Catenaa, Sunday, October 19, 2025-Telstra has advanced its AI capabilities by testing a quantum-based machine learning processor, dubbed Watermelon, in collaboration with Sydney’s Silicon Quantum Computing (SQC).
The system is designed to enhance predictive network performance and detect anomalies faster than classical computing models.
Over the past year, Telstra applied Watermelon to its existing AI/ML infrastructure, which monitors latency, bandwidth, and other network metrics to preempt performance issues.
Using quantum reservoir computing, Watermelon feeds outputs back into the neural network, improving memory and prediction accuracy, particularly when handling sparse or noisy data.
Telstra and SQC reported that Watermelon could train and fine-tune a quantum reservoir in days, a task that would take weeks using conventional deep learning models.
The system also operates efficiently without heavily relying on GPUs, which are typically required to power large-scale AI models.
Shailin Sehgal, Telstra’s group executive for global networks and technology, said the company is exploring quantum computing to create smarter connectivity experiences, from personalized services to proactive issue prevention.
SQC chief executive Michelle Simmons described the project as a rare real-world application of quantum systems in telecommunications, highlighting the processor’s ability to uncover complex relationships within classical data.
Industry analysts note that as AI consumption is projected to reach up to 326 terawatt-hours annually by 2028, quantum systems like Watermelon could help reduce environmental impact by completing intensive computations in a fraction of the time.
