A cross-industry team spanning quantum, banking, and technology sectors demonstrates that cryptographic-grade randomness can be generated and continuously verified on a production-ready chip.
‍

A team of researchers working on quantum technology, drawn from banking and finance, commercial technology, and academic sectors, has published a peer-reviewed paper in a prestigious journal, PRX Quantum, demonstrating the world's first fully integrated, self-testing quantum random number generator (QRNG) on a silicon photonic chip, QDice^®. 

The paper, titled "Self-testing quantum randomness expansion on an integrated photonic chip", addresses a critical problem in modern security infrastructure: how does the user certify that the output of their random number generator is private and secure, especially when the performance of hardware components drift over time?

The Problem with Randomness

Random number generation underpins nearly every layer of digital security, from encrypting financial transactions and protecting cloud infrastructure to securing communications across public and private networks.

Trust is the hidden fineprint of every random number generator in production today. Whether classical or quantum, every device shipped to date has come with an implicit contract: the user must trust that the hardware was correctly characterised upon manufacture and trust that the hardware’s performance stays that way throughout its entire life cycle. However, neither can be guaranteed in practice as components age and conditions change. Furthermore, devices that drift or are tampered with can begin producing predictable output, and nothing in the existing verification stack will catch it. For security teams in banking, critical infrastructure, and enterprise technology, this is not a theoretical concern. It is an unresolved gap at the foundation of their cryptographic infrastructure.

How QDice^® Changes Everything

Published in PRX Quantum on 5 June 2026, the peer-reviewed paper 'Self-testing quantum randomness expansion on an integrated photonic chip' introduces QDice^®: a fully integrated, self-testing quantum random number generator (QRNG) built on a single silicon photonic chip.

QDice^® addresses the above-mentioned issues through a measurement-device-independent protocol, a design choice that reframes where trust is required. Rather than demanding users to trust every internal component, the protocol treats the measurement devices as untrusted and uncharacterised “black boxes”. During operation, the measurement devices undergo a test and are given a score. When the score meets the required threshold, certified random bits are produced. When it does not, the protocol aborts, ensuring security.
‍

“QDice^® sits in a very sweet spot between the security versus practicality trade-off, making it one of the most secure QRNGs that are commercially available in the market.”

Dr Ng Hong Jie, Senior Software Engineer, Squareroot8 Technologies

‍

The chip was fabricated at Singapore’s Advanced Micro Foundry. It operates at room temperature, requires no cryogenic cooling, and is fabricated on a standard eight-inch silicon-on-insulator wafer process, which is the same production platform used in data centres and telecommunications hardware today. This makes scaling up through existing semiconductor foundries commercially viable.

Where QDice^® Sits on the Security Spectrum

This work demonstrated the implementation of a QRNG with an unprecedented level of security on a single chip. The security analysis assumes that the threat actor may hold quantum correlations with the QDICE^® chip and the characterisation of the entire chip other than the quantum light source may be untrusted. Such a security framework has no equivalent classical analogue, as it harnesses the essence of quantum information science. This means that QDICE^® QRNG is offering protection that no classical device could – a unique value proposition that cannot be ignored in the quantum era.

Comparing QDICE^® with fully-trusted QRNGs, the current trade-off is clear: speed. While a typical QRNG often operates above 100 gigabits per second, the QDICE^® produces random numbers at 64 bits per second. While the numbers may be alarming, it should not be surprising, as working in a tighter security framework means less randomness can be certified, which naturally impedes its speed. To close the speed gap, the team has fabricated photodiodes that would improve the detector’s efficiency from the current 69 per cent to 92.4 per cent. This improvement would translate to a speed of 68 megabits per second, demonstrating the commercial viability of the QDICE^® technology.
‍

“Random numbers play a central role in modern security and digital trust. Quantum theory allows us to check for signatures in these numbers to verify their secrecy. What we have done here is turn quantum light into verifiably secure random numbers with the entirety of the process within a single chip smaller than the size of your fingernail. This will undoubtedly be a major step towards mass adoption of quantum technologies.”
‍
‍Dr Goh Koon Tong, Co-Founder and Chief Technologist, Squareroot8 Technologies

‍
What Comes Next

The implications reach well beyond any single sector. From financial systems, cloud infrastructure, government networks, and medical devices to AI decision pipelines or connected hardware at scale, the inability to continuously verify randomness is a structural vulnerability. QDice^® is the first device to close that gap on a production-ready chip. The same silicon platform that makes it manufacturable at scale also makes it integrable into data centre hardware, network security appliances, and, as the technology matures, into the compact devices that underpin the Internet of Things.

‍

The research was conducted by a cross-industry team drawn from the quantum technology sector, global banking and financial services, multinational technology, and academic research institutions around the world, reflecting the breadth of industries for which this technology has direct relevance.

‍

Published paper: APS Journals

‍