#quantumprocessingunit

IBM quantum centric supercomputing

IBM and RIKEN researchers used the Fugaku supercomputer and a quantum processor to solve complex chemical equations. This cooperation's closed-loop methodology allowed the two systems to work together and share data continuously. Using a hybrid method, the scientists mapped iron-sulfur compound electrical structures with unprecedented precision. This milestone enables real scientific applications by showing that quantum-centric supercomputing can scale. The work shows a shift toward seamless quantum-classical coordination to enhance computational power and efficiency.

Quantum-centric supercomputer chemistry beyond exact solutions

RIKEN and IBM demonstrated quantum-centric supercomputing (QCSC) at a scale never before feasible, advancing HPC. In a collaborative research initiative, the teams coordinated the Fugaku supercomputer, one of the world's most powerful classical systems, with an IBM Quantum Heron processor on-site. This partnership accomplished the largest and most exact quantum chemistry experiment on a quantum computer, marking a turning point in quantum advantage research.

IBM Director of Research Jay Gambetta presented the experiment at Supercomputing Asia 2026 on January 29. It sought to determine two iron-sulfur molecules' complex electrical structure. Since electron distribution and behavior impact a molecule's interactions and reactions with its surroundings, understanding these structures is a basic chemistry challenge. A very exact answer by the study team revealed that quantum and classical resources can be combined in a smooth, “closed-loop” execution to solve issues that are too tough for precise classical procedures.

An Innovative Method: “Closed-Loop” Workflow

This requires creating and executing a closed-loop workflow. This experiment fed data back and forth in an uninterrupted, iterative cycle, unlike conventional hybrid setups, where a quantum processing unit (QPU) finishes a task before sending results back for classical processing. Practical quantum computing in HPC, where real-world applications require close integration of various compute kinds, is more like this orchestration.

Given its enormity, the orchestration is technically complex. Researchers had to build a complicated job assignment method to use the Fugaku supercomputer and Heron processor during computing. Any idle time wastes valuable runtime for other investigations since classical and quantum resources are expensive and valuable. As a billion-dollar machine, Fugaku must maximize its uptime and avoid “sitting around” during quantum steps. The new approach lowered “time-to-solution” by running both computers as near to simultaneously as possible.

SQD: Hybrid Algorithm Power

The discovery was based on hybrid quantum-classical algorithms sample-based quantum diagonalization (SQD). These algorithms divide a problem into conventional and quantum components. The quantum computer unlatches the hardest aspect of the problem, allowing the classical supercomputer to “turn the handle and open the door,” like the “lifting pin in a lockpicking set,” according to the researchers.

Electronic structure simulations have a vast number of electron combinations that grows exponentially with molecular complexity. This vast region was sampled using the IBM Quantum Heron processor in the SQD workflow to identify traditional computer focus areas. With this understanding, Fugaku reached a conclusion. The systems achieved results similar to state-of-the-art classical approximation techniques and far more accurate than quantum approaches when working together.

Heron-Fugaku Hardware Cooperation

This milestone's hardware is cutting-edge conventional and quantum technology. The 2020–2021 world's fastest computer, Fugaku, provided the experiment's classical foundation. The huge device has 158,976 48-core chips. The system revealed future quantum-centric supercomputing architecture when paired with RIKEN's IBM Quantum Heron CPU.

Director of the Quantum-HPC Hybrid Platform Division at the RIKEN Center for Computational Science, Mitsuhisa Sato, called the achievement “very exciting development for hybrid computing”. The process was designed for Fugaku's architecture, but the researchers found it could be utilized in other cloud-based HPC setups. This shows that standard HPC infrastructures worldwide can interface with quantum computers.

Toward future

Despite this success, the route to full-scale quantum advantage continues. The RIKEN-IBM team expects to integrate GPUs as accelerators into quantum-classical processes next. GPUs using hybrid algorithms like SQD may speed up the process, according to recent studies.

