Q.ANT is poised to transform the high-performance computing (HPC) sector with its photonic Native Processing Server (NPS), making its debut at ISC 2025. This event will feature the first live demonstration of functional photonic computing, highlighting its potential to drastically improve energy and computational efficiency for AI, physics simulations, and other scientific applications.
The NPS, powered by Q.ANT's Light Empowered Native Arithmetic's (LENA) architecture, achieves up to 30 times the energy efficiency of conventional technologies. It features 16-bit floating point precision with 99.7% accuracy, requires 40–50% fewer operations for the same output, and eliminates the need for active cooling, offering significant cost and energy savings.
At the core of Q.ANT's system is a proprietary thin-film lithium niobate (TFLN) photonic chip, enabling high-speed, low-loss optical modulation without thermal crosstalk. This innovation not only reduces energy-intensive cooling needs but also allows for up to 100x greater compute density per rack in data centers and up to 90x lower power consumption per application.
Bob Sorensen of Hyperion Research notes Q.ANT's success in addressing integration and precision challenges in photonic computing. The NPS's high accuracy in nonlinear and mathematical operations makes it a strong contender against digital processors, especially in AI inference, physics simulations, and image analysis.
Q.ANT's photonic architecture ensures easy integration with existing digital infrastructure, supporting popular frameworks like TensorFlow, PyTorch, and Keras. This compatibility allows AI and HPC early adopters to benefit from photonic computing without significant system changes.
Dr. Michael Förtsch, Q.ANT's CEO, highlights the technology's potential to revolutionize HPC economics by enabling more efficient, scalable, and sustainable computing solutions through light-based mathematical transformations.
The photonic NPS is ideal for data-heavy applications, including scientific simulations, advanced image processing, and large-scale AI inference and training, offering simplified AI model architectures and reduced system demands by computing nonlinear functions and Fourier transforms directly with light.
