This French DeepTech Startup Just Unlocked a Faster Path to AI Hardware

Jan de Vries · 2026-09-22

Here's a problem you might not think about every day, but it's absolutely crucial for the AI revolution: getting light into and out of a photonic chip is ridiculously expensive and inefficient. In fact, the packaging—the process of connecting fibers and components—chews up about half the total cost. That's a massive bottleneck. But a French startup called Lightspring just raised a significant war chest to change that. They've secured over $3.8 million in total funding. A $3.4 million Seed round, led by Atlantic and OVNI Capital with participation from Concept Ventures, Ixcore, Plug and Play Ventures and angels, plus a $380,000 grant from winning France's i-Lab DeepTech competition. Their goal? To move a breakthrough 3D optical coupling process from the research lab to the factory floor. ### Why Packaging is the AI Hardware Achilles' Heel Think about the race for faster AI. It's not just about software or raw computing power anymore. It's about the physical hardware that data screams through. Photonic chips, which use light instead of electricity, are becoming essential. But there's a catch. “Photonics is scaling into AI infrastructure fast, but optical packaging remains a slow and expensive step,” explains Quentin Calleja, Principal at Atlantic. “Lightspring has developed a technology that finally makes it scalable.” The issue is the current method. Aligning hair-thin optical fibers to a chip requires micron-level precision, done manually or with complex machines. It's slow, it loses a lot of light signal (2-3 decibels worth), and it simply doesn't scale. It's like trying to build a highway system one tiny, perfectly placed pebble at a time. ### The Lightspring Solution: 3D Printing the Interface Founded in 2026 in Besançon by Grégoire Bonnat, Adrià Grabulosa and Daniel Brunner, Lightspring took a radically different approach. Instead of assembling parts, they build the optical interface directly onto the chip itself. They use a form of 3D lithography, a bit like 3D printing at a microscopic scale. This single step creates the lenses and waveguides needed to channel light. Here's what that means in practice: - **Massive Cost Reduction:** Packaging goes from being 50% of the chip's cost to just another manufacturing step. - **Near-Zero Signal Loss:** They're targeting less than 0.5 dB of loss per connection, a huge leap from the current standard. - **Speed and Flexibility:** Design cycles shrink from several months to under two weeks, and the process works with different materials and equipment. - **Eliminates Manual Alignment:** The need for finicky, active alignment is gone. It's reproducible every single time. “The whole industry is working on making photonic chips faster, yet light still loses most of its energy the moment it enters or leaves the device,” says CEO Grégoire Bonnat. “We decided to attack that specific problem: to build the optical interface the same way the rest of the chip is built, in one reproducible process step.” ### The Bigger Picture for European Tech and AI Infrastructure This isn't just a niche engineering win. It's about enabling the next generation of AI infrastructure. As optics move closer to the compute core to reduce latency and boost speed, solving the packaging puzzle becomes non-negotiable. The funding will help Lightspring qualify their process as an industrial building block. Their core IP is a Process Design Kit (PDK) for 3D optical coupling—essentially the rulebook that lets other companies design chips meant for their packaging method. The tech springs from EU-funded research at the renowned FEMTO-ST Institute, showing how deep-tech research can translate into market-ready solutions. For professionals watching the European startup scene, Lightspring represents a key trend: deep-tech ventures tackling foundational hardware problems. They're not just making an app; they're reshaping the physical layer that future technologies will depend on. It's a hard path, but as this funding shows, it's a path where focused innovation can attract serious backing and potentially redefine an entire category.