C12 Quantum Computing

In order to improve fault-tolerant quantum computing, Classiq and C12 have integrated the Callisto digital twin into their quantum software platform. Developers can construct, tune, and simulate CNT spin qubit quantum algorithms using this cooperation. This new modality expands Classiq's ecology beyond superconducting, trapped-ion, neutral atom, cat, and photonic backends.

Why Carbon Nanotubes?

Unlike silicon or diamond-based solid-state technologies, C12 uses a distinct hardware approach. They hang ultra-pure carbon nanotubes over gate electrodes to contain electron spin qubits.

Purity and near-one-dimensional structure are the main benefits of this “carbon path”. C12 wants CNTs to:

Minimise Noise: The material's 1D structure reduces charge and magnetic noise, which restricts qubit performance.

Lower noise levels increase coherence times, the “holy grail” for scaling quantum systems.

Improve Connectivity: Circuit quantum electrodynamics (cQED) uses a superconducting microwave resonator as a quantum bus to communicate qubits across vast distances.

Defining the ‘Digital Twin’: Callisto Discovery Edition

The Callisto Discovery Edition, a high-fidelity digital counterpart of C12's quantum processing unit, is key to this interaction. Digital twins are complicated mathematical models that replicate hardware's physical properties, gate fidelities, and noise profiles, unlike basic simulators in quantum computing.

The Callisto emulator on Classiq lets researchers simulate up to 13 noisy qubits using C12's physical features. We need high-fidelity models for:

Realistic Noise Modeling: The twin models carbon nanotube-relevant phonon interaction, charge noise, and relaxation events.

Advanced Operations: It provides mid-circuit measurements and noisy initialization, essential for error-correction testing. Performance Benchmarking: Before sending a pulse to a dilution refrigerator, developers can test an algorithm on real hardware.

Software Synthesis and Hardware-Agnostic Coding Classiq's synthesis engine and Qmod modeling language simplify quantum programming. Users provide high-level functional needs instead of physically assembling logic gates.

The Classiq engine can now “compile” high-level models for the CNT backend with C12 integration. Program automatically:

Explains CNT architecture's unique relationship. Adjusts the circuit for the C12 hardware's unique gate settings.

Optimizes performance and reduces mistakes.

This advance brings the industry closer to hardware-agnostic. Developers may write code once and deploy it across superconducting, photonic, and now carbon nanotube technology.

A Hardware-Software Co-Design Strategy

This relationship validates spin qubits for the quantum community. Classiq and C12 promote “hardware-software co-design” by releasing the Callisto digital twin early.

The digital twin can influence hardware development instead of waiting for optimal, fault-tolerant technology before designing software. A Classiq spokeswoman said the industry “can no longer ignore” C12's carbon nanotube technology.

Commercialization Path

Although an emulator, the Callisto Discovery Edition bridges C12's upcoming hardware processors. Businesses and academic institutions can quickly use Classiq algorithms and error-mitigation strategies to physical CNT QPUs as they grow.

Quantum Error Mitigation Makes Molecular Simulations Scalable and Affordable

Tiled M0

Quantum processing has the potential to revolutionise computing, but noise and errors in hardware must be solved before quantum computers can be built. A group of scientists from Southern Denmark, Copenhagen, the Technical University of Denmark, and Southampton took a major step towards reliable quantum calculations. The crew created “tiled M0,” a unique, cost-effective method for eliminating errors.

This unique strategy dramatically reduces computer resources needed to correct errors, enabling more complex and exact simulations on near-term quantum technologies. Successful molecular energy calculations on benzene and lithium hydride demonstrated the findings.

Variable Quantum Eigensolver with Tiled Noise Mitigation

The Ansatz-based gate and readout error reduction technology M0 has evolved into tiled M0. It is designed for tiled Ansae quantum circuits, which commonly use hardware-efficient circuits, tUPS, and QNP.

Tiled M0's noise characterisation is effective due to its locality approximation to M0 and the insertion of quantum chemical Ansatz elements. Tiled Ansätze' distinctive structure is used in this approximation. Tiled M0 characterises noise in discrete Ansatz tiles rather than the entire system.

