Q.ANT

Q.ANT GmbH is a German deep-tech scale-up developing photonic processing solutions that compute natively with light, positioning itself as a scalable alternative to transistor-based (CMOS) computing. Founded in 2018 by Michael Förtsch, who serves as CEO, the company is headquartered in Stuttgart, Germany. Its stated mission is to unlock AI's potential through a fundamental rethink of computing, with the belief that the future of AI computing runs on light. The company's flagship offering is the Q.ANT Native Processing Server (NPS), described as the world's first commercial photonic AI processor. The NPS is a 19-inch rack-mountable server that executes demanding AI and HPC workloads directly in the optical domain, communicating seamlessly with existing computing infrastructure. Q.ANT reports that the system is deployed at LRZ and Jülich, two major German supercomputing centers. The company claims the platform can deliver up to 30x higher energy efficiency and 50x faster computation, with higher computational density, reduced infrastructure complexity, lower cooling requirements, and reduced operational costs for data centers. Q.ANT's technology targets workloads where photonic processors structurally outperform CMOS, particularly large-scale nonlinear networks. According to the company, a single optical element can replace 100 to 1,000 transistors for the same nonlinear function, and native nonlinear operations reduce the number of model parameters needed for higher accuracy per parameter and training budget. The second generation of its photonic processor (NPU Gen 2) is presented as delivering 8 GOPS sustained throughput on nonlinear functions and providing a foundation for wavelength multiplexing as a path to optical scaling. On the software side, Q.ANT provides the Q.ANT Photonic Algorithms Library (Q.PAL), C/C++ and Python APIs, and integration with frameworks such as PyTorch, turning photonic computing into real applications spanning AI training and inference, machine learning, physics simulations, and time-series analysis. In collaboration with the Institute for Microelectronics Stuttgart (IMS CHIPS), Q.ANT operates its own pilot line for photonic chips based on the thin-film lithium niobate (TFLN) material system. The company has been recognized as a Sample Vendor in three Gartner Hype Cycle 2025 reports.

Q.ANT GmbH was founded in 2018 by Michael Förtsch and is headquartered in Stuttgart, Germany. The company emerged as a deep-tech scale-up focused on photonic processing, aiming to build a scalable alternative to transistor-based computing by computing natively with light. Its founding premise was that unlocking AI's true potential requires a fundamental rethink of computing itself, and that the future of AI computing runs on light. From its early years, Q.ANT pursued an integrated hardware-and-software approach. On the hardware side, it developed analog photonic co-processors optimized for complex computations, and in collaboration with the Institute for Microelectronics Stuttgart (IMS CHIPS) it established its own pilot line for photonic chips based on the thin-film lithium niobate (TFLN) material system. This pilot line gave the company in-house capability to fabricate photonic integrated circuits suited to its native computing architecture. A major commercial milestone was the introduction of the Q.ANT Native Processing Server (NPS), which the company describes as the world's first commercial photonic AI processor. The NPS is a 19-inch rack-mountable server that executes demanding AI and HPC workloads directly in the optical domain while communicating seamlessly with existing computing infrastructure. Q.ANT reports deployments at LRZ and Jülich, two leading German supercomputing centers, giving the technology early real-world validation in high-performance computing environments. The company subsequently unveiled a second generation of its photonic processor, NPU Gen 2, targeting the next wave of AI and HPC. Q.ANT states that NPU Gen 2 delivers 8 GOPS sustained throughput on nonlinear functions — workloads where photonic processors structurally outperform CMOS — and introduces native nonlinear operations in which one optical element replaces 100 to 1,000 transistors for the same function. The generation also lays the foundation for wavelength multiplexing as the basis for the next generation of optical scaling. Q.ANT frames the trajectory as a path toward up to 30x higher energy efficiency and 50x faster computation, unprecedented compute density and bandwidth, and radical reductions in operational costs for data centers. Alongside hardware, Q.ANT built out a software stack to make photonic computing usable in real applications. The Q.ANT Photonic Algorithms Library (Q.PAL) serves as the interface to the NPU, complemented by C/C++ and Python APIs and integration with frameworks such as PyTorch. The software targets AI training and inference, machine learning, physics simulations, and time-series analysis. Q.ANT argues that photonic processors are the natural hardware fit for large-scale nonlinear algorithms, allowing models with fewer parameters to achieve higher accuracy per parameter and training budget. By 2025, Q.ANT had grown into a recognized player in the emerging photonic computing space, employing a multinational team and holding a portfolio of patents. The company was recognized as a Sample Vendor in three Gartner Hype Cycle 2025 reports, reflecting its visibility in data center infrastructure and related technology categories. Its leadership includes founder and CEO Michael Förtsch and VP Software Utz Bacher, and it maintains an advisory board including partners from the venture capital community.

• 2025

  Recognized as a Sample Vendor in three Gartner Hype Cycle 2025 reports.

• 2018

  Q.ANT founded by Michael Förtsch; headquartered in Stuttgart, Germany.

• Established own pilot line for photonic chips based on thin-film lithium niobate (TFLN) in collaboration with IMS CHIPS.

• Introduced the Q.ANT Native Processing Server (NPS), described as the world's first commercial photonic AI processor, deployed at LRZ and Jülich.

• Unveiled second-generation photonic processor (NPU Gen 2) with 8 GOPS sustained throughput on nonlinear functions and native nonlinear operations.

Deep-tech scale-up delivering the world's first commercial photonic AI processor, positioned as an energy-efficient, scalable alternative to CMOS for AI and HPC data centers.

• Q.ANT Native Processing Server (NPS)

  19-inch rack-mountable server and the first commercial photonic AI processor, executing AI and HPC workloads directly in the optical domain.

• NPU Gen 2

  Second-generation photonic processor delivering 8 GOPS sustained throughput on nonlinear functions, native nonlinear operations, and a foundation for wavelength multiplexing.

• Q.ANT Photonic Algorithms Library (Q.PAL)

  Software library and interface to the NPU, with C/C++ and Python APIs and PyTorch integration for AI training, inference, machine learning, physics simulations, and time-series analysis.

1. 1Q.ANT – Photonic Computing for efficient AI and HPCqant.com
2. 2About Us - Pioneering Photonic Technology - Q.ANTqant.com
3. 3The photonic AI accelerator - Q.ANT (software)qant.com
4. 4The photonic AI accelerator - Q.ANT (photonic computing)qant.com
5. 5Q.ANT GmbH - Wikidatawikidata.org