Microsoft's Expanding Quantum Domain
November 19, 2024
In January of 2022 I wrote a post about cloud-based quantum computing access and noted three providers in the space (Amazon, Microsoft (MSFT) and QC Ware) which were offering access to systems from six Quantum Computing providers. Microsoft’s role, at the time, of providing access to systems from three providers, made them seem like an early but minor player in the quantum field. I was recently encouraged to re-look at Microsoft’s quantum efforts and was quite impressed how far they had advanced to-date, hence this post now more fully showcasing their quantum chops. During a call with Microsoft’s Chetan Nayak, a Technical Fellow and General Manager for Quantum Hardware, and Lindsay Bayne, a Director on the Growth Innovation and Strategy Team, I learned a great deal about their latest quantum efforts and partnerships, including some profound insights regarding their topological computing approach.
The graphic below provides a high-level view of Microsoft’s latest approach to quantum which includes their Azure Quantum platform, their AI and High Performance Computing (HPC) resources and their internally developed Quantum Computer efforts focused on a bold and more error resistant modality known as Topological qubits:
Graphic courtesy of Microsoft
Here is a clearer overview of each of these complementary resource areas:
1. Microsoft Azure Quantum: MSFT’s cloud-based computing service provides access to quantum hardware, software and solutions. It provides access to the Quantum Computing machines of IonQ, Toshiba, Pasqal, QCI, 1Qbit, Rigetti, Quantinuum, and Atom Computing. It also includes Azure Quantum Elements which contains Generative Chemistry and Accelerate DFT (density function theory), both designed to accelerate research in chemistry and materials science.
2. Microsoft AI and HPC: MSFT’s AI solutions including Copilot, their AI-powered digital assistant as well as their broad presence in HPC or High Performance Computing, a cloud-based solution that combines computing, networking, and storage resources with workload orchestration services.
3. Topological Quantum Computing: Although still in early development, MSFT is working on their own quantum hardware, focusing on leveraging topological quantum qubits as their platform. Topological quantum computers utilize quasi-particles known as anyons, which are braided together and are thought to provide much more robust error protection than other modalities.
According to Microsoft, they are pioneering a new computing paradigm by bringing the power of the cloud and AI together with quantum. The Azure Quantum platform enables seamless execution of quantum applications that leverages hardware across a variety of qubit architectures and chips, while offering integration with cloud HPC and AI. Microsoft now offers a synergistic assortment of compute assets that are intended to address Quantum Computing resources today, as well as provide a seamless platform to expand those resources as quantum machines and algorithms continue to evolve. Let’s dig in a bit deeper on each of these primary quantum resources.
Azure Quantum
Microsoft's Azure Quantum platform is at the forefront of making quantum computing accessible to a broad audience. This innovative cloud service allows researchers, developers, and businesses to harness the power of quantum technology without needing extensive quantum physics knowledge or needing to invest in on-premises hardware.
Azure Quantum offers a comprehensive suite of tools and services that bridge the gap between classical and quantum computing enabling users to access quantum hardware, software, and pre-built solutions through a user-friendly interface, enabling them to explore quantum algorithms and potential applications in various fields. By providing access to a number of different quantum computing modalities (e.g., trapped ions, neutral atoms and superconducting qubits, as highlighted on the above, and which will soon include Atom Computing) Azure Quantum supports multiple quantum hardware technologies, giving users the freedom to experiment with different approaches. This diversity allows for comparison and selection of the most suitable technology for specific problems.
The platform includes Q# (pronounced Q sharp), Microsoft’s a programming language specifically designed for quantum computing, which is part of their Quantum Development Kit (QDK) and is designed to simplify the process of writing quantum algorithms. Additionally, Azure Quantum offers simulators and tutorials to help users understand and experiment with quantum concepts without the need for physical quantum hardware. Specifically, Azure Quantum includes a Generative Chemistry AI feature which leverages advanced AI models trained on vast datasets of existing compounds to design and discover new molecules with desired properties. By using AI-driven workflows, scientists can generate novel molecular candidates and rapidly evaluate their potential applications, significantly reducing the time required for traditional experimental methods. They also include Accelerated Density Function Theory (DFT) which is a computational method used to solve quantum mechanical problems related to the electronic structure of molecules. Accelerated DFT offers high-speed calculations without compromising accuracy, performing simulations an order of magnitude faster than traditional DFT codes. This capability is particularly beneficial in scaling chemical discovery pipelines, allowing researchers to simulate quantum-mechanical properties at unprecedented speeds.