The computing demands are greatly reduced by this mechanism. Localisation approximation reduces the cost of the Quantum Processing Unit (QPU) needed for noise characterisation exponentially. The systems under study benefit from the method's consistent characterisation cost regardless of system size or circuit complexity.

Displaying Precision and Efficiency

Researchers verified the tiled M0 technique by calculating molecular ground state energy. They examined tUPS Ansatz systems with two to twelve qubits. Water (H2O), butadiene, benzene, LiH, and H2 were tested.

The strategy worked on IBM's quantum hardware and in noisy quantum processor simulations. Energy comparisons with and without tiled M0 reveal that the approach improves accuracy. Despite the lower computing cost, the results show little to no accuracy loss compared to the M0 technique. Quantum circuits were optimised to find the lowest energy state and operated 100,000 times for statistical significance.

Some lithium hydride simulations attained chemical accuracy, and the method lowered energy error dramatically. Due of its high precision and minimal processing requirements, the approach may be suitable for near-term quantum applications. Its scalability and usability are improved by its layer depth independence in tiled Ansätze.

Limitations and Future Resilience

Even while tiling M0 worked well in hydrogen and butadiene, the researchers found that quantum hardware noise can limit its effectiveness. A thorough analysis found that chemical and noise level affect accuracy.

High noise levels could overwhelm the error mitigation mechanism, and the scientists saw occasions where it failed. Quantum experiments also limited benzene and water improvements. The approach's noise instability on existing hardware is due to quantum hardware noise drifts and fluctuations, according to the authors. The scientists studied quantum computation oscillations to learn how noise impacts accuracy.

Amazon Braket Introduces Quantum Processing Unit Spending Limits

AWS Spending Limit

Amazon Braket's new expenditure restriction function lets customers actively manage Quantum Processing Unit (QPU) prices. When studying quantum computing applications, research institutes, educational institutions, and development teams demanded better QPU expenditure control. This innovation quickly meets their needs.

With spending limitations, customers can set maximum device expenditure boundaries. Amazon Braket automatically checks task submissions against these caps. To avoid unintended overspending, jobs that exceed the budget are rejected before creation. Every AWS region that supports Amazon Braket now offers free expenditure constraints.

How Spending Limits Work A QPU device is linked to a maximum dollar amount to limit spending. When a limit is set, Braket tracks the device's total charges.

Amazon Braket verifies every task creation request against the limit. Braket immediately rejects and reports a validation error if a task's expected cost exceeds the expenditure limit.

The system estimates quantum work costs using the QPU list price and shot count. This predicted sum is removed from the expenditure cap when creating the job. Also, the system allows budget replenishment based on execution efficiency:

Using fewer successful shots than desired returns the unused anticipated cost to the expenditure limit.

Cancelling a task before completion returns the projected cost to the spending limit. Customers can modify spending caps as their needs change.

Governance of scope and cost

Users should know the full scope of this new feature: QPU cost limits apply only to on-demand quantum computations.

The feature excludes Amazon Braket-related costs like:

Simulators.

Managed notebooks.

Cases of hybrid jobs.

Braket Direct bookings produce quantum tasks.

Customers should continue using AWS Budgets as part of AWS Cost Control for comprehensive cost control across Amazon Web Services (AWS), including simulators and traditional compute components.

Additionally, consumers can limit their spending time. For managing AWS credits with expiration dates or ensuring payment cycles, this functionality restricts task submissions to the set interval.

Key Industry and Academic Applications

QPU expenditure caps benefit several key clientele groups:

Research Institutions and Budget Allocation: Research organisations must limit spending to manage quantum computing budgets across programs and clients. Pawsey Supercomputing Research Centre Chief Technology Officer Ugo Varetto underlined the necessity of having capabilities that allow their community of over 4,000 researchers to share budgets fairly and regulate costs. This prevents one workload from using unexpected research community resources. Research teams can set conservative bounds early on and change them for larger trials to minimise costly errors.

