Optical inspection is seeing the emergence of platforms natively designed around artificial intelligence, thus Delvitech, a Swiss deeptech specializing in 3D AOI, is developing a system capable of analyzing complex electronic boards, including difficult-to-inspect components such as transparent adhesives or plastic parts. The objective is to improve visibility across all assemblies and integrate predictive mechanisms intended to anticipate errors.
The diversification of materials, miniaturization and increased production rates are creating increased demand for this type of solution. Delvitech’s self-learning approach seeks to reduce reliance on manual setup and deliver faster inspection in industries such as automotive, telecom, EMS, IoT, pharmaceutical, defense or industrial automation.
Roberto Gatti, founder and CEO of Delvitech, says “We have developed a future-proof neural network architecture for a native AI platform. Our flexible and agnostic system is applicable to multiple markets and industries. With our proprietary predictive technology, we are moving from error detection to error prevention in optical inspection. »
A market in transition
The 3D optical inspection market remains dominated by historical players such as Koh Young, Omron, Viscom, Mirtec or CyberOptics, whose solutions are mainly based on rule-based models. The arrival of AI-native systems is shaking up the cards in a sector where precision, adaptability and reduced commissioning times are becoming essential. The integration of predictive capabilities represents a notable departure from traditional methodologies.
Fundraising to address the semiconductor market
Delvitech has just raised 40 million dollars, or 34.5 million euros, from EGS Beteiligungen AG, Creadd Ventures and European private investors. Founded by Roberto Gatti, the company is preparing an expansion into the semiconductor sector. This financing will support the increase in industrial capacities, with a factory in India planned for 2026 and a target of more than 200 machines produced per month in 2030.
