the city-state.