In addition to their own platforms and resources, MSFT also provides Azure Quantum users with access to certain third-party toolkits. For example, users can access InQuantoTM, Quantinuum’s state-of-the-art quantum computational chemistry platform which allows users to experiment with various quantum algorithms and advanced techniques on current quantum hardware, including such essential chemistry algorithms as the Variational Quantum Eigensolver (VQE) and its variants. It has been used by companies such as BMW, Honeywell and TotalEnergies for applications such as fuel cell development, materials science and carbon capture.
There are two recently described collaborations, which both highlight the diversity of the Azure Quantum approach as well as showcase the cutting-edge advances MSFT is helping its partners achieve, namely with Quantinuum and Atom Computing.
Partnership with Atom Computing
Atom Computing is one of the leading neutral atom based Quantum Computing companies, and was the first company to exceed 1,000 qubits in a gate-based Quantum Computer. Although MSFT has been collaborating with Atom for a few months, their latest announcement this week is a jaw-dropper. Namely, they have created and entangled 24 logical qubits, a new world record. Not only that, but they also demonstrated the ability to detect and correct errors while performing computations on 28 logical qubits. This is so significant that diving down a bit deeper will be useful to help describe how important this achievement is.
While the distinction between “logical” and “physical” qubits warrants its own post, suffice it to say that the creation of logical qubits is a convenient work-around for the noisiness inherent in most existing Quantum Computing platforms. Over-simplified, you can think of it as using many physical qubits to support and provide certain redundancies to create one logical qubit. Classical computers are only able to simulate quantum computers with about 50 fully entangled qubits, meaning that once we have more than 50 logical qubits, it will no longer be possible to obtain results using classical computer based emulators [1], and therefore represents the point where quantum advantage will become prevalent.
By partnering and leveraging each company’s strengths, MSFT and Atom were able to detect and correct losses of neutral-atom qubits during experiments. This enabled the team to announce the demonstration of 99.963% single-gate fidelities and 99.56% two-gate fidelities, the highest two-gate fidelities ever achieved with a neutral atom system. With this achievement Atom, with MSFT’s help, has now demonstrated all of the key ingredients required to perform robust quantum error correction, including a large number of high-fidelity qubits with long coherence times, all-to-all connectivity and mid-circuit measurement with qubit reset and reuse.
Partnership with Quantinuum
While Quantinuum’s H-Series of Quantum Computers, using trapped ions for qubits, has been publicly available on Azure Quantum since February 2021, a number of exciting new achievements showcase the benefits of this partnership. For example, this past September MSFT and Quantinuum announced the demonstration of the best performing logical qubits on record at the time, including 12 logical qubits on their 56-physical qubit H2 machine. In fact, all 12 logical qubits were entangled in a complex arrangement known as a “cat state” (or Greenberger-Horne-Zeilinger (GHZ) state for those with a bit more technical background) which improved the error rate by 22x. However, this was far more than just a technical exhibition, they used the set-up to prepare the ground state of an important catalytic intermediate, and the measurement outputs were then integrated with AI to correctly estimate results. This is the first time that HPC, AI and quantum-computing hardware have been applied together to solve an actual scientific problem. While this is not quite “quantum advantage” since the results could also be achieved using just classical computers, it’s an important step towards useful quantum results. Below is a graphic highlighting the current Quantinuum-Microsoft partnership:
AI (Copilot) and HPC Integrated with Azure Quantum
Copilot helps Azure Quantum users leverage natural language to reason through complex chemistry and materials science problems. Scientists are able to accomplish complex tasks such as generating underlying calculations and simulations, querying and visualizing data, and getting guided answers to complicated processes. It can help users write code for today’s Quantum Computers via its fully integrated browser-based interface available to try for free and with a built-in code editor, quantum simulator and seamless code compilation.