Education and Training: Amazon Braket lets educational institutions and training programs set spending caps that match course budgets for quantum computing courses. This prevents pupils from accidentally accumulating large charges and gives them practical experience with quantum devices. If a student mistakenly sets up a task with too many pictures, the spending limit stops the assignment and immediately reports the expenses.

Development Teams and Platform Builders: Quantum algorithm development teams benefit from spending caps during exploratory work. The Amazon Braket API allows companies that develop platforms based on Amazon Braket to programmatically set expenditure constraints for quantum computing and automatically enforce these restrictions on their customers.

Start-up and Resource Management Customers can set expenditure limits using the Amazon Braket Management Console, CLI, or SDK. The AWS CLI and SDK offer powerful automation and custom application integration methods.

Amazon Braket Management Console users can view QPU-specific expenditure limits in real time on a visual dashboard. This page displays the specified limit, current spending, queued spending (estimated device queue job expenses), and remaining spend. Due to this visibility, users may

Amazon Braket Introduces Quantum Processing Unit Spending Limits

AWS Spending Limit

Amazon Braket's new expenditure restriction function lets customers actively manage Quantum Processing Unit (QPU) prices. When studying quantum computing applications, research institutes, educational institutions, and development teams demanded better QPU expenditure control. This innovation quickly meets their needs.

With spending limitations, customers can set maximum device expenditure boundaries. Amazon Braket automatically checks task submissions against these caps. To avoid unintended overspending, jobs that exceed the budget are rejected before creation. Every AWS region that supports Amazon Braket now offers free expenditure constraints.

How Spending Limits Work A QPU device is linked to a maximum dollar amount to limit spending. When a limit is set, Braket tracks the device's total charges.

Amazon Braket verifies every task creation request against the limit. Braket immediately rejects and reports a validation error if a task's expected cost exceeds the expenditure limit.

The system estimates quantum work costs using the QPU list price and shot count. This predicted sum is removed from the expenditure cap when creating the job. Also, the system allows budget replenishment based on execution efficiency:

Using fewer successful shots than desired returns the unused anticipated cost to the expenditure limit.

Cancelling a task before completion returns the projected cost to the spending limit. Customers can modify spending caps as their needs change.

Governance of scope and cost

Users should know the full scope of this new feature: QPU cost limits apply only to on-demand quantum computations.

The feature excludes Amazon Braket-related costs like:

Simulators.

Managed notebooks.

Cases of hybrid jobs.

Braket Direct bookings produce quantum tasks.

Customers should continue using AWS Budgets as part of AWS Cost Control for comprehensive cost control across Amazon Web Services (AWS), including simulators and traditional compute components.

Additionally, consumers can limit their spending time. For managing AWS credits with expiration dates or ensuring payment cycles, this functionality restricts task submissions to the set interval.

Key Industry and Academic Applications

QPU expenditure caps benefit several key clientele groups:

Research Institutions and Budget Allocation: Research organisations must limit spending to manage quantum computing budgets across programs and clients. Pawsey Supercomputing Research Centre Chief Technology Officer Ugo Varetto underlined the necessity of having capabilities that allow their community of over 4,000 researchers to share budgets fairly and regulate costs. This prevents one workload from using unexpected research community resources. Research teams can set conservative bounds early on and change them for larger trials to minimise costly errors.

Education and Training: Amazon Braket lets educational institutions and training programs set spending caps that match course budgets for quantum computing courses. This prevents pupils from accidentally accumulating large charges and gives them practical experience with quantum devices. If a student mistakenly sets up a task with too many pictures, the spending limit stops the assignment and immediately reports the expenses.

Development Teams and Platform Builders: Quantum algorithm development teams benefit from spending caps during exploratory work. The Amazon Braket API allows companies that develop platforms based on Amazon Braket to programmatically set expenditure constraints for quantum computing and automatically enforce these restrictions on their customers.

Start-up and Resource Management Customers can set expenditure limits using the Amazon Braket Management Console, CLI, or SDK. The AWS CLI and SDK offer powerful automation and custom application integration methods.

