Lightspring raises €3.1M Seed to industrialise 3D photonic chip packaging with two-photon polymerisation

Lightspring (Besançon) has raised €3.1 million in a Seed round from Atlantic and OVNI Capital, with Concept Ventures, Ixcore, and Plug and Play Ventures also participating. The round is complemented by a €350,000 i-Lab grant, bringing total disclosed financing above €3.5 million. Founded in 2026 by Grégoire Bonnat, Adrià Grabulosa, and Daniel Brunner — who brings a background from the CNRS — the company is applying two-photon polymerisation to the packaging problem that has long constrained photonic chip manufacturing at scale.

The packaging bottleneck in photonics

Photonic integrated circuits (PICs) — chips that use light rather than electrons to move data — offer compelling advantages for high-bandwidth, low-power optical interconnects. Silicon photonics has been positioned as the solution to data centre interconnect density for over a decade. The physics has never been the limiting factor; the manufacturing has. Specifically: the packaging step, where optical fibres must be aligned to waveguides on the chip surface with tolerances measured in hundreds of nanometres. Manual alignment is slow, expensive, and fundamentally incompatible with the throughput that semiconductor manufacturing economics demand.

Two-photon polymerisation (2PP) is a laser-based 3D writing technique that can build free-form optical structures — lenses, waveguides, coupling elements — directly on the chip surface with nanoscale precision. Rather than aligning a pre-fabricated fibre to a waveguide opening, the process builds a custom coupling structure in place, removing the alignment tolerance as a constraint. The question Lightspring is attempting to answer is whether 2PP can be made fast enough, reliable enough, and cost-effective enough to function as an industrial packaging process rather than a laboratory demonstration.

Why now

The demand side has changed. AI data centres are driving a step-change in optical interconnect requirements: the bandwidth density of copper-based electrical interconnects is reaching physical limits at the speeds and distances that large-scale AI inference infrastructure requires. Photonic chips that would have competed with copper on price alone — losing — are now competing on capability, and the comparison is increasingly favourable. That shift is creating procurement interest that justifies investing in packaging technology that previously looked like pre-commercial research.

The i-Lab grant — awarded by Bpifrance and evaluated against scientific and technological criteria — provides independent validation that the technology has cleared a peer-review equivalent for public research funding. Daniel Brunner's CNRS background connects the company to the research infrastructure where 2PP for photonics has been developed over the past decade, and where Lightspring's IP presumably originates.

The 18-month question is whether laboratory-grade 2PP throughput can be scaled to production-relevant cycle times and yields without sacrificing the precision that makes the technique valuable. That transition has defeated earlier deep-tech packaging approaches; whether 2PP's fundamental advantages survive industrialisation is the technical thesis that the Seed must prove.