EY, one of the Big Four global consulting companies, has installed a quantum computer on-site, led by EY Canada. The machine is physically located in Canada, but the initiative is global. EY describes the installation as an expansion of its worldwide quantum computing capabilities, with applications developed and tested in Canada intended to scale through the EY network.
EY is not a university, a national laboratory or a quantum hardware manufacturer. It is one of the world’s largest professional-services organizations. Its business is helping companies and governments decide how technology should be governed, procured and deployed. Now it has brought the technology itself in-house.
Susan Etlinger, EY’s Director of AI Thought Leadership, summarized the rationale in three phrases: “Think data sovereignty, client zero experience, and quantum readiness.”
Those are all legitimate reasons for the investment. But the most important may be the one that is hardest to quantify. EY is buying its way into the learning curve.
What is quantum computing?
It’s faster, but not just faster. Think of it as almost a hybrid between deterministic (conventional computing) and artificial intelligence computing. Conventional computing stores and processes information in bits, each represented as a zero or a one. A quantum computer uses quantum bits, or qubits, which can be prepared in combinations of states and manipulated through quantum effects including superposition, entanglement and interference.
This is different from brute force variable computation. A quantum computer tries every possible answer at once and then selects the right one, a common but misleading explanation. Quantum algorithms are designed so that probability amplitudes associated with useful answers are strengthened while others cancel through interference. When the qubits are measured, the machine returns classical results drawn from the resulting probability distribution. In this specific ways it has similarities to generative AI.
That approach can produce major advantages for certain mathematical structures, but it isn’t useful for all applications and it doesn’t necessarily make them faster. It does allow more functions to occur simultaneously.
Email, accounting systems, databases, web applications and most analytics will continue to run on classical computers. Quantum machines are being developed for narrower classes of problems where classical computation becomes prohibitively difficult, including the simulation of quantum systems, some forms of optimization, cryptanalysis and specialized mathematical operations.
The most promising long-term commercial applications include:
● simulating molecules, chemical reactions and materials
● designing drugs, catalysts, batteries and industrial materials
● solving selected scheduling, routing and resource-allocation problems
● improving parts of financial modelling and risk analysis
● performing certain searches, sampling tasks and linear-algebra operations
● breaking some current public-key cryptography once sufficiently powerful fault-tolerant machines exist
Potential is a long way off from present capability. Current systems remain noisy, difficult to scale and highly specialized. For most business problems, the best classical systems still win on cost, reliability and performance.
What EY has actually installed
EY has called the new system an “on-site quantum computer.” It has not publicly identified the manufacturer, qubit architecture, qubit count, error rates, benchmark performance, acquisition cost or contractual structure.
It has also not said whether EY owns the machine outright, leases dedicated capacity, or hosts a vendor-managed system at an EY-controlled location. On-site, in-house, dedicated and owned are not interchangeable terms. Different quantum architectures also require very different physical environments.
Superconducting systems use electrical circuits cooled inside dilution refrigerators to temperatures close to absolute zero. Trapped-ion systems hold charged atoms in electromagnetic fields and manipulate them with lasers. Neutral-atom systems arrange atoms with optical tweezers. Photonic systems encode and process information using light.
An on-site installation can range from a relatively compact educational system to a laboratory-scale cryogenic machine with extensive control equipment.A complete quantum environment generally includes:
| Layer | What it does | Classical comparison |
| Quantum processing unit | Performs quantum operations | Specialized accelerator |
| Qubit environment | Keeps the physical qubits stable | Cooling and facility infrastructure, but far more specialized |
| Control and readout systems | Sends instructions and measures results | Accelerator control electronics |
| Classical CPUs and GPUs | Prepares problems and processes outputs | Host computing |
| Quantum runtime and software | Compiles circuits, schedules jobs and manages execution | Runtime and orchestration stack |
| Network and identity controls | Determines who can use the machine and what it can connect to | Private infrastructure and access management |
| Calibration and operations | Maintains performance within narrow physical tolerances | A combination of site reliability engineering and experimental physics |
The quantum processor is only one part of the system.
Quantum computers still need ordinary computers
A quantum computer doesn’t replace CPUs or GPUs. It operates more like a specialized accelerator inside a larger classical workflow.
The classical system prepares the problem and translates the relevant portion into a quantum circuit. The quantum processor executes the circuit, usually many times, and returns measurements. Classical software interprets the results and may decide what the quantum system should run next.
