ETH Zurich spinout Mosaic SoC raises $3.8 million from Fonderful to build AI recognition chip for AR glasses — TFN

AI News


Zurich-based semiconductor startup Mosaic SoC has raised $3.8 million in pre-seed funding led by Founderful with participation from the Kick Foundation.

The company is building the next generation of perception chips that allow devices to see, track, and interpret their surroundings in real time while consuming little power. Its technology is aimed at wearable and mobile products such as AR glasses and smartphone cameras, which are always limited by battery life, heat generation, and size.

Smart devices are packed with cameras and sensors, but many still struggle to truly understand the world around them. Next-generation devices are expected to do more than just capture photos and videos. It needs to understand space, recognize objects, map its surroundings, and react instantly to user actions.

Mosaic SoC was founded by Moritz Scherer and Alfio Di Mauro, PhD students from ETH Zurich. They recognized the growing gap between the demand for on-device intelligence and the limitations of existing hardware.

Instead of relying on power-hungry general-purpose processors, we designed a purpose-built recognition chip to serve as a baseline layer of spatial intelligence. Hardware manufacturers can integrate the chip and build additional applications on top of it.

The company says its integrated circuits process visual and location sensor data, allowing devices to understand their location and what’s around them. Internally, this process is described as converting space into signals. This feature could unlock new experiences for smart glasses. The device can create a live map of the room, identify objects, and even remember where the user last placed an item. You can also generate floor plans in real time.

In smartphones, this chip acts as a coprocessor connected to the front camera. Continuously track scenes and classify objects using only a fraction of the power of traditional processors. This means your phone can only start recording when certain people, gestures, or events appear, instead of running the camera at full load all the time.

Its commercial model is to sell chips. But the company believes that by shipping a complete software application layer with the silicon, it can reduce the engineering burden for manufacturers and allow partners to integrate recognition systems more quickly than building them from scratch.

In its first year, the business is already generating revenue through non-recurring engineering contracts with manufacturing partners. The company expects revenue from large-scale chip sales to increase further as commercial products launch.

Mosaic SoC says its main technological advantage lies in its architecture. While many competing solutions such as Axelera AI, Hailo, and Syntiant use one or two ARM-based cores, its unique design uses eight or more cores to maximize performance per watt and deliver practical always-on awareness in compact devices.

In addition to hardware, the company also develops deployment tools and compilers to help firmware developers optimize applications for its chips. Its long-term goals are clear. It’s about becoming the standard intelligence layer for wearables and mobile devices, enabling products that are always recognizable without sacrificing battery life, comfort, or design.

The new capital will be used to accelerate chip development, expand the engineering team, enhance software tools, and prepare the first products for commercial launch with hardware partners.





Source link