But the Azure Quantum team determined early on that tackling these urgent, real-world problems will require quantum computers that employ a million qubits or more. To date, public demonstrations of gate-based quantum computing have used fewer than 130.
And Microsoft’s experts expect that many of today’s qubits have limitations that will make it difficult to achieve the scale necessary to support commercial quantum applications.
That’s why Azure Quantum has focused on developing topological qubits, which are expected to be faster, smaller and less prone to losing information than other types of qubits currently under development.
Microsoft believes creating a more stable topological qubit is the clearest and fastest path to building an industrial-scale quantum machine.
But, until now, the downside of pursuing a topological qubit was that no one was sure it was possible to harness the underlying quantum physics to produce them.
“The fact that we have done this thing that’s very, very hard and can now make devices that produce this topological phase shows that we have a very talented team that is up to the challenge and can tackle the next critical steps,” said Chetan Nayak, a Microsoft distinguished engineer who leads the quantum hardware program.
“This proves the key aspects of this elusive physics and now it’s full steam ahead to the topological qubit,” he said.
Chetan Nayak, Microsoft distinguished engineer who leads Azure Quantum’s hardware program. Photo by John Brecher for Microsoft.
A high-risk, high-reward approach
The quantum industry is currently pursuing many different approaches to developing qubits. When qubits can be maintained in an optimal state, a quantum computer can theoretically exploit the peculiarities of quantum mechanics — like superposition, entanglement and interference — to solve certain problems with many variables and possible solutions in a fraction of the time it would take classical computers.
But no quantum computer yet exists on a scale that can deliver on the promise of solving complicated real-world problems.
Today’s Azure Quantum customers have been able to realize early yet meaningful benefits from quantum technology, such as using quantum principles in classical algorithms to speed up optimization solutions.
They can also begin to learn how to program durable quantum solutions that can be tested and run on the industry’s current generation of quantum hardware.
Every decision and every investment in Azure Quantum has been focused on one long-term goal: developing a quantum machine and supporting an ecosystem that allows Azure customers to solve real-world, enterprise-scale problems with the technology.
This quantum machine is designed to work hand-in-hand with Azure’s classical computing resources to offer customers new capabilities. For example, a chemical company might be able to design new catalysts in a matter of weeks rather than decades in a lab. Scientists may be able to unlock nature’s secrets to more sustainably harvest light and improve photovoltaics for cleaner energy.
Whoever develops a commercial quantum accelerator first will have a strong competitive advantage, along with their customers, Alam said. It’s yet another way that Azure plans to continue to deliver best-in-class cloud services and empower its enterprise customers to deliver breakthroughs in their industries.
For example, a million-qubit quantum computer should be able to accurately simulate complex molecules in pursuit of new chemical catalysts that a classical computer even the size of the entire solar system would not be able to model, Microsoft experts say.
“This is the next big breakthrough in computing — there’s no confusion about that in the minds of the corporate world,” Alam said.
But to build a commercially useful quantum computer, its qubits need to perform well across three key dimensions: reliability, speed and size.
Quantum states are, by their nature, extremely fragile and prone to disruption, making it difficult to maintain a state where qubits can perform computations reliably. To offer advantages over classical computing, qubits also need to process information quickly.
And a quantum machine’s components can’t be so large that they would fill a warehouse or a football field, which will make systems built on certain types of qubits difficult to scale.
“You can go build a qubit — that’s not a problem. But we know that to get to millions of qubits working together, which is what’s really needed to unlock new materials and do the practical applications that we want to do, you need to nail these three things at the same time,” said Lauri Sainiemi, Microsoft’s general manager for fabrication.
Lauri Sainiemi, Azure Quantum’s general manager for fabrication. Photo by John Brecher for Microsoft.
One challenge in developing a quantum computer is that qubits readily collapse and decohere when they encounter environmental noise such as heat, stray subatomic particles or magnetic fields. Information is lost, and the qubits are no longer useful for computation.
Errors start to occur, and the quantum computer has to devote even more unreliable qubits to correcting them. It’s like trying to keep an entire room of plates spinning on chopsticks when the smallest perturbation can cause one plate to become unbalanced and start crashing into all the others.
