One-third of the large global companies reviewed by McKinsey allocated more than $10 million to quantum-computing initiatives in 2025, even though near-term returns remain difficult to measure. Most of that spending went into applications, integration and skills rather than buying quantum computers.
The figure comes from a selection of large companies analysed for the McKinsey Quantum Technology Monitor 2026. It should not be read as evidence that one-third of all businesses are making investments on this scale.
Corporate spending can also mean very different things. A bank testing algorithms through a cloud service is making a different bet from IBM, which plans to invest more than $10 billion over five years in research, manufacturing, capital expenditure, partnerships and acquisitions.
McKinsey separately estimates that investment in quantum-technology start-ups reached $12.6 billion in 2025, with 90% going to quantum-computing companies. That is funding for suppliers, not evidence that their business customers are already earning a return.
What companies are paying for
A conventional computer stores information in bits, normally represented as either 0 or 1. A quantum computer uses quantum bits, or qubits, which can be controlled through effects such as superposition and interference. Carefully designed algorithms may use those effects to tackle certain calculations in ways that conventional machines cannot easily copy.
This does not make a quantum computer a faster replacement for every computer. It is unlikely to improve ordinary tasks such as payroll, email or spreadsheets. The strongest candidates involve narrower problems, including molecular simulation and some forms of optimisation.
Current machines are also prone to errors. Qubits are easily disturbed by their surroundings, so researchers use error correction to protect information. That can require many physical qubits to create a smaller number of more dependable logical qubits.
For most businesses, buying readiness therefore means hiring or training specialists, identifying suitable problems, developing algorithms, testing cloud-based machines and working out how quantum services might connect with existing systems. The company may never own the hardware.
A March 2026 OECD paper on quantum readiness, based on organisations in ten countries, described this as an exploratory process. It identified four recurring barriers: immature technology, unclear business applications, high access and training costs, and a shortage of people who understand both quantum computing and a specific industry.
Why learning early can still make business sense
A company cannot know whether quantum computing suits its problems by reading a supplier presentation. It needs a classical baseline, meaning the best result its existing computers can produce, and a clearly defined test for whether a quantum or hybrid approach performs better.
Chemical and pharmaceutical companies are investigating molecular simulation. Transport businesses are examining routing and scheduling problems. Financial institutions are testing risk models and portfolio calculations. These remain areas of experimentation rather than proof that quantum systems are routinely beating conventional software.
Starting early may still have a practical benefit. Quantum algorithms differ from ordinary software, and employees who can connect quantum methods with chemistry, logistics or finance take time to train. A small programme can also reveal that a proposed application is unsuitable before a company commits a much larger budget.
McKinsey estimates that quantum computing could create between $1.3 trillion and $2.7 trillion in worldwide economic value by 2035. This is a consultancy estimate of possible benefits across industries, not projected revenue for quantum-computing vendors or a guaranteed return for companies investing today.
The OECD offers a useful discipline for these programmes: firms need a specific business case and a planned series of tests. Without those, quantum work can remain an isolated experiment that absorbs money without improving the organisation’s ability to make a decision.
A laboratory milestone is not a business return
Scientific progress explains why some companies are reluctant to wait. In July 2026, IBM and University of Chicago researchers reported an experiment that they said met fundamental criteria for quantum advantage. The term means that a quantum machine has performed a computation beyond the practical reach of leading conventional methods.
The team executed what the University of Chicago described as a 70-logical-qubit demonstration while using error-detection methods to assess the reliability of the result. Its supporting paper was posted to arXiv, which means it was publicly available as a preprint rather than a peer-reviewed journal article.
The experiment involved a specialised circuit-sampling problem. It did not discover a medicine, improve a delivery network or increase an investment return. It showed progress in performing and checking a difficult quantum computation, not that a general commercial advantage had arrived.
That difference matters when approving a budget. A scientific benchmark can make future applications more plausible, but it does not supply the revenue estimate, operating cost or implementation plan needed for an investment decision.
Cybersecurity requires a separate decision
A sufficiently capable future quantum computer could break some forms of public-key cryptography used to protect communications and confirm digital identities. Businesses do not need to buy quantum-computing access to respond to this threat.
Post-quantum cryptography uses mathematical methods for establishing encryption keys and creating digital signatures that are designed to resist attacks from both conventional and quantum computers. These methods run on conventional systems. In 2024, the US National Institute of Standards and Technology published its first three standards, and its current guidance says organisations should begin migrating now.
That security work belongs in the technology budget even if a company has no credible use for quantum calculations. It involves finding vulnerable cryptography in software and equipment, deciding what must be replaced first and testing compatible products.
Large companies with a genuine computational bottleneck may justify a staged quantum programme. Smaller businesses may reasonably limit their spending to monitoring the field, developing selected skills and preparing their encryption systems.
The size of the budget is not evidence of readiness. A useful programme should produce a named business problem, a measurable comparison with conventional computing and a stopping point if the experiment does not justify further spending.
