The Next Phase for Quantum Innovation: “IBM’s Starling will solve the hardest problem in quantum computing” – Vasques

As Quantum computing evolves, IBM is advancing Quantum innovations to develop new supercomputer systems that solve complex issues across major industries.

Xavier Vasques, Vice President, Chief Technology Officer, and Director of R&D at IBM France, shares insights in an interview with Ndubuisi Micheal Obineme, Managing Editor, The Energy Republic, on the company’s trajectory for Quantum innovations and action plans to launch new Quantum computers in 2026 and beyond.

TER: Please tell us about IBM’s Quantum computing roadmap.

Vasques: We have been building quantum computers for over a decade. We put our first quantum computer on the cloud in 2016 with five qubits; but today, our Heron processors carry 156 qubits, and we operate the largest fleet of quantum systems in the world (more than 90).

Since 2020, we have published our roadmap publicly and delivered against its mandate year after year. The roadmap has now reached an inflection point: we have moved from chasing qubit counts to engineering fault tolerance.

Our public plan is to deliver IBM Quantum Starling in 2029, the world’s first large-scale, fault-tolerant quantum computer, with 200 logical, error-corrected qubits able to run on the order of 100 million quantum operations.

We get there through a sequence of modular processors, each providing one essential ingredient: Loon for the high-connectivity components and couplers, Kookaburra as the first fault-tolerant module that combines logic and memory, and Cockatoo to entangle modules together.

Beyond Starling, we expect a quantum-centric supercomputer by 2033, IBM Quantum Blue Jay, capable of running on the order of a billion gates across roughly 2,000 logical qubits.

In the near term, we are confident that partners running on IBM systems will demonstrate quantum advantage in 2026, the point at which a quantum computer will be able to solve a problem more efficiently than any known classical-only method to date.

TER: What is IBM doing in France, and are there any specific projects that you would like to share with us?

Vasques: France has built one of the most dynamic quantum ecosystems in Europe through its national quantum strategy, and IBM is part of that fabric on several levels. Our open-source software stack, Qiskit, is the most widely used quantum development framework in the country, across universities, national research organisations and industry.

French researchers and enterprises access our global fleet through the IBM Quantum Network, and, importantly for European clients, they can run workloads within the European Union: our first European quantum data centre is operated in Ehningen, Germany, with all job data processed inside EU borders.

In France, we also have a team in Montpellier dedicated to Qiskit, supporting our users. We work hand in hand with the French ecosystem. A good example is our partnership with Pasqal, a French company building a common, hardware-agnostic approach to quantum-centric supercomputing, where machines and classical high-performance computing can work together in a single workflow.

TER: How does IBM’s approach to Quantum computing differ from other technologies, particularly in areas such as hardware, software, and ecosystem development?

Vasques: On hardware, we are betting on superconducting transmon qubits combined with a modular architecture: rather than building one impossibly large chip, we connect smaller modules with couplers and with a new family of quantum error-correcting codes, qLDPC, that cut the overhead by roughly 90 percent, paired with real-time decoding. This is what turns fault tolerance into an engineering problem rather than a theoretical one.

On software, Qiskit is the most adopted quantum development stack in the world (it is now used by nearly 70% of quantum developers and has enabled the execution of more than 4 trillion quantum circuits on quantum computers), and we are layering AI assistance on top of it to automate parts of quantum code generation. It is open and hardware-agnostic, and it can also drive other machines, not only ours.

On ecosystem, the scale speaks for itself: we operate more than 90 quantum systems globally, the largest fleet in the industry, and our network spans more than 300 organisations running real workloads today. We are also extending beyond single machines, developing the interconnect and networking layer needed to link quantum processors into distributed, quantum-centric systems.

The differentiator is the full stack, from the chip to the software, delivered openly and on schedule.

TER: IBM is also investing $10 billion over the next five years to advance Quantum computing. What key technological and business outcomes does IBM expect this investment to achieve, and how will it impact the Quantum computing landscape?

Vasques: On June 2, 2026, we announced that we will invest more than $10 billion in quantum computing over the next five years, spanning research and development, capital expenditure, manufacturing scaling, ecosystem partnerships and acquisitions. The intent is to accelerate the roadmap beyond Starling, to industrialise quantum, moving it from leading-edge systems toward fault-tolerant scale.

A concrete part of that is manufacturing. Through a letter of intent with the U.S. Department of Commerce, IBM is establishing Anderon, a dedicated quantum chip foundry, with $1 billion from the Department matched by $1 billion from IBM. The outcomes we expect are threefold: demonstrate quantum advantage on useful problems in 2026, secure the manufacturing and supply chain required to scale, and deliver fault tolerance in 2029. For the wider landscape, the signal is clear: the quantum era is no longer a research promise; it now has a time-bound industrial trajectory.

TER: As IBM plans to launch Quantum Starling by 2029, what challenges will it resolve, and which applications, including industries, could benefit from these advancements?

Vasques: The single hardest problem in quantum computing is error: qubits are fragile, and genuinely useful algorithms require long, deep circuits. Starling resolves this by running error-corrected logical qubits at scale, which allows more than 100 million operations across 200 logical qubits. It unlocks problems classical computers cannot reach.

For an energy and industrial audience, the most relevant applications are in chemistry and materials science, designing better catalysts, battery materials without scarce heavy metals, and carbon-capture chemistry, and in large-scale optimisation such as grid balancing, logistics and risk management. These are exactly the problems where the number of possibilities explodes beyond what classical machines can handle.
We already see early signals across industries on today’s systems.

With the Cleveland Clinic and RIKEN, we modelled a 12,635-atom protein, the largest ever simulated on a quantum system; Moderna has explored mRNA structure on 80 qubits; HSBC reported around a 34 percent improvement in bond-trading prediction; and Q-CTRL achieved large speedups in materials simulation.