At ETH Zurich, an experiment with superconducting qubits certifies truly unpredictable numbers, with implications for data security.

Cybersecurity relies on an often invisible premise: the availability of truly unpredictable numbers. Cryptographic keys, electronic signatures, identity protocols, authentication systems, and confidential communications depend on the quality of the randomness used to generate them. For many everyday applications, a seemingly random sequence is sufficient. For the security and privacy, however, even a minimal deviation can become a weak point.

The case presented by theETH in Zurich concerns precisely this technical issue. A research group led by Renato Renner e Andreas Wallraff, in Department of Physics of the Swiss university, has created certifiably perfect random numbers through a quantum experiment. The news, published on 27 May 2026 and linked to an article on Nature, indicates a relevant passage for the cryptography and for trusted architectures based on physical hardware.

The point is not just to produce random bits, i.e. sequences of 0 e 1The most delicate issue is demonstrating that those bits are not affected by instrument imperfections, residual correlations, systematic errors, or statistical biases. The Zurich experiment addresses the problem with a combination of entangled superconducting qubits, Bell test and a protocol of randomness amplification, that is, amplification of randomness.

Why randomness is a critical industrial resource

In industrial practice, a random number generator is not a neutral component. It is part of the security chain. If it produces even partially predictable sequences, it can weaken keys, tokens, signatures, or encryption procedures.Applied Cryptography Group ETH notes that the security of many cryptosystems depends on perfect randomness, especially in key generation, but that this condition is difficult to guarantee in real-world settings.

Modern generators can exploit physical phenomena, including quantum effects such as the reflection of photons off beam splitters. However, according to the ETH press release, even these devices are not automatically immune to systematic errors. In operational terms, this means that a sequence may appear random to a standard statistical test and still be vulnerable to more sophisticated analysis or unobserved defects in the device.

“It may seem strange, but it is almost impossible to create a perfect coin or die,”

says Renato Renner.

The coin metaphor is useful because it makes the problem accessible. A physical object, even one designed with high symmetry, is susceptible to minimal irregularities in mass, friction, spin, or surface. In the computational world, this defect can take various forms: correlations between successive bits, non-uniform noise, component instability, dependence on environmental conditions, or insufficient quality of the initial entropy.

For businesses, banks, public infrastructures, and cloud providers, the consequences are tangible. The robustness of a cryptographic solution depends not only on the algorithm, but also on the random material that powers key generation. From this perspective, ETH research should not be viewed as an abstract exercise in quantum physics, but as an advancement in research and development with potential repercussions on high-security systems.

The role of the Bell test and superconducting qubits

The solution adopted in Zurich uses two superconducting chips, cooled to temperatures close to absolute zero. Each chip represents a quibit, the elementary unit of quantum information, capable of assuming states 0, 1 or an overlap of both. The two chips are connected by a long cryogenic tube 30 meters, through which microwave photons travel, capable of creating entanglement.

The entanglement It is a quantum correlation that has no direct classical equivalent. In this experiment, it allows measurements on the two qubits to be linked without any exchange of information at the speed of light during the measurement window. The distance of 30 meters It serves precisely to support this condition of causal separation, which is necessary to prevent the result from being explainable through ordinary communications between devices.

The Bell test comes into play here as a certification tool. It doesn't just produce data, but allows us to verify that the observed correlations are not compatible with a local classical explanation. The article on Nature specifies that the amplification of randomness is device-independent: does not require detailed assumptions about the internal workings of the devices, but depends on performing a loophole-free Bell test within a specific experimental regime, with high Bell violation and high repetition rate.

“This was made possible by an improved Bell test, with high quality and high speed of data production,”

explains Andreas Wallraff.

This step is important because it separates random number generation from blind trust in the device. In many industrial contexts, hardware is treated as an element certified by audits, standards, or manufacturing processes. Here, however, certification derives from observable physical behavior: if the system passes the conditions required by the protocol, the resulting randomness can be certified even without knowing every microscopic detail of the device.