In addition to integrating AI into Azure, MSFT is bringing together all of these inter-related HPC and quantum technologies into a purpose-built platform that leverages the complementary strengths of both AI for large-scale data processing and quantum for complex calculations and unprecedented accuracy. This strong compute foundation offers a secure, unified and scalable hybrid computing environment that enables users to develop best-in-class solutions for tackling problems that are difficult or even intractable on classical computers. MSFT is integrating hardware from a diverse set of Quantum Computing providers with their quantum control, processing and error correction software and adding capabilities for Copilot assisted workflows, developer tools, classical supercomputing and multi-modal AI models. Quite an integrated and robust set of tools and solutions.
Topological Quantum Computing
Most current implementations of qubits, or the basic processing units of Quantum Computers, utilize particles such as atoms (neutral atoms or trapped ions), electrons (superconducting) or photons (photonic), but all are highly susceptible to noise and decoherence because they encode the information on the particles themselves. Topological Quantum Computing seeks to implement a more resilient qubit by utilizing something known as anyons to store quantum information. Anyons are neither fermions (atoms and electrons are comprised of fermions) nor bosons (photons are bosons) but are quasiparticles that reside in two-dimensional space. Anyons cannot intersect or merge, which enables their paths to form stable braids in space-time. With such anyons, the quantum information is encoded not in the quasiparticles themselves, but in the manner in which they are braided. This makes topological Quantum Computing more resistant to decoherence and noise which plagues other qubit implementations.
As Chetan Nayak described to me, you can think of topological qubits as being analogous to a mobius strip. Consider the following examples:
The strip on the left is untwisted, while the strip on the right has a twist. It’s possible to stretch and compress the strips without impacting whether or not there is a twist. The only way to remove (or add) a twist is to physically cut or break the strip, and then reattach the ends, which takes considerable energy to do.
A Deeper Dive (feel free to skip this section of you are not interested in further technical details of topological qubits)
Topological qubits, like those being developed by Microsoft, utilize Majorana zero modes (MZMs), which are special quantum states that can exist in certain materials under very specific conditions:
1. They exist at zero energy, hence the “zero mode” part of the name;
2. They behave as their own antiparticles; and
3. They can be thought of as “half” of an electron
They are created using a combination of semiconducting nanowires, superconductors, and magnetic fields to create conditions where MZMs can exist at the ends of the nanowires. The quantum information is stored in the collective state of pairs of these MZM’s and the non-local storage is a key to the topological protection. The graphic below shows how MZM’s are created:
Graphic: R. M. Lutchyn, J. D. Sau, S. Das Sarma, Nature 556 74, 2018
The graphic depicts a nanowire of a semiconductor, such as indium arsenide or indium antimonide, that has strong spin-orbit coupling and places it in contact with an s-wave superconductor, such as aluminum, in the presence of an external magnetic field B. The nanowire device experiences a topological nontrivial phase with exponentially decaying Majorana bound states at both ends. To perform computations, the positions of the MZMs are manipulated in a process called “braiding” which is analogous to weaving strands in a braid. The graphic below depicts topological braiding for computing:
Graphic: Nature.com, “Computing with Quantum Knots,” Graham P. Collings, Scientific American 294,56-53 (2006). Credit Georg Retseck.
The advantages of topological qubits include:
1. Error resistance – the topological nature of these qubits makes them inherently resistant to many types of errors that plague traditional qubits;
2. Scalability – Because of the error resistance, topological qubits should require far fewer physical qubits to create logical qubits, making scale-up easier; and
3. Long coherence times – The topological protection should allow quantum states to remain coherent longer, giving more time for complex computation.
Although there are many advantages to using topological qubits, there are also substantial existing challenges including that these particles are extremely challenging to create and detect. There have been a number of starts and stops regarding topological qubits including some high-profile retractions and controversies in the field, highlighting the difficulty of working with these exotic quantum states.
