From the 2000s onwards, new generations of high-performance computers (AI-centric) have pushed back the boundaries of machine learning, enabling applications to work in hundreds of different languages, using text, images and video. This computing power is now available to the widest possible audience, thanks in particular to advances in hybrid cloud computing.

Today, the development of quantum-centric supercomputers is paving the way for the potential resolution of problems hitherto unsolvable with classical computers, such as the simulation of natural phenomena (materials, chemistry, physics) or the analysis of data with complex hidden structures, or which are difficult to represent in a classical way (such as spectral properties, graph and group properties, or feature selection).

High-performance quantum computing will also make it possible to significantly increase the speed of resolution of certain complex calculations, such as search and optimization algorithms. Although we're not talking about exponential acceleration (as in the case of simulation or complex data sets), quadratically accelerated quantum applications could nonetheless help reduce the computing resources required, and thus bring significant value to businesses and end-users.

To successfully create this quantum supercomputer, IBM has been relying since 2020 on a reliable and proven development roadmap[https://research.ibm.com/blog/ibm-quantum-roadmap-2025], updated in 2022, including the foundations of the necessary infrastructure. The three cornerstones of this new architecture should be available and announced by the end of 2023:

• The "IBM Heron" processor which, with its 133 qubits, is a complete overhaul of the previous generations ("IBM Eagle" with 127 qubits and "IBM Osprey" with 433 qubits which are already available and accessible), to enable a very significant improvement in the quality of operations performed, but also compatibility with a modular architecture.
• The IBM Quantum System Two, the new winning system in the Innovation by Design 2023 category[https://www.fastcompany.com/90924596/ibm-transformed-quantum-computing-into-modular-computer] by Fast Company. This modular system will enable up to three IBM quantum processors to be housed in the same vacuum chamber. This will open the door to huge scalability through quantum and classical communication processes.
• A middleware layer dedicated to quantum computing, including tools for decomposing an application, running it in parallel on several quantum processors and then reconstructing the results. Here again, the aim is to make the solution efficient on a large scale. This layer is already in beta access through the IBM Quantum Platform[https://quantum-computing.ibm.com].

These three cornerstones of the quantum-centric architecture are not, however, IBM's only points of progress in recent months. IBM wants to bring useful quantum computing to its private and public partners as soon as possible.

The announcement in June 2023 of IBM's first quantum datacenter in Europe (scheduled to open in 2024) to facilitate access for companies, research institutes and government agencies to explore how best to apply quantum to their industry, is a first step forward.

As a second step forward, the article published in June 2023 by UC Berkeley and IBM researchers in Nature[https://research.ibm.com/blog/utility-toward-useful-quantum] demonstrates a real and pragmatic path towards the usefulness of quantum computing in the medium term. The IBM Eagle processor used in this demonstration will thus be generalized in all our systems currently deployed, to enable our partners to identify right now use cases that would be too complex for a classical computer.

Finally, IBM is pursuing a steady path towards the quantum advantage and the universal quantum computer, based on today's technologies. The announcement of theoretical work on a new type of error-correction code[https://research.ibm.com/blog/error-correction-codes] in August 2023 provides a glimpse into the future. This new published code is part of the Low Density Parity Check-Code family and, compared with the surface code, offers a more than 10-fold reduction in the number of physical qubits required, but is otherwise comparable in terms of result, albeit requiring a specific hardware architecture. With these assumptions, the level of error correction that would require 1,000 to 2,000 qubits for the surface code can be achieved with just 144 qubits in this new code.

In anticipation of future announcements and realizations at the end of 2023, it's time for you too to come and carry out your first experiment with a quantum computer. Whether via Qiskit, the open-source Software Development Kit giving access to IBM technologies (but also those of other players), or via free access to simulators and real quantum machines on the IBM Quantum Platform.