Amazon Braket Management Console users can view QPU-specific expenditure limits in real time on a visual dashboard. This page displays the specified limit, current spending, queued spending (estimated device queue job expenses), and remaining spend. Due to this visibility, users may make educated decisions about workload management and restriction management.

Please note that recognized researchers can apply for AWS Cloud Credits for Research to support their Amazon Braket experiments. The Amazon Braket Spending limitations page is where customers should start using this service.

make educated decisions about workload management and restriction management.

The University of Naples Federico II has Italy's largest quantum computer, a milestone for European quantum research. The system's successful deployment, powered by a 64-qubit QuantWare QPU, confirms the Quantum Open Architecture (QOA) paradigm, a modular method that democratises top-tier quantum computing. The 1224-founded organisation hosted this milestone, which shows a major quantum sector change. Instead of expensive, proprietary “full-stack” systems from a single vendor, universities and research labs can build powerful quantum computers using specialised components from numerous manufacturers.

New Quantum System Building Model

The Quantum Open Architecture paradigm is redefining a field dominated by government labs and tech companies. QOA allows universities to acquire QPUs off the shelf, saving time and money on advanced quantum capabilities.

This method has advantages, according to Professor Francesco Tafuri, who runs the Quantum Computing Napoli (QCN) Laboratory, where the computer is. He said “a powerful, commercially available, and ready for integration processor was necessary to build Italy’s largest quantum computer.” “QuantWare's Tenor QPU accelerated its timeline, allowing us to focus on system and application development.” This modular strategy avoids the costly options of buying a closed “black box” system from a big supplier or building a custom processor from scratch. Putting their machine together from its parts helped the University of Naples team understand it better and optimise and explore. Technology Behind Milestone The Naples system uses QuantWare's 64-qubit Tenor QPU, which uses superconducting transmon qubits like Google and IBM. These processors operate at temperatures below interstellar space, thus specialised dilution freezers are needed to chill the circuits to millikelvin, where quantum properties like superposition and entanglement arise. Open architecture addresses the massive engineering challenges of constructing a quantum technology stack by enabling specialisation. Due to this division of labour, QuantWare can refine processors while others work on software, control electronics, or cryogenic engineering. This is similar to how Intel focusses on processors while others build equipment and software in the traditional computing market. A spin-off from TU Delft's QuTech centre, QuantWare has become the world's largest QPU provider with clients in over 20 countries. The company's unique VIO 3D scaling architecture uses three-dimensional chip stacking to overcome planar chip connection restrictions and enable future “MegaQubit-scale” devices for quantum advantage.

Quantum democratisation and ecosystem development

Clifford Circuit Initialization

Clifford Circuit Initialisation Revolutionises Quantum Computing: More Effective Algorithms

Fraunhofer ITWM researchers Théo Lisart-Liebermann and Arcesio Castanadena Medina developed and demonstrated Clifford Circuit Initialisation, a crucial step towards quantum computing. The Quantum Approximate Optimisation Algorithm (QAOA) and Variationally Quantum Eigensolver (VQE) employ complex quantum circuits, although this could improve their optimisation. In “Clifford Accelerated Adaptive QAOA,” they offer a unique technique that improves quantum-classical interactions and reduces QPU calls by combining classical simulation capabilities.

Clifford Circuit Initialisation improves parametric quantum circuit (PQC) parameter guesses. It exploits the intrinsic efficiency of Clifford Group gate circuits, which the Gottesmann-Knill theorem allows classical hardware to simulate quickly. The researchers found a way to increase circuit parameter initialisation and optimisation efficiency by exploring the parameter space with a smaller set of “Clifford-expressible points” (Clifford Points).

This invention is easily integrated into dynamic circuit reconfiguration methods like ADAPT-QAOA, which optimise QAOA performance by iteratively changing gate designs. The researchers' Clifford approximations improved ADAPT-QAOA at many levels.

