Can Light Solve AI’s Insatiable Energy Problem?

AUSTIN, TX – April 23, 2026 – As the artificial intelligence boom strains global power grids and pushes silicon technology to its physical limits, a German pioneer is making a bold entry into the U.S. market, armed with a radical solution: computing with light.

Q.ANT, a Stuttgart-based startup, today announced the opening of its U.S. headquarters in Austin, Texas. The move is headlined by the appointment of semiconductor titan Bruno Spruth, formerly the Vice President of Power Processor Development at IBM, as its new Chief Technology Officer. The company is betting that its photonic processors, which run complex calculations using light instead of electricity, can offer a way out of the escalating energy and heat crisis plaguing the world’s data centers.

The AI Energy Dilemma

The rapid advancement of AI has come at a staggering environmental cost. Data centers, the backbone of the digital world, are projected to more than double their electricity consumption by 2030, largely driven by the demands of training and running AI models. Cooling systems alone can account for up to 40% of a facility's energy use, a direct consequence of the immense heat generated by billions of transistors packed onto conventional silicon chips.

Q.ANT aims to break this cycle. The company has developed what it calls Native Processing Units (NPUs) that execute mathematical operations directly in the optical domain. Because photons, the particles of light, do not encounter electrical resistance like electrons, the process generates negligible heat. This fundamental difference allows for claims of up to 30 times greater energy efficiency and 50 times the performance of conventional processors for specific AI and high-performance computing (HPC) workloads.

"Q.ANT’s U.S. expansion offers hyperscalers, data centers, and innovators the opportunity to explore how they can reduce energy consumption and improve compute performance through our Native Processing Server," said Michael Förtsch, CEO of Q.ANT. "Bruno has spent his career at the center of modern computing. He knows its limits, and that the future of innovation requires reinvention. That is exactly what we are doing at Q.ANT."

This isn't just theoretical. In 2025, the company deployed the world's first commercial photonic processor in a live production environment at Germany's prestigious Leibniz Supercomputing Centre (LRZ). There, its processors are already tackling real-world problems in climate modeling, medical imaging, and fusion energy research, demonstrating the technology's viability beyond the lab.

Beyond Silicon: A New Computing Architecture

At the heart of Q.ANT's technology are photonic integrated circuits built on a cutting-edge material platform: Thin-Film Lithium Niobate (TFLN). Often dubbed a "holy grail" material by materials scientists, TFLN boasts an exceptional ability to manipulate light at high speeds with minimal energy loss. This allows Q.ANT to build processors where a single optical element can perform the work of thousands of transistors for certain calculations, dramatically accelerating tasks that are bottlenecks for traditional GPUs, such as nonlinear mathematical operations common in advanced AI.

A significant breakthrough for the company has been achieving 16-bit floating-point precision, a level of accuracy essential for training modern neural networks. This has long been a hurdle for optical computing, and its achievement signals a new level of maturity for the technology, moving it from a niche concept to a practical tool for AI developers.

"Photonics is not an incremental step forward — it is a different way to compute entirely," said Bruno Spruth, Q.ANT's new CTO. "Q.ANT has built something the industry has needed for a long time, and it is time to bring this technology into the U.S. market at scale."

The company's Native Processing Server is designed for seamless integration, connecting to standard data center racks via a PCIe interface and functioning as a co-processor alongside existing CPUs and GPUs. This "plug-and-play" approach aims to lower the barrier to adoption for an industry built on decades of x86 architecture. However, the path to widespread adoption is not without challenges. The supply chain for TFLN wafers is still nascent, and a broader ecosystem of software and developer tools is needed to fully unlock the potential of light-based computing.

Austin's New Tech Star and a Veteran's Next Act

Q.ANT’s choice of Austin for its U.S. base is a strategic move that places it at the heart of America's "Silicon Hills." The city boasts a dense ecosystem of semiconductor giants, a world-class talent pipeline from institutions like the University of Texas at Austin, and a vibrant culture of tech innovation. The company plans to hire 20 employees in software, photonics, and system design over the next six months, with plans to eventually localize chip manufacturing in the U.S.

The appointment of Bruno Spruth is perhaps the strongest signal of Q.ANT’s commercial ambitions. Spruth spent 16 years in senior leadership at IBM, culminating in his role leading the global development of the POWER processor architecture—a cornerstone of some of the world's most powerful supercomputers and mission-critical systems. His decision to leave the pinnacle of traditional high-performance computing to lead a photonic computing startup is a powerful validation of the emerging technology.

"With its combination of top-tier technical universities, a mature semiconductor ecosystem, and supportive regulatory environment, Austin is the ideal home for Q.ANT’s U.S. headquarters,” said Spruth.

This U.S. expansion is fueled by significant financial backing. Q.ANT recently closed an $80 million Series A funding round from a syndicate of high-profile investors, including the Duquesne Family Office, deep-tech visionary Hermann Hauser, and imec.xpand, the venture arm of the world-leading semiconductor research hub imec. This strong investor confidence underscores the growing belief that photonic computing is no longer a distant dream, but a critical and commercially viable technology for powering the next generation of artificial intelligence.