Introduction
AI is leaving the data centre. The next generation of intelligent devices (sensors, wearables, hearing aids, autonomous robots) must think on their own, on a tiny battery. Synthara, a Zurich-based spin-out from ETH Zurich and the University of Zurich's Institute of Neuroinformatics, builds the chip technology that makes this possible.
How it works?
In conventional chips, data travels constantly between memory and processor. That back-and-forth, known as the memory bottleneck, costs both time and energy. Synthara's ComputeRAM removes it by performing the computation directly inside the memory, bringing server-class AI performance to devices small enough to fit in your ear.
Built to scale through licensing
Synthara doesn't manufacture chips. It licenses its architecture to chipmakers, which build it into their own products. This capital-light model lets one technology reach millions of devices, and it is already being applied with industry players such as Bosch.
Why we funded?
Three things convinced us. First, the team: CEO Manu V Nair and CTO Alessandro Aimar are joined by Chairman Sean Mitchell, co-founder of Movidius, the edge-AI chip company Intel acquired in 2016. Second, the timing: as AI moves to the edge, energy efficiency becomes the limiting factor, and that is exactly the problem Synthara solves. Third, the backing: its 2024 round was led by Vsquared Ventures alongside High-Tech Gründerfonds, complemented by EU and Swiss grants. Synthara is European deep tech in the semiconductor space, one of the areas where Europe most needs its own capabilities.
