Strengthening IT capabilities for critical security functions
In digital systems, processing information is a fundamental task. Different types of computation typically benefit from different types of processors or semiconductor chips. A particularly demanding set of tasks are known as ‘randomness-intensive workloads’. These are computations whose performance and security depend critically on consuming large volumes of high-quality random numbers. Typical use cases would include cryptography, secure communications and data security. “The harder the numbers are to predict, the harder the protection is to break,” explains RPU(opens in new window) project coordinator Carlos Abellan from Quside(opens in new window) in Spain. “The quality of the randomness directly determines how secure a system is.”
Computer capacity for cryptographic execution
These workloads often have specific needs that existing computer platforms are unable to fully address. “Current solutions on the market fall short on two fronts,” says Abellan. “The quality and performance of their hardware randomness generation, and the compute capacity needed for cryptographic execution.” To address this, Quside has developed the Randomness Processing Unit (RPU). This is a hardware accelerator designed specifically for randomness-intensive workloads, particularly cryptographic applications. “What makes the RPU different is that it combines two capabilities in a single platform,” adds José Martinez, chief architect for the RPU project. “First, it includes built-in quantum random number generators that produce genuinely unpredictable randomness at high speed. Second, the RPU includes reprogrammable hardware, allowing users to adapt the platform to their own cryptographic algorithms, security protocols and application requirements.”
Optimised photonic chips
The RPU project, supported by the European Innovation Council(opens in new window), sought to advance this technology further through the development of photonic quantum random number generation chips. These were optimised for the power, size and performance needs of different hardware accelerators and applications. “Two targets in particular were pursued,” explains Abellan. “These were a platform for high-performance applications, for which an optimised photonic chip was developed, and a highly scalable embedded device, for which a compact, integrated solution combining photonics with a first silicon chip for control electronics was built.” At the technical level, the project involved designing photonic integrated chips to generate quantum randomness, and developing the hardware platform and architecture to run randomness-intensive workloads. Applications and integrations were built for artificial intelligence and cryptographic software, a task that was carried out alongside potential end users.
From data centres to space systems
The project produced four main results. “First, we developed a new photonic chip that halves the power consumption of randomness generation and cuts the control requirements fivefold,” says Abellan. “Second, we paved the way for the Garnet X5, an optimised RPU device for cryptographic applications, which will be commercialised.” Early prototypes are expected to be available in late 2026, with commercial launch planned for 2027. Quside was also able to build a wide range of integrations with cryptographic libraries, both open-source and vendor-based. “This means that adopting the technology requires almost no integration effort,” remarks Abellan. “Finally, we manufactured our first ASIC (a custom control chip), designed to run a highly optimised RPU in future embedded and edge devices. Together, these product lines will enable the deployment of RPU technology across a broad range of applications, from data centres and telecom infrastructure to edge, industrial and space systems.” In this way, the RPU project aims to become a key building block for next-generation security architectures across Europe, supporting critical sectors such as space, defence, telecommunications, data centres, critical infrastructure and government systems.
