Integrating quantum key distribution with photonic neural networks for secure and efficient AI computing in defense and cloud systems

Secure and efficient computing of AI is highly essential in modern defense and cloud infrastructure, for which traditional electronic architecture often faces performance and security bottlenecks. This paper presents an integrated framework that incorporates QKD with photonic neural networks to realize quantum-secured ultrafast AI computation. The integrated framework establishes QKD-secured channels for securing AI data transmission while using optical neural networks for high-speed inference and low-energy processing. Experimental simulation confirms the integrated approach attains as high as 40% latency reduction and 35% energy saving when compared with state-of-the-art conventional AI systems secured by classical cryptography. Moreover, the integrated QKD enables provably secure key exchange resistant against both classical and quantum attacks. This work thus points out the potential of quantum-photonic integration as a base for scalable, secure, and high-efficiency AI computing both in defense and cloud applications.