Q.ANT has announced the availability of its second-generation Native Processing Unit, the NPU 2, featuring enhanced nonlinear processing capabilities that deliver substantial gains in energy efficiency and performance for artificial intelligence and high-performance computing workloads. The new processor performs nonlinear mathematics natively in light, enabling entirely new classes of AI and scientific applications including physical AI, advanced robotics, next-generation computer vision, industrial intelligence, and physics-based simulation.
According to Dr. Michael Förtsch, CEO of Q.ANT, the company offers a new class of processors that enable performance gains beyond incremental improvements of digital counterparts. For years, AI has raced ahead of our ability to power it, with energy becoming the new frontier. With our NPUs, we've changed the equation, he stated. The NPU 2 demonstrates that performance and sustainability are not opposing forces but rather one and the same, representing what the company describes as a new beginning rather than an evolution.
The timing of this advancement comes as AI acceleration reaches the physical limits of silicon technology. Each new generation of GPUs consumes more power and water while producing more heat, with cooling systems accounting for up to 40 percent of total data-center energy consumption. Photonic processing fundamentally changes this equation by using light that travels faster, generates almost no heat, and can execute complex functions in a single optical step that would require thousands of transistors in traditional CMOS chips. Q.ANT's architecture delivers up to 30 times lower energy use and 50 times higher performance for complex AI and HPC workloads.
Q.ANT will debut its second-generation Native Processing Unit at Supercomputing 2025 in St. Louis from November 17-21. At the LRZ booth #535, the company will run a live image-based AI learning demonstration powered by the Q.ANT Photonic Algorithm Library on its photonic processors. The Q.PAL offers developers efficient, nonlinear algorithms and functions for complex workloads, continuously enhanced and optimized by Q.ANT for application-oriented photonic processing. The demonstration will show how Q.ANT's photonic processors achieve more accurate results with fewer parameters and fewer operations compared to conventional CPU-based systems.
Visitors can test how the NPU learns images within seconds using a nonlinear neural network, marking significant progress from simple digit recognition to image classification and learning within just one year. Dr. Förtsch emphasized that photonic computing is scaling much faster than CMOS technology, achieving in one year what took ten years for digital computing. The second generation of our Native Processing Unit shows how rapidly this transition is happening and why efficient, light-based computation will drive the next wave of AI and HPC, he explained.
The enhanced NPU 2 features an improved nonlinear processing core with optimized analog units for nonlinear network models that dramatically reduce parameter counts and training depth while improving accuracy for image learning, classification, and physics simulation. The system is delivered as a turnkey 19-inch rack-mountable server called the Native Processing Server NPS, containing multiple NPU 2 processors and integrating seamlessly with existing CPUs and GPUs via PCIe and C/C++/Python APIs, making photonic acceleration immediately deployable in HPC and data-center environments.
In practical applications such as manufacturing, logistics, and inspection, photonic processors can execute nonlinear neural networks far more efficiently. This enables visual AI to recognize defects, track objects, and optimize inventories with fewer parameters, dramatically reducing energy costs and making computer vision systems economically viable for tasks previously considered too compute-intensive. Photonic processors will accelerate next-generation AI architectures, including hybrid models that combine statistical reasoning with physical modeling, advancing domains such as drug discovery, materials design, and adaptive optimization where both nonlinear complexity and extreme energy efficiency are essential. Q.ANT servers equipped with NPU 2 processors are available to order now, with customer shipments scheduled for the first half of 2026. Additional information about the company and its technology can be found at https://www.qant.com.