Three Improvement Pillars

The study finds Clifford approximations advantageous in three areas:

Clifford Point pre-optimization gives ADAPT non-trivial gate selection behaviour, which may accelerate convergence. Early results suggest that this can accelerate initial convergence for challenges like the Transverse Field Ising Model (TFIM). As the TFIM Hamiltonian's gz control parameter grows, single-qubit Z-gates' benefits become more apparent. The Clifford Point projection on the Z-basis decreases continuous optimisation errors in some cases. The MaxCut problem is more complicated. Some pre-optimization methods failed, perhaps leading ADAPT to enter local minima. This suggests that MaxCut may need momentum transfer or objective function data for Clifford Point optimisation.

Fully Classical and Parallel Operator Selection: ADAPT operator selection that is fully parallel and classical is a major development. Clifford circuit assessments can be mimicked on classical hardware, so operator selection does not require costly QPU calls. Due to Clifford Point selection, better choices were made when extending the QAOA mixer layer for MaxCut, resulting in convergence behaviour at lower parameter counts. It recommended using two-qubit RZZ gates instead of single-qubit RY rotations, which improve expressivity and circuit performance. Similar benefits were reported for TFIM. This allows for substantial quantum-classical integration, which reduces computing time and cost by offloading workloads without quantum speedup to classical hardware.

Optimisation using T-gate Error Approximation: One of the most significant breakthroughs is the treatment of T-gates, non-Clifford gates essential to universal quantum processing. The researchers showed that low-rank stabiliser decomposition to apply a 10–30% error approximation on T-gates improves MaxCut and TFIM convergence quality. This unexpected finding suggests that T-gates are overused in modern quantum circuit design. This realisation could enable aggressive circuit compilation optimisations, reducing quantum resource requirements for complex algorithm implementation. The T-gate approximation improved convergence even when MaxCut's Clifford Point pre-optimization yielded contradictory results.

Towards Scalable Quantum Algorithms

This study advances the development of scalable and effective quantum algorithms. By actively integrating classical approximations, the team has improved hybrid quantum-classical computation prospects by better handling the trainability-expressivity trade-off in PQC design.

The researchers acknowledge that MaxCut and TFIM benefits depend on parameters and issue topologies and are problem-dependent. Future study will create automated methods to detect and reduce T-gate over-representation in quantum circuits and test them with different issue formats and larger system sizes. To explain the observed traits, more theoretical research is needed, notably on Clifford Point operator selection.

Quantum annealers are generally employed to determine classical spin systems' low-energy configurations. They address optimisation issues using quantum mechanical notions, unlike universal digital quantum computers.

Their typical operation and distinguishing features:

Adiabatic Evolution: Quantum annealers gradually transform a quantum system from a known initial state, usually a superposition of all possible states, to a final state with the answer. This is adiabatic evolution. A problem-specific Ising Hamiltonian controls its evolution. The term “adiabatic” refers to a slow, steady change in conditions that aims to minimise the goal problem's energy while keeping the system in its ground state.

The D-Wave system implements a transverse-field Ising model with a time-dependent Hamiltonian. A longitudinal energy scale that depicts the classical problem and a transverse field controlled by a function A(s) that encourages quantum fluctuations make up this Hamiltonian. These energy scales evolve according to the annealing schedule parameter, which runs from 0 to 1.

In contrast to ideal zero-temperature quantum computers, quantum annealers use Boltzmann sampling at finite temperatures. The spin configurations at the end of the anneal should follow a Boltzmann distribution if the system stays in contact with its thermal environment (such as the refrigerator or QPU) long enough. This makes them promising for describing complex classical and quantum thermodynamics.

Quantum annealers have received interest for their ability to provide finite-temperature criticality that classical technologies cannot. Their ability to capture classical and quantum phase transitions makes them useful for equilibrium statistical mechanics and finite-temperature criticality.

Despite their potential, quantum annealers have struggled to determine finite-temperature criticality for several reasons:

Noise and Thermal Variations: QPU temperature changes may affect Boltzmann sampling accuracy.