Many current experimental algorithms are hybrid. Variational algorithms such as the Variational Quantum Eigensolver and the Quantum Approximate Optimization Algorithm use a classical optimizer and a quantum circuit in an iterative loop.
A TechTimes article about EY’s innovative approach argues that putting the QPU on-site solves a fundamental latency problem in these loops, which may or may not be the case. Latency can matter, particularly when algorithms require many interactions between classical and quantum resources. But a cloud provider can place classical compute beside its own QPU and expose the combined service remotely. Queue times, circuit execution, measurement volume, calibration, compilation and algorithm design can matter more than the internet round trip. Current variational algorithms also face convergence and noise problems that physical proximity alone does not solve.
The stronger engineering argument and business rationale for EY’s deployment is control over the complete development environment, and hands-on experience. Dedicated local access can make it easier to run repeated experiments, characterize hardware behaviour, integrate the system with approved classical resources, test security controls and train staff without competing for time on a public fleet.
Why EY wants this now
Cloud access remains the logical starting point for most organizations. It avoids the expense and complexity of maintaining specialized hardware, provides access to multiple qubit architectures and allows users to move to newer processors as the technology changes.
Dedicated access
EY already has cloud access. The firm joined the IBM Quantum Network in 2023. The new installation adds capabilities that ordinary shared access does not provide.
Quantum hardware is scarce and changes rapidly. Dedicated access allows teams to use the same physical system repeatedly, study its errors and calibration patterns, and build an internal development rhythm around it. The immediate value is not necessarily faster calculation. It is faster organizational learning.
Proprietary application development
The circuit, optimization constraints, model structure, outputs and surrounding classical workflow can reveal commercially sensitive information even when raw customer records never reach the QPU.
An internally controlled environment gives EY more options for isolating that work and protecting the intellectual property created around it.
Talent
Quantum expertise cannot be acquired instantly when demand appears. Physicists, mathematicians, software developers, cybersecurity specialists and industry experts have to learn to work together.
Operating hardware gives EY a recruitment and training asset. It also gives technical staff a reason to remain with a professional-services firm rather than moving to a hardware company, research institute or hyperscaler. It is not the first consulting firm to do so, but the first based in North America.
| Firm | What it has done | Physical quantum computer under its control? |
| Eviden/Atos | Installed a five-qubit IQM Spark at its Angers factory in 2024 for employee and client experimentation | Yes |
| EY | Installed an undisclosed on-site quantum computer led by EY Canada in 2026 | Yes, although ownership and vendor-management terms are undisclosed |
| TCS | Partnering with IBM and the Andhra Pradesh government on an IBM Quantum System Two at India’s Quantum Valley Tech Park | Real hardware involvement, but not a private TCS installation |
| Capgemini | Operates Q-Lab facilities and serves as an IBM Quantum Hub, giving clients access to IBM systems | No publicly disclosed local QPU |
| BCG | BCG X joined the QuEra Quantum Alliance to develop enterprise and government applications | No publicly disclosed local QPU |
| Booz Allen Hamilton | Runs a quantum practice, develops quantum software and has invested in hardware company SEEQC | No publicly disclosed internal QPU |
| CGI | Builds proofs of concept using D-Wave, Quantinuum, Azure Quantum and AWS Braket | No publicly disclosed internal QPU |
| McKinsey | Conducts market research, develops client strategies and advises quantum companies and governments | No publicly disclosed internal QPU |
| Bain | Advises clients on quantum readiness, pilots and investment timing | No publicly disclosed internal QPU |
| Accenture | Offers quantum strategy, security and experimentation services through technology partnerships | No publicly disclosed internal QPU |
Commercial credibility
EY wants to advise banks, governments, energy companies, manufacturers and pharmaceutical companies about quantum adoption. Direct operating experience makes that advice more credible.
It also creates a client demonstration environment. EY can show executives what an actual hybrid workflow requires, what fails, what the security boundaries look like and how little of the process resembles the magical language often used in quantum marketing.
Option value
The machine does not have to deliver a near-term return comparable with a production server fleet to be strategically useful.
If commercially useful quantum computing arrives gradually, EY develops expertise along the way. If progress accelerates, it has trained people, governance processes, supplier relationships and candidate applications. If the field advances more slowly, EY still acquires practical evidence about what does not work and avoids advising clients entirely from vendor demonstrations.