Microsoft’s Current Topological Efforts
Microsoft has created a roadmap to track the milestones they expect to hit along this journey as depicted below:
Microsoft has achieved this crucial first step, creating and controlling Majoranas. They are continuing to advance their systems and technology and aim to advance through these stages as promptly as they can. I look forward to tracking, and report on, their progress.
TL;DR Key Takeaways:
Microsoft has a robust and ever-growing cloud-based quantum platform that brings together quantum tools and third-party quantum hardware in a synergistic manner, and seamlessly integrates AI and High Performance Computing resources, making it a robust one-stop-shop.
Not only does Microsoft host third-party Quantum Computers for market leaders including IonQ, Quantinuum, Pasqal, Rigetti and others, it actively collaborates with these partners and augments their offering with virtualization and other tools that facilitate error detection and correction.
Examples of these partnerships include the creation of a cat state on Quantinuum’s H-Series Quantum Computers by entangling 12 logical qubits on their 56-physcial qubit H2 machine and using this to solve an actual scientific problem.
And today (11/19/24) Microsoft and Atom jointly announced they have created and entangled 24 logical qubits, a new world record. Not only that, but they also demonstrated the ability to detect and correct errors while performing computations on 28 logical qubits. Jaw dropping achievements that signal continuing breakthroughs.
Microsoft is also working to create their own Quantum Computer, utilizing topological qubits, which would be much more resistant to decoherence and noise which plagues other qubit implementations. They have created and controlled Majorana Zero Modes, which is the first step in creating such topological qubits.
If you hadn’t thought about Microsoft when considering the leading Quantum Computing players, you should begin doing so.
References:
“Atom Computing: Building Quantum Supercomputers with Microsoft,” Atom-computing.com, Sept. 10, 2024.
Bauer, Bela, “Azure Quantum innovation: Efficient error correction of topological qubits with Floquet codes,” Microsoft.com, May 5, 2022.
Ezratty, d’Olivier, “Understanding Quantum Technologies 2024,” www.oezratty.net, accessed Nov. 12, 2024.
Finke, Doug, and Shaw, David, “A Deeper Dive Into Microsoft’s Topological Quantum Computer Roadmap,” Quantum Computing Report by GQI, Sept. 21, 2023.
Langston, Jennifer, “In a historic milestone, Azure Quantum demonstrates formerly elusive physics needed to build a scalable topological qubits,” Microsoft.com, Mar. 14, 2022.
Li, Alexander, “Topological Quantum Computing,” Medium.com, May 16, 2019.
Nayak, Chetan and Bayne, Lindsay , telephonic interview conducted by the author on Oct. 31, 2024.
Nayak, Chetan, “Microsoft has demonstrated the underlying physics required to create a new kind of qubit,” Microsoft.com, Mar. 14, 2022.
Nayak, Chetan, “Microsoft achieves first milestone towards a quantum supercomputer,” Microsoft.com, June 21, 2023.
Raevenlord, “The Future is Quantum: Microsoft Releases Free Preview of Q# Development Kit”, Techpowerup.com, Dec. 12, 2017.
Svore, Krysta, “Microsoft and Atom Computing offer a commercial quantum machine with the largest number of entangled logical qubits on record,” Microsoft.com, Nov. 19, 2024.
Svore, Krysta, “Microsoft and Quantinuum create 12 logical qubits and demonstrate a hybrid, end-to-end chemistry simulation,” Microsoft.com, Sept. 10, 2024.
Topological Quantum Computer, Wikipedia.org, Mar. 23, 2006.
Yearwood, Torri, “Topological Quantum Computing,” Medium.com, Jan. 31, 2020.
Zander, Jason, “At Microsoft, we’re accelerating scientific discovery with a first-of-its-kind commercial offering,” Linkedin.com, Nov. 19, 2024.
Zander, Jason, “Microsoft announces the best performing logical qubits on record and will provide priority access to reliable quantum hardware in Azure Quantum,” Microsoft.com, Sept. 10, 2024.
[1] The author recognizes that this is an over-simplification and that many other conditions must be present before quantum machines surpass classical emulators, but the point is that this is a tremendous advancement that makes true commercial quantum advantage now seem well within reach.