Microsoft’s approach has been to pursue a topological qubit that has built-in protection from environmental noise, which means it should take far fewer qubits to perform useful computation and correct errors.
Topological qubits should also be able to process information quickly, and one can fit more than a million on a wafer that’s smaller than the security chip on a credit card.
To create topological protection, quantum information can be encoded in a pair of Majorana zero modes that are physically separated. This makes a topological qubit more immune to environmental noise, which can’t interact with or destroy the information when it encounters just one.
The only way to unlock the quantum information is to look at the combined state of both Majorana zero modes at the same time. Taking these measurements in a strategic way enables both quantum operations and creates inherent protection for the qubit.
But first, the Azure Quantum team needed to demonstrate how to reliably create the topological phase that confers these stability, speed and size advantages. They developed a process that layers semiconducting and superconducting materials onto a device in an extremely controlled and atomically precise way.
In the presence of specific magnetic fields and voltages, the devices can produce a topological phase with a pair of Majorana zero modes — characterized by telltale energy signatures that will appear at either end of a nanowire under the right conditions — and a measurable topological gap.
When exploring what architectures would meet the requirements to run practical quantum applications, Microsoft’s quantum experts came to believe that a topological qubit was the only building block that checked all three boxes for a quantum computer that could achieve the necessary scale for practical use.
But they also knew that deciding to invest in this challenging topological approach was a little like choosing to climb a mountain straight up from the trailhead to ultimately be rewarded with an easier walk along the ridgeline, rather than taking the easier path of staying in the valley only later to reach a cliff that blocks upward progress.
“Microsoft has taken this very risky but high reward approach in trying to make a qubit which on the theory side looks like the very best qubit you can get. But the challenge was that nobody has really seen these Majorana zero modes in real life,” said Peter Krogstrup, scientific director of Microsoft’s Quantum Materials Lab in Lyngby, Denmark.
“But we have done that now, and that’s super exciting. We have to continue to evolve our engineering capabilities, but it really looks like there is a path towards scalable quantum computing now.”
Microsoft Hardware Engineer Ajuan Cui and Senior Researcher Mohana Rajpalke collaborate inside Microsoft’s Quantum Materials Lab. Photo by John Brecher for Microsoft.
‘It was suddenly wow’
Roman Lutchyn remembers eating lunch in a hotel when he got an email last year from a colleague who had been asked to analyze the measurements from an experiment on the team’s newest device design.
Earlier, they had worked with quantum experts to develop a checklist of all the things they’d need to see in the data to convince themselves that they’d truly achieved the topological breakthrough.
Lutchyn, a Microsoft partner research manager with expertise in quantum simulation, picked this colleague to analyze the data because he had historically been a healthy skeptic on the team.
He also hadn’t been involved in designing or testing this particular device, which involves sending an electrical current through the system and seeing how the materials respond. This time, the colleague agreed, the data checked all the boxes they’d been looking for.
There were a pair of telltale energy signatures called zero bias peaks, which indicate the presence of Majorana zero modes at both ends of a nanowire that’s been tuned into a topological phase. Previously, that signature had only been seen at one end of the wire and not in combination.
There was also another pattern in the electrical conductance data that provided evidence of a topological gap, which is measurement that quantifies how immune the topological phase is to environmental disturbance.
The team needed to see the gap closing and reopening — along with the simultaneous appearance of the two zero bias peaks — which they did clearly for the first time.
“It was suddenly wow. We looked at the data, and this was it,” Lutchyn said.
In consultation with outside experts in the quantum field, the Azure Quantum team wanted to set the highest possible bar and clearly establish what objective criteria would show that they’d established the long-sought-after topological phase.
In particular, they wanted to avoid uncertainties such as those that led to the retraction of a 2018 Nature paper in which authors seeking Majorana zero modes relied on data that turned out to be incomplete or misleading.
That’s why, Alam said, the hardware team has invited an external council that includes some of the world’s leading experts in the quantum field to review the latest results in detail and offer feedback and validation for the discovery.
The Azure Quantum team understood that simply seeing one piece of evidence in isolation would not be sufficient. But they say the accumulation of data from their latest device designs — seeing all the patterns they’ve been looking for in conjunction with one another and on multiple devices — makes a much more compelling case.