For a global consultancy, that knowledge can itself be sold. Data residency is real. Data sovereignty requires more. EY says in-house operation can help organizations address regulatory, privacy, security and industry requirements that cloud alternatives cannot always meet.
That is carefully worded and defensible. It should not be expanded into the claim that a Toronto installation automatically resolves sovereignty for regulated clients. Physical location answers only one question: where does the computation occur?
Data sovereignty requires several more:
● Who owns or controls the hardware?
● Who maintains it and can access it remotely?
● What telemetry, logs, circuit data and diagnostic information leave the site?
● Where are the classical preprocessing and storage systems?
● Which legal entity controls the environment?
● Which country’s laws can compel the vendor or operator?
● How are encryption keys, backups and administrative identities governed?
● Does the client’s actual regulatory regime permit the proposed arrangement?
The deployment is a global capability with a Canadian physical jurisdiction versus a universal sovereign-computing solution.
EY ran this playbook with AI and has learned lessons
EY’s move fits a broader pattern. The firm tends to place emerging technology inside its own operations early, then turn what it learns into client services.
In September 2023, EY launched EY.ai after investing US$1.4 billion in the platform and related capabilities. The system had been in development for 18 months, and EY piloted its generative AI tools with 4,200 technology employees before releasing EY.ai EYQ, its private generative AI environment.
EY then deployed EYQ to approximately 300,000 people. The firm now reports adoption above 81 percent and more than 116 million prompts processed. It has since expanded its AI infrastructure into an enterprise agentic platform designed to support a global workforce of roughly 400,000 people.
The value of Client Zero (deploying technology internally first) is the lessons learned during the deployment. It forced the firm to confront permissions, proprietary information, security, training, workflow redesign, model integration, employee behaviour and governance at enterprise scale.
It also exposed the limits of internal control. In May 2026, EY Canada removed a report on cybersecurity and loyalty programs after researchers found fabricated and inaccurate citations, invented data and references to material that did not exist. The Financial Times independently verified several of the problems. EY said it was reviewing how the report had been published and that it was not connected to client work.
It was, however, an EY governance failure in an organization that had already deployed controlled AI infrastructure at enormous scale. The lesson is important: private infrastructure does not verify output. Access controls do not create factual accuracy. A governed platform and a governed work product are separate things.
EY was not unique. Deloitte, KPMG and PwC have also faced cases involving inaccurate or apparently AI-generated professional material. The entire Big Four has discovered that adopting a technology early also means encountering its failure modes in public.
Client Zero includes learning from the failures. With AI, EY learned that infrastructure, models, workflows and publication controls have to be governed separately. With quantum, it can begin learning the equivalent lessons before clients attempt broad deployments.
EY is early, but the other Big Four firms are not absent
PwC, Deloitte and KPMG all developed quantum practices before EY’s installation. Deloitte advises on quantum strategy, applications and post-quantum cybersecurity. PwC works on use cases, risk and quantum-safe migration. KPMG has published quantum-readiness frameworks covering cybersecurity, skills, investment and infrastructure. All four firms have alliances, researchers and client projects.That makes EY the earliest physical adopter among the Big Four.
Early adoption provides:
● operational knowledge that competitors cannot obtain entirely from presentations
● a head start in recruiting and training scarce talent
● a place to develop proprietary applications and controls
● greater credibility with clients considering dedicated infrastructure
● influence over vendors, standards and emerging implementation practices
It also imposes costs:
● the hardware may become obsolete quickly
● EY may commit attention to an architecture that does not prevail
● useful client workloads may develop more slowly than expected
● operating and securing the environment may cost more than cloud access
● marketing may run ahead of demonstrated performance
The other Big Four firms may be conservative because cloud access preserves flexibility. Waiting allows them to use better hardware later, avoid maintenance and compare multiple architectures without committing to one installation.
EY is making a different bet. It is treating operational experience as a compounding asset whose value begins before the hardware produces economic advantage.
Who else is building on-site quantum capacity?
