Quantum Computers: IBM, Google, Intel, Microsoft

It’s Quantum Wednesday! I decided to write a weekly article on quantum computing, such an exciting field in rapid R&D acceleration. This is the third article following Quantum Computers: Optimism and Pessimism and Quantum Computers: Models of Computation.

This week I will review the players investing R&D in quantum computing to contrast their approach, current state and future goals. Specifically, I will focus on tech leaders like IBM, Google, Intel, and Microsoft.

IBM

IBM’s research focuses on superconducting qubits, an approach that leverages IBM’s pre-existing experience with semiconductors and silicon wafer technology.

• Computation Model: Universal Gate Model
• Modality: Superconducting Qubits
• Maximum qubits: 50 physical qubits (announcement, pictures)
• Near term applications research: quantum simulation, quantum heuristics for optimization, quantum machine learning

IBM Q Experience is a cloud-hosted service to experiment with a 5-qubit quantum computer. Quantum algorithms can be designed with a graphical user interface (1) or through the Qasm code (2). Then the algorithm can be simulated (3) or scheduled to run in the quantum computer (4). The following series of image shows a very basic experiment that creates entanglement between two qubits so that the state of the system cannot be determined by looking at individual qubits. Specifically, it creates the Bell state, or EPR pair.

The final step is running in a real quantum computer. You can observe the noise and imperfection in the results as the states 01 and 10 are observed on some of the runs. This simple circuit has more than 10% error. This illustrate the critical importance of quantum error correction that is needed to make error-less logical qubits versus error-prone physical qubits.

Google

Google’s research focuses mainly on Bristlecone, a quantum processor based on superconducting qubits. The purpose of this system is to provide a testbed for research into system error rates and scalability of Google qubit technology, as well as applications in quantum simulation, optimization, and machine learning.

• Computation Model: Universal Gate Model
• Modality: Superconducting Qubits
• Maximum qubits: 72 physical qubits (announcement)
• Near term applications research: quantum simulation, discrete optimization, quantum machine learning
• Focused verticals: automotive, aerospace, chemistry, pharmaceuticals, defense

In another initiative, Google participates in a Quantum Artificial Intelligence Lab (NASA, Google, USRA) to experiment with a 1,097-qubit D-Wave quantum annealer. The goal is to explore the potential for quantum computers to tackle optimization problems that are difficult or impossible for traditional supercomputers to handle.

• Computation Model: Quantum Annealing
• Qubits: 1,097 D-Wave system
• Near term applications research: discrete optimization

Intel

Intel’s research focuses both on superconducting qubits and spin qubits, an alternative structure that leverages Intel’s leading expertise in manufacturing silicon transistors. The reason Intel is researching this alternative is that superconducting qubits are quite large physically and thus require big coolers which makes it more difficult to scale the technology. Spin qubits are small and strong, can work at higher temperature, and resembles the knowhow of traditional transistor manufacturing.

• Computation Model: Universal Gate Model
• Modality: Superconducting Qubits
• Maximum qubits: 49 physical qubits (announcement)

Microsoft

Microsoft’s research is along the completely different modality of topological qubits to solve the inherent problems of other qubits modalities such as superconducting qubits. These other approaches are very sensitive to noise and require very complex error correction. “One of our qubits will be as powerful as 1,000 or 10,000 of the noisier qubits,” Microsoft’s Julie Love, director of quantum computing business development. This means that scaling a commercial quantum computer with 1000 logical qubits will not require millions of physical qubits like other avenues of research.

Creating such a topological qubit is Microsoft’s objective for 2018 and they believe that scaling up the system to commercial size can be done in the next few years. In the latest development, Microsoft demonstrated that they can create half-electron quasi-particles that are the basis of topological qubits. The next step is braiding this basic unit in order for the system to behave as a full qubit.

Microsoft also focuses on a full stack mindset by developing not only a quantum development kit but also the Q# programming language that can be used to code quantum algorithms in a modality-agnostic language.

Who do you think will win the quantum race? Next week we will explore what startups such as IonQ, Rigetti, QCI, and D-Wave System are researching in this space.