Hardware Limitations: Qubit connectivity, coupling noise, and annealing biases might impair measurement reliability and repeatability.

Critical Slowing Down and Freeze-out: Critical areas incur excessive annealing durations due to system reaction time divergence. A “freeze-out” event can also occur, slowing dynamics and locking the system in metastable configurations.

In Two-Dimensional Ising Model, Quantum Annealers Resolve Finite-Temperature Criticality

A team led by Francesco Campaioli from RMIT University and the Universitá degli Studi di Padova and Gianluca Teza from the Max Planck Institute for the Physics of Complex Systems has improved our understanding of how physical systems change states at realistic temperatures.

In Finite-temperature criticality through quantum annealing, they show how to map and analyse the behaviour of a magnetic material going through a phase transition using a quantum annealer, capturing the entire range of critical behaviour at finite temperatures. This research provides a better knowledge of equilibrium and non-equilibrium processes and opens up new avenues for studying complex systems like magnetism and maybe biological systems.

Critical phenomenon study challenges:

Traditional methods struggle with entanglement, which hinders tensor network approaches in higher dimensions, and critical slowing down, which reduces sampling efficiency.

Even though quantum computing and simulation platforms like digital quantum processors and analogue platforms are promising, the system's response time, QPU temperature fluctuations, and hardware limitations make it difficult to precisely identify finite-temperature criticality with quantum annealers.

Overcoming these obstacles:

The researchers demonstrated that quantum annealers may precisely resolve finite-temperature criticality in complex many-body systems with careful programming.

As a benchmark, they chose the 2D ferromagnetic Ising model, which has a critical point. A novel embedding method reduced edge effects, hardware asymmetries, and magnetic leakage by mapping the simulated system to the annealer's physical design. This requires building physically coupled qubits into logical “superspins” and adding periodic boundary constraints to eliminate edge effects to emulate lattice connection and increase coherence. A calibration approach using local field offsets corrected magnetic leakage biases in superspins with high-amplitude ferromagnetic couplings.

By calibrating the QPU temperature and changing the model's energy scale, they inserted lattices with over 2500 spins. Most of the qubit count was covered by the largest D-Wave Advantage systems. Combined with randomised gauge transformations, this embedding strategy balanced hardware spatial asymmetries.

They ended the anneal at s*=0.5 to keep the system in a dynamic region suitable for thermal equilibrium sampling and avoid freeze-out, which locks the system in metastable configurations.

They also performed in-situ temperature assessments before each batch of samples, a real-time temperature compensation approach not used in previous quantum annealing Boltzmann sampling studies. This allowed consistent finite-temperature sampling across runs and system sizes.

Key discoveries and demonstrations:

The researchers made two vital measurements to demonstrate that they could measure and adjust key solvable model parameters:

After a sudden change, they observed magnetisation throughout a non-equilibrium relaxation and found that a transverse field accelerated convergence to the symmetry-broken ground state. The system often stayed in metastable states and relaxed slowly at small transverse fields.

After starting the system in a metastable state, a fully polarised state opposite the longitudinal field, they assessed the possibility of tunnelling over an energy barrier. The mechanism remained imprisoned at small transverse fields, but the success chance increased drastically as the field extended, suggesting barrier crossing. This underlines how the transverse field helps people escape metastability. At experimental temperature, the mechanism tunnelled through a 60-fold larger barrier than thermal energy. The transverse field levels studied were well below the critical field to ensure that the observed transition was a significant crossing between locally stable configurations rather than a smooth crossover between disordered states.

wider implications: This work establishes a scalable and systematic strategy to finite-temperature criticality research with quantum annealers. Using the great connectivity and coupling tunability of existing annealing architectures, this method can be applied to many classical and quantum systems that can be mapped to the Ising model. These include:

Frustrated magnets

Spinning glasses

Lattice gauge theories

Quantum gravity models

Non-equilibrium quantum thermodynamics.

Abnormal relaxation.