EY is unusual among consultancies, but it is part of a broader move to place quantum systems beside institutional computing infrastructure.
| Organization | Installation | Strategic purpose |
| Cleveland Clinic | IBM Quantum System One installed on its Cleveland campus in 2023 | Dedicated healthcare and life-sciences research |
| RIKEN, Japan | Quantinuum trapped-ion system installed at its Wako campus and integrated with the Fugaku supercomputer environment | Hybrid quantum and high-performance computing research |
| OVHcloud, France | Quandela photonic system installed in OVHcloud infrastructure | Internal capability and customer quantum access |
| VTT, Finland | Successive superconducting systems developed with IQM | National research, industrial access and domestic capability |
| Leibniz Supercomputing Centre, Germany | IQM superconducting system integrated into a high-performance computing centre | Quantum and HPC integration |
| Forschungszentrum Jalich, Germany | Multiple quantum systems and testbeds integrated through the Jalich supercomputing environment | Multi-architecture research and hybrid computing |
| CINECA, Italy | IQM system selected for deployment with its supercomputing infrastructure |
The pattern is revealing. Most disclosed on-site systems are still located at research institutes, supercomputing centres, healthcare research campuses or cloud providers. They are infrastructure for experimentation and capability-building, not ordinary enterprise production equipment.
Vendors are now productizing this model. IQM sells complete on-premises superconducting systems. Rigetti markets systems for local deployment. Quandela offers photonic machines designed for data-centre environments. IBM installs and manages dedicated systems at selected customer sites. Quantinuum has demonstrated that a trapped-ion system can be deployed beside national supercomputing infrastructure.
This is a real market, but not yet a mature corporate procurement category. Both the Canadian and US governments are investing in quantum infrastructure, although neither has made public sector ownership of complete quantum computers a focus. Canada launched its C$360 million National Quantum Strategy in 2023 to support computing hardware and software, communications, sensing, talent and commercialization. It has since committed another C$92 million through the Canadian Quantum Champions Program to four domestic hardware companies, Anyon Systems, Nord Quantique, Photonic and Xanadu, with the explicit goal of anchoring quantum companies, infrastructure and talent in Canada. Federal departments are also funding testbeds, secure communications, quantum-safe cryptography and defence applications, but Canada has not announced a general-purpose federal quantum computing facility comparable with its major classical supercomputing centres. The United States began earlier and has built a much larger federally coordinated research system under the National Quantum Initiative Act. The Department of Energy operates five National Quantum Information Science Research Centers, national-laboratory testbeds, quantum networking facilities and programs that give researchers access to commercial quantum systems alongside federal supercomputers. Its 2027 budget request also proposes exploring a dedicated quantum-computing user facility capable of eventually hosting a scientifically useful fault-tolerant system. In both countries, current federal policy is therefore focused less on using quantum computers for ordinary government operations than on building domestic suppliers, research infrastructure, security readiness and the capacity to integrate future quantum machines with high-performance computing.
Europe is a significantly more mature market. China has taken a more state-directed and infrastructure-led approach than either Canada or the United States, although the scale of its spending is unusually opaque and the frequently cited US$15 billion figure cannot be verified from a public national budget. Quantum technology has been designated a strategic national priority across successive five-year plans, with government-backed work concentrated around the Chinese Academy of Sciences, the University of Science and Technology of China, the national quantum laboratory complex in Hefei, state-owned telecommunications companies and domestic hardware suppliers. China has built physical infrastructure across several layers: the Micius quantum science satellite and associated ground stations, a roughly 2,000-kilometre Beijing-Shanghai quantum-communications backbone, experimental photonic and superconducting processors, and cloud platforms providing access to domestically manufactured machines. The 72-qubit Origin Wukong, launched in 2024, is a programmable, deliverable superconducting computer built with Chinese hardware and software and made remotely available to users internationally. Chinese researchers have also produced the Jiuzhang photonic systems and Zuchongzhi superconducting processors, while state-controlled China Telecom has invested directly in quantum companies and services. China is therefore not merely funding research or purchasing access from commercial vendors. It is attempting to build a vertically integrated national quantum stack encompassing chips, control systems, computers, networks, satellites, cloud access and state-backed commercial deployment. That makes its strategy closer to sovereign industrial capacity-building than the research-network model used by the United States or Canada’s smaller supplier-development strategy.
Why Canada is a logical place to put it
Canada is not merely the jurisdiction EY selected for privacy reasons. It has one of the world’s deepest quantum ecosystems. The country’s strength is unusually broad.
The University of Waterloo’s Institute for Quantum Computing, founded in 2002, has helped train more than 3,000 researchers. Waterloo is also home to the Perimeter Institute for Theoretical Physics and a dense network of quantum researchers and companies.