“If you only see bits or pieces, it can be hard to tell what you’re looking at,” said Judith Suter, a Microsoft senior researcher who works in the Quantum Materials Lab.
“If you find a single bone in the desert, it’s tough to tell what animal it came from. But if you find a whole skeleton put together, you can look at it and say, ‘Ah yes, that’s a fox.’”
Roman Lutchyn, Microsoft partner research manager who leads Azure Quantum’s simulation team. Photo by John Brecher for Microsoft.
From experimentation to industrial design
In the last year, the Azure Quantum hardware team has moved from a largely experimental approach — testing theories in the lab and learning by trial and error — to simulating, designing and engineering materials with specific requirements for optimal performance.
“We are not motivated by scientific discovery alone. We are in the business of building products that deliver value and empower our customers to do the once unimaginable,” said Alam, who helped drive a cultural shift across the program that many agree has helped accelerate the team’s recent progress.
“Building a quantum computer is similar to sending someone to the moon or adventuring to Mars. It has the same level of complexity — or more — and requires a team of experts all working very closely together, where the mission is much greater than the individual parts,” Alam said.
For years, the company’s quantum research relied on largely academic approaches that encouraged multiple teams to test whatever theories they saw as most promising, based on deep expertise in quantum physics but also a little bit of intuition and guesswork.
It requires setting up and running experiment after experiment in a lab, which can be time consuming and sometimes make it difficult to quickly isolate what contributed to a success or failure.
Using Azure’s massive computing capabilities, Lutchyn and other researchers at Microsoft’s Station Q lab in Santa Barbara have developed new quantum simulation capabilities to complement the team’s invaluable academic research.
This now allows the hardware team to model and predict how different device designs — from the materials that are used to the dimensions of each component to how qubits can be linked together — influence quantum behavior.
This ability to iteratively test different scenarios and tweak individual parameters in simulation has allowed the team to isolate what characteristics are the most important drivers of performance.
“This brings the program to the next level because it’s switching from an experimental and scientific approach to more of an industrial and engineering approach,” Lutchyn said. “We now have far more consistency.
You can say ‘this is the recipe and here are the specs you need to hit’ and then more predictably you get what you expect to see.”
Experts at Microsoft’s Copenhagen Quantum Materials Lab and elsewhere have also spent the last several years inventing or optimizing fabrication techniques that now allow them to engineer and make devices with atomic level precision.
Figuring out how to assemble key elements of the latest device in a high vacuum environment has also allowed hardware teams to achieve purity levels that were impossible with conventional fabrication techniques.
These and other fabrication advances were also instrumental in realizing Microsoft’s latest breakthrough, the team says, allowing the people making the hardware to match and physically realize the ideal specifications generated by the design and simulation team.
“We are now led by designs that are based on simulations, not just someone batting ideas around in a conference room,” said Nayak. “And now we have the unique growth and fabrication technologies to bring those ideas to life.
It doesn’t matter if you have the best designs in the world — if you can’t make them, they just stay on paper.”
Peter Krogstrup, scientific director of Microsoft’s Quantum Materials Lab in Denmark. Photo by John Brecher for Microsoft.
On the qubit engineering path
To be clear, Microsoft’s quantum leaders say, there is far more challenging work ahead on the path to creating a scalable quantum computer.
But these same simulation, design and fabrication capabilities will continue to benefit the Azure Quantum team as they tackle next steps: figuring out how to make a topological gap more robust and stable, entangling Majorana building blocks to make a qubit, processing information with qubits that can perform meaningful computation and connecting qubits that must operate at temperatures colder than outer space into a scalable machine.
But the most important scientific question mark has now been erased, the team says. And the next set of problems on the horizon, while still difficult, lie in slightly less uncharted territory.
“There’s no fundamental obstacle to producing a topological qubit anymore,” said Sainiemi. “This definitely doesn’t mean that we’re done — we still have tons of work to do. But the fundamental part has been demonstrated, and now we’re on more of an engineering path and that’s what we’ll continue to pursue.”
Top image: Postdoctoral researcher Xiaojing Zhao works in Microsoft’s Quantum Materials Lab, where an important milestone towards creating a topological qubit and scalable quantum computer has been demonstrated. Photo by John Brecher for Microsoft.