Mosaic SoC: Pioneering Real-Time AI Chip Performance

20.01.2026 11:30, Rita Longobardi Meet Alfio Di Mauro, founder of Mosaic SoC. The engineering startup is enabling faster, more efficient AI processing for AR, VR, robotics, and autonomous systems. In March, Alfio will join nine other innovators on a business development and investor roadshow in Barcelona. Name: Alfio Di Mauro Location: Zurich Nationality: Italian Graduated from: ETH Zurich Founding team members: Alfio Di Mauro, Moritz Scherer Number of employees: 6 Money raised: CHF 150k (Venture Kick) What does your product or solution do, and what makes it unique? Mosaic SoC delivers a purpose-built perception chip and integrated software stack designed for always-on spatial intelligence. Our architecture tightly combines general-purpose compute, on-chip memory, and dedicated accelerators in a compact form factor suitable for wearables, mobile devices, and robotic platforms. This enables real-time vision and AI at a fraction of the power consumption of conventional solutions. Paired with Mosaic’s perception software, including SLAM, object detection, and neural inference, OEMs get a plug-and-play platform that reduces integration effort and accelerates time to market, making it possible to build smaller, more efficient products that continuously understand their surroundings. What trend or shift in your industry is currently creating the biggest opportunity for you? The major shift is where computation happens. Industries are moving from cloud-centric solutions to intelligence embedded in devices. Real-time perception, low latency, privacy, and energy efficiency highlight cloud limitations, creating demand for advanced edge computing. Augmented reality illustrates this trend, now overcoming long-standing power, latency, and form-factor constraints with solutions like ours. How did the idea for your startup originate? The idea grew out of our research work at ETH, where we were deeply involved in advanced system-on-chip architecture design. Through this work, we collaborated closely with companies tackling some of the most demanding real-world use cases, which consistently exposed fundamental limitations in existing compute solutions at the hardware level. Evaluating the state of the art and pushing beyond it was not optional, but necessary to make these applications viable. That experience gave us a first-principles understanding of the problems and ultimately led to a clear vision for building a company around it. Which market are you addressing, and what potential do you see for your startup in that market? We are initially targeting smart glasses, where our low-power technology fits strict size and power limits. Early design decisions will shape market leaders as the wearables market matures. This entry point lets our platform prove itself before expanding into smartphones, industrial robotics, and other adjacent markets. What impact do you want your technology to have five years from now? We want our technology to enable a new class of applications that were not feasible just a few years ago. At its core, the impact is about bringing always-on, real-time physical context to intelligent systems that today are largely confined to cloud infrastructure. By embedding perception directly into devices, these systems will be able to observe the world from a human perspective and interact meaningfully with the physical environment. In our future, you will be able to ask your glasses where you left your keys... What major challenges have you faced so far? Building a fabless semiconductor company comes with a distinct set of challenges. Many of the typical startup difficulties are amplified by the capital-intensive nature of the business and by long development and sales cycles. Success requires anticipating market demand three to four years in advance, engaging the right decision-makers within large multinational organizations, and securing design wins with a long-term perspective. What motivates you on tough days? Besides the fact that designing silicon is genuinely fun, what keeps me motivated is the scale and substance of the challenge. We are rethinking edge compute architecture at a fundamental level, which carries real risk but also the potential for lasting impact. Just as important, we have a clear and practical strategy for delivering this technology to customers without forcing them to change how they work today. When we explain both what we are building and how we plan to deploy it, the value is immediately clear, and that reinforces why the effort is worth it. Why did you decide to join the Venture Leaders Roadshow, and what are you most excited about? The roadshow is a strong opportunity for us to engage directly with potential customers. Events like this are where established companies actively look for technologies to integrate into future product generations. Given the long development cycles in our industry, it is essential for us to be present and show what their products could enable three to four years from now.