A Singapore-based deep tech startup called SixSense has developed an AI-powered platform that helps semiconductor manufacturers predict and detect potential chip defects on production lines in real time.

It has raised $8.5 million in Series A bringing its total funding to around $12 million. The round was led by Peak XV’s Surge (formerly Sequoia India & SEA), with participation from Alpha Intelligence Capital, FEBE, and others.

Founded in 2018 by engineers Akanksha Jagwani (CTO) and Avni Agarwal (CEO), SixSense aims to address a fundamental challenge in semiconductor manufacturing: converting raw production data, from defect images to equipment signals, into real-time insights that help factories prevent quality issues and improve yield.

Despite the sheer volume of data generated on the fab floor, what stood out to the co-founders was a surprising lack of real-time intelligence.

Akanksha brings a deep understanding of manufacturing, quality control, and software automation through her experience building automation solutions for manufacturers like Hyundai Motors and GE and led product development at startups like Embibe. Agarwal adds technical experience from her time at Visa, where she built large-scale data analytics systems, some of which were later protected as trade secrets. A skilled coder with a strong background in mathematics, she had long been interested in applying AI to traditional industries beyond fintech.

Together, the duo evaluated sectors from aviation to automotive before landing on semiconductors. Despite the semiconductor industry’s reputation for precision, inspection processes remain largely manual and fragmented, Agarwal told TechCrunch. After speaking with more than 50 engineers, it became clear there’s significant room to modernize how quality checks are done, she added.

Fabs today are filled with dashboards, SPC charts, and inline inspection systems, but most only display data without further analysis, Agarwal said. “The burden of using it