It is reported that during 2025, Atom Computing and Microsoft will deliver the first 1000-qubit quantum computers to customers. These will be local systems on cold neutral atoms, for which Microsoft has created a software platform for organizing hybrid quantum-classical calculations, as well as basic error correction algorithms. These developments will make it possible to create an array of 50 logical qubits on a computer for arbitrary calculations.
Image source: Atom Computing
Atom Computing emphasizes that they have developed a local computing system available for direct use by customers. Qubits are created from cooled ytterbium atoms. The atoms are brought to a temperature close to absolute zero in one vacuum chamber, and then, using optical tweezers, are transferred to a second vacuum chamber, where they are held in traps in configurations created according to a given algorithm.
Atom configurations are a circuit of connected gates—logical structures designed to process the states of input qubits. In this year’s 1000-qubit system (more precisely, 1200 qubits), developers propose using about 50 logical qubits to create a circuit for executing a quantum algorithm. It’s important to note that a number of competing companies use cold neutral atoms to simulate quantum processes. This is also a kind of calculation, but such platforms cannot be called universal. Atom Computing initially found this approach unacceptable and organizes its neutral cold atoms into gates.
«Atom Computing has never been interested in analog computing [simulations] at all,” says Remy Notermans, the company’s director of strategic planning. “The reason is that when we talk about long-term fault-tolerant quantum computing, what we really need is gate-based computers. Obviously, I can’t read the minds of our competitors, but we have seen a window of opportunity where analog computing with neutral atoms is very interesting. However, to achieve the long-term goal of creating a fault-tolerant quantum computer, analog computing is not suitable.”
Cold neutral atoms used as qubits have their pros and cons. The main disadvantage is that operations with them are noticeably slower. However, this is offset by one of the longest coherence times in the industry, during which computational algorithms can be run and errors corrected. Moreover, the company’s focus on atomic spins rather than electron spins maximized coherence times. Today they have the longest running time in the industry, and their 99.6% computational accuracy for two-qubit gates is the highest in the industry.
It is believed that quantum calculations will become absolutely error-free after the creation of a computer with 1000 logical qubits out of 1 million physical ones. In practice, everything may not be so simple, but the goal is clear – to try to create systems with the largest possible number of logical qubits due to the redundancy of physical ones. This year, Atom Computing and Microsoft will offer customers more than 50 logical qubits for more than 1,000 physical ones. At the next stage, the companies plan to begin delivering systems with more than 100 logical qubits for more than 10 thousand physical ones. The company has no bottlenecks preventing the platform from scaling, and is ready for development.
It is interesting to note that a technology giant such as Microsoft is participating in this project as a partner. At the same time, Microsoft has its own quantum platforms and even cloud services in this area. However, in the process of delivering and deploying Atom Computing quantum computers, Microsoft will be fully responsible for its part – for error correction and creating a hybrid stack with classical computers using the Microsoft Azure Quantum package. This package includes tools to virtualize qubits and make it easier to work with them.
Over time, Atom Computing promises to begin supplying quantum computers for the deployment of cloud services. However, for this it is necessary to create next-generation platforms that can bring practical benefits. The company views modern quantum systems, as well as solutions in the coming years, as a testing ground for debugging algorithms and searching for optimal quantum technologies. It will take many more years for a quantum computer to gain practical value.