Toronto is home to Xanadu, which builds photonic quantum computers and created PennyLane, a widely used open-source quantum software framework. British Columbia produced D-Wave, one of the earliest commercial quantum-computing companies and a pioneer in quantum annealing.
Quebec has major research clusters around Université de Sherbrooke, Institut quantique and the DistriQ quantum innovation zone. Canadian firms also span full-stack superconducting systems, error correction, photonics, quantum networking and quantum-safe security. They include Anyon Systems, Nord Quantique, Photonic, evolutionQ, Crypto4A and ISARA.
Canada’s National Quantum Strategy, launched in 2023 with C$360 million in dedicated funding, is organized around computing, communications and sensing, supported through research, talent and commercialization. In late 2025, the federal government added a first phase of support for domestic quantum computing companies including Anyon Systems, Nord Quantique, Photonic and Xanadu.
The ecosystem also includes sophisticated potential users. Canadian banks, governments, defence organizations, utilities, universities and pension funds manage the kinds of optimization, security and risk problems quantum companies want to address.
This gives EY access to researchers, startups, trained personnel, government programs and regulated clients in the same country as the installation.
Canada’s weakness has often been commercialization and domestic adoption rather than research quality. EY’s installation helps close that gap. It gives a major global enterprise a reason to build applications, talent and supplier relationships around Canadian quantum capacity.
What executives should do now
Most companies do not need a quantum computer, or even access to one.
But they should become capable of making an informed decision about quantum computing before a vendor, competitor or cryptographic deadline makes the decision for them.
1. Separate computing readiness from security readiness
Post-quantum cryptography is an immediate enterprise program. Organizations should inventory cryptographic assets, identify data that must remain confidential for many years, map supplier dependencies and plan migration to standardized quantum-resistant algorithms.
Buying quantum computing capacity is not required to begin that work.
2. Learn the problem classes, not the slogans
Executives do not need to become quantum physicists. They do need to understand why a proposed problem might have a quantum advantage.
Ask which known quantum algorithm applies, what the best classical alternative is, how performance will be measured, how data will be encoded and whether the claimed advantage includes preprocessing, error mitigation and repeated execution.
3. Build a portfolio of candidate problems
Start with expensive, recurring problems in chemistry, materials, optimization, risk or simulation. Rank them by business value, mathematical suitability, data readiness and the likelihood that quantum hardware can eventually improve the result.
Do not begin with a quantum solution in search of a problem.
4. Use cloud systems before considering ownership
Cloud platforms let teams compare superconducting, trapped-ion, neutral-atom, photonic and annealing systems. That flexibility is valuable while the technology is changing quickly.
Dedicated hardware becomes a serious question only when access, sovereignty, integration, intellectual property, research intensity or talent development justifies the cost.
5. Require classical benchmarks
Every quantum experiment should have a strong classical baseline. A demonstration is not commercially useful because it ran on qubits. It matters only if it improves cost, speed, accuracy, energy use or an otherwise unreachable capability after the complete workflow is counted.
6. Educate a cross-functional team
Quantum decisions should not sit only with innovation staff. The learning group should include technology, cybersecurity, data, legal, procurement, risk and the business unit that owns the candidate problem.
Canadian executives can draw on the Institute for Quantum Computing, Perimeter Institute, Institut quantique, Quantum Industry Canada, federal quantum-readiness materials and direct engagement with domestic hardware, software and security companies.
7. Treat vendor claims as hypotheses
Ask for architecture, logical and physical qubit counts, fidelity, error rates, circuit depth, uptime, queueing, benchmark methodology and the exact boundary of any performance comparison.
If those details are absent, the claim is not yet ready for an investment decision.
Joe Depa, EY’s Global Chief Innovation Officer, captured the strategic distinction: “AI changes how work gets done, while quantum may change what’s possible.”
It now has to decide who may use the machine, what may be processed on it, how it connects to classical infrastructure, how results are verified, how errors are understood, how vendor access is controlled, how intellectual property is protected and which use cases survive contact with real hardware.
Sources
● EY announcement, July 29, 2026
● EYQ enterprise adoption case study
● Government of Canada National Quantum Strategy
● Government of Canada quantum computing readiness guide
● Cleveland Clinic and IBM on-site installation
● Quantinuum installation at RIKEN
● Quandela quantum computing on OVHcloud
● Institute for Quantum Computing
