As the AI boom pushes silicon chips and data centres to their limits, Stuttgart-based Q.ANT is changing the game by developing light-powered processors that significantly reduce energy costs while dramatically increasing computing power.
Their €62 million Series A funding round was co-led by Cherry Ventures, UVC Partners, and imec.xpand, with participation from additional deep-tech investors, including L-Bank, Verve Ventures, Grazia Equity, EXF Alpha, LEA Partners, Onsight Ventures, and TRUMPF. This represents the biggest investment in European photonic computing to date, signalling investor belief that the next major computing breakthrough may come through analogue and photonic technology rather than electrical and digital approaches.
While the company keeps its precise valuation under wraps, it notes: “Our €62 million Series A ranks among Europe’s largest deep tech rounds for photonic computing, and reflects investor confidence in the market need for native, light-based architectures.”
The rapid rise of AI and high-performance computing (HPC) has placed immense pressure on data centres. Energy use is nearing critical levels: “AI data centres are projected to consume up to 17% of US electricity by 2030, and operational costs are rising as a single GPU now consumes as much power as a kitchen oven. Data centre energy consumption is expected to increase by 160% this decade,” explains Q.ANT to TFN.
However, the issue extends beyond energy consumption. “Traditional chips aren’t designed for the complexity of modern AI. An 8-bit multiplication on a CMOS processor requires 1,200 transistors; on Q.ANT’s photonic chip, it’s just one optical element, providing up to 30 times higher energy efficiency. Even common operations like a Fourier transform, which use over a million transistors on silicon, are performed in light with a single component.”
Behind Q.ANT: Redefining data centre hardware
To address this bottleneck, Q.ANT has developed the world’s first commercial photonic processor, built on thin-film lithium niobate (TFLN). Instead of shuffling electrons, these chips compute directly with light, offering radical new efficiency at speeds that traditional architectures simply can’t match.
“Q.ANT delivers photonic processors based on thin-film lithium niobate, allowing direct analogue computation with light. This approach skips the power-hungry translation to digital bits, solving non-linear equations natively, critical for AI,” the team shares. “What we deliver today is a turnkey system that integrates seamlessly into a data centre. This is unique in the photonic computing industry. Nobody else has that.”
Q.ANT’s Native Processing Server is a plug-and-play system, compatible with today’s software stacks and programming languages, including C++ and Python. Even the installation feels comfortingly familiar to seasoned data centre managers: “Our servers come without water cooling and installation costs, you don’t need to consider special load capacity of your datacenter floor: it’s a bit like installing servers in the past, just 4 screws and 2 cables.”
For developers, the goal is effortless adoption. “We make it very easy to use so that you can program it with one click. Developers want to run their code without worrying about the hardware. Together with our partners, we will ensure that the optimal processor is automatically selected for the task at hand and the developer gets their response function back promptly.”
Vision born in research, forged in industry
Q.ANT’s leap is the result of years of foresight and “hands-on industry experience, as well as our extensive work with Thin Film Lithium Niobate, a material optimal for computing with light.” The company was founded in 2018 by Dr. Michael Förtsch, a physicist and alumnus of the Max Planck Institute, who spun the team out of the German manufacturing giant TRUMPF.
“We have been developing photonic chip technology since 2018. Since 2022, we have specifically been building AI accelerators because, even before AI became widely used, we foresaw an impending energy and performance problem for when AI would enter the mainstream market,” says Förtsch.
Q.ANT’s 100+ employees hail from over 19 nationalities, with ages ranging from 19 to 71. Over 30% of the workforce are women, including in crucial roles like engineering and leadership, a diversity rarely seen in deep-tech startups.
Making the impossible possible
While several corporations worldwide are pursuing photonic computing, notable names include Lightmatter (US), Celestial AI (US), Arago (France), and Oriole (Israel). Most of these companies are focused on integrating optics for interconnects or remain at the demonstrator stage.
“Competitors in the international context used a different material, and after some time, they had to look for other applications because their processors did not work with the quality that the market would accept. They did not reach the precision. Before Q.ANT, 5-bit precision was the maximum. Since we have been around, 16-bit is no problem. We also have the prospect of 32-bit precision,” the company says.
“Unlike most ‘photonic’ or quantum competitors still in R&D, Q.ANT’s first-generation processors are easy to integrate into today’s data centres,” thanks in part to their manufacturing on a repurposed 90nm CMOS line in Germany. This innovative approach proves scalability and underscores European chip sovereignty.
Physics simulations, complex image recognition, and previously cost- or time-prohibitive AI applications get a new lease of life: “Software users realise that they can run certain applications faster than before. Applications that were once too complex to program or took too long to calculate are now feasible. Our processor can be used to create training sets that were neither economically viable nor time-relevant in the past.”
For the customer, the value is as practical as it is profound: “The data centre operator has lower electricity costs – that’s an economic added value… Software users realise that they can run certain applications faster than what used to be too complex to program or took too long to calculate.”
What’s next for Q.ANT?
With the product in the hands of early adopters, Q.ANT’s sights are firmly set on rapid market integration. “Over the next five to seven years, Q.ANT’s focus is to establish photonic computing as a new foundation for AI and high-performance computing. Our roadmap is ambitious.”
The company plans to push the boundaries of its Native Processing Server, improving precision, speed, and density, and to replicate its pilot line at new foundries globally. “We are seamlessly integrated into the existing AI software stack and support programming languages like C++ and Python; we are confident to win [developers] over.”
Ultimately, Q.ANT envisions itself as a chip company, not a manufacturer, aiming to license its photonic technology as the market matures: “By the end of this decade, our aim is for photonic co-processors to be standard in major data centres, working alongside CPUs and GPUs to deliver the performance and efficiency today’s workloads demand.”