From imperfect randomness to certifiable sequences

The technique used is called randomness amplificationThe principle is to transform an imperfect initial source into a much more reliable sequence, up to the certifiably perfect randomness described by ETH. In the experimental case, the choice of measurement basis on the two qubits depends on a non-ideal initial generator; the measurement results are then processed with an algorithm developed by Renner's group.

According to the abstract published by NatureRealistic quantum devices are inherently imperfect and generate random bits that must be improved for applications such as cryptographic key generation. The experimental demonstration was made possible by the combination of theoretical advances, which brought the protocol to a realizable regime, and experimental progress in superconducting circuits.

A central aspect is the comparison with classical computer science. Nature points out that randomness amplification has been proven impossible by classical means alone; therefore, the experiment is presented as a case of quantum advantage, that is, the use of a quantum property to perform a task that cannot be achieved through classical information.

The distinction is relevant for the industrial transferabilityIt's not enough to say that a generator uses quantum phenomena. The key is to determine what guarantees it offers, what assumptions it requires, and what level of certification it produces. In a market where quantum vocabulary is often widely used, the difference between quantum generation, device-independent certification, and formal randomness amplification is substantial.

The Nature study also reports that the data and code for randomness extraction and analysis have been deposited in the ETH Zurich Research Collection and made publicly available. This element is important for the scientific community because it allows for verification, methodological comparisons, and possible subsequent developments on protocols, equipment, and evaluation metrics.

Implications for security, identity, and infrastructure

The applications cited by ETH include encryption of sensitive communications, digital identities, public randomization services, lotteries, and blockchain applications. In all these cases, the value lies not only in the generation of bits, but also in the ability to rely on a physically certified source. The university compares the potential role of this work to that of atomic clocks for time measurement: a technical benchmark on which other systems can build trust.

For the post-quantum cryptography And for quantum-secure communications, the issue is even more sensitive. Robust algorithms and advanced protocols can be compromised by a weak random basis. The quality of entropy is therefore not an accessory, but a design prerequisite. In the medium term, certified sources could become specialized components for data centers, high-security network nodes, government infrastructure, and trust services.

Application maturity, however, must be assessed with caution. The experiment requires superconducting chips, cryogenic cooling, controlled physical connections, and highly precise measurement conditions. These elements are far from routine integration into standard consumer devices or enterprise equipment. The immediate value is therefore primarily scientific and infrastructural: it demonstrates that a threshold previously considered theoretical can be reached in the laboratory, opening the door to engineering optimizations.

For security companies, the operational message is twofold. On the one hand, the demand for verifiable randomness could grow with the growth of critical systems based on identities, automated transactions, software signatures, and encrypted communications. On the other, source certification will become an increasingly important technical criterion in specifications, regulated systems, and services where proof of reliability is as important as performance.

A cautious advance, but with reference value

ETH's work doesn't eliminate the practical problems of cybersecurity. It's no substitute for proper protocol design, key management, algorithm updates, or endpoint protection. However, it adds a piece to the physical foundation of trust: the ability to obtain a sequence of random bits certified not only by statistical tests, but by observable quantum properties.

The most concrete prospect concerns the evolution of randomness generation devices for highly secure environments. Industries such as finance, telecommunications, defense, public identity, and cloud infrastructure could view these developments not as immediate products, but as a reference for future standards, security benchmarks, and hybrid architectures combining quantum physics and applied cryptography.

The news also has broader significance for the European tech industry. It shows how a fundamental physics result can impact regulated markets, trust models, and data protection infrastructures. In this sense, the contribution of theETH in Zurich falls within a line of research in which quantum hardware, teoria dell'informazione e computational security converge on measurable industrial problems.

Perfect randomness remains a demanding concept, difficult to produce and even more difficult to certify. However, the Swiss experiment suggests that, at least under controlled conditions, quantum physics can transform an imperfect source into a reliable resource for systems that depend on unpredictability. For an economy increasingly based on encrypted communications, verifiable identities, and connected infrastructures, this is a progress worth following without fanfare, but with attention.

Here are three insights that might interest you:

Jiuzhang 4.0: 25,6 microseconds to challenge supercomputers
Votegral, the electronic vote that overcomes the challenge of coercion
Those defects in semiconductors useful for quantum supercomputers