Quantum circuit compilation

The programmable framework can also explore quenched disorder and non-equilibrium dynamics, allowing researchers to study important phenomena that are difficult to study using standard methods. The work finds that accurately calibrated quantum annealers can discover finite-temperature criticality in complex many-body systems, circumventing hardware constraints like the freeze-out point and QPU native topology.

EIF European Investment Fund

Quantonation II, France-based Quantonation's second venture capital fund, received €30 million from the European Investment Fund (EIF) to strengthen Europe's leadership in cutting-edge science. This capital infusion marks a turning point in InvestEU's attempts to expand deep-physics and quantum technology enterprises in Europe.

Quantonation II: Science-Market Bridge

Quantonation II, with a €200 million target fund, highlights early-stage companies developing game-changing technologies like secure communications, high-precision sensing, and quantum computing. The fund supports 25 enterprises and five venture studios worldwide to commercialise key particle physics and  quantum mechanics findings.

Dual financial backing and in-depth specialised knowledge underpin the fund's strategy. Quantonation's Christophe Jurczak, Charles Beigbeder, Olivier Tonneau, and Will Zeng lead portfolio firms with capital and subject expertise.

EIF Strategic Vision: Deep-Tech Financing Gap Closure

The EIF's investment tackles Europe's perennial deep-tech funding need. Quantum and related fields require substantial R&D and scientific skills, yet tech startup techniques often take precedence.

EIF CEO Marjut Falkstedt said Quantonation II shows how EIF supports Europe's deep-tech goals. The €30 million investment addresses a major financing need and supports EIF's InvestEU goal of closing the equity gap in priority innovation industries.

Rising Quantum Role for France

Quantum Insider's national roundup found June 2025 to be a milestone month for French quantum activity, which matches with Quantonation II's spending. France has promoted quantum infrastructure, worldwide relationships, and technology in addition to cash.

Commercial Innovations & Ecosystem Growth

The neutral-atom quantum hardware business Pasqal launched its first North American quantum fabrication unit in Sherbrooke, Quebec. It also sold Distriq of Canada a 100-qubit QPU.

The company also bought Canadian photonic integrated circuit leader AEPONYX to achieve fault-tolerant quantum systems.

Orange Business and Toshiba launched quantum-safe network services in Paris using post-quantum cryptography and QKD.

French IT giants Alice & Bob won support after being named to the national IT Next40/120 list.

Research and Global Partnerships

The EuroQCS-France consortium began remote access to a 12-qubit Quandela photonic quantum computer for European academics before system deployments.

Leaders at the France-Singapore Frontier Technologies forum highlighted greater collaboration in AI, quantum, and renewable energy, enhancing France's worldwide influence.

This burgeoning government-business ecosystem illustrates France and Europe's interest in quantum innovation. This makes Quantonation II support appropriate and possibly revolutionary.

Scaling Deep-Tech Commercialisation: Implications for Europe's Quantum Future

Quantonation II helps science move from lab to market. Startups receive finance and specialised support to scale hardware production, develop quantum algorithms, and safeguard intellectual property. These talents are crucial for building future computing, sensing, materials, and cybersecurity enterprises.

Technology sovereignty reinforcement

The investment supports Europe's strategic goal of technical autonomy for key infrastructures. Quantum technology can enable secure communications and material science advances too critical for other superpowers. EIF support shows a policy to develop domestic champions.

Global Talent and Capital Attraction

A strong deep-tech venture capitalist presence lends credibility and opportunities. EIF's €30 million anchor and Quantonation II's €200 million target should attract top tech and international LPs interested in mission-driven science.

Promoting Startup-Institution Collaboration

Quantonation II integrates startups with venture studios to stimulate collaboration between academics, industry, government labs, and larger businesses. This ecosystem approach accelerates R&D and creates profitable applications.

The Last Shot

The EIF's €30 million Quantonation II investment is a key step in Europe's quantum strategy. The fund combines financial power with domain expertise to turn quantum research into strategic and commercially viable solutions. This measure reinforces sovereignty in a technological field with long-term effects and shows Europe's desire to become a deep tech hub.

In conclusion