Aylight raises €4.5M pre-seed to build chip-scale lasers for AI data centres
What's the deal? Zürich-based deeptech startup AylightDealroom has a profile for this one. Try Dealroom → has closed a €4.5 million pre-seed round to develop chip-scale lasers that generate multiple wavelengths from a single chip. The round was co-led by Elaia and Swisscom VenturesDealroom has a profile for this one. Try Dealroom →, with participation from Verve VenturesDealroom has a profile for this one. Try Dealroom → and Plug and Play.
What's the endgame? Founded in 2025 by Bahareh MarzbanDealroom has a profile for this one. Try Dealroom → and Dmitry Kazakov following research at ETH Zürich, Aylight makes multiwavelength lasers for AI data-centre optical interconnects and high-precision frequency-modulated continuous-wave (FMCW) sensing. Its architecture, based on a frequency-modulated comb, produces precisely spaced wavelengths from one chip, cutting the need for multiple discrete lasers.
Why now? The company said demand for more efficient optical interconnects is rising as AI infrastructure scales and data transfer between chips becomes a bigger factor in data-centre performance. "We started from a problem rather than a technology: the laser had become one of the constraints on scaling AI infrastructure," said Marzban, co-founder and chief executive officer.
The funding will support Aylight's first semiconductor-foundry prototypes and expand its research and development team. The lasers are designed to be built using existing semiconductor photonics foundries.
Beyond interconnects, the technology targets semiconductor inspection, metrology, industrial automation, and precision robotics, where high-resolution 3D sensing is required.
The signal: The round is sizeable for a pre-seed and notable for its backers — Elaia and Swisscom Ventures rarely co-lead at this stage — signalling investor conviction in a capital-intensive area. It places Aylight in an emerging cohort building the data-movement technologies that AI chips and data centres increasingly depend on.
Read more: Tech.eu
Image credit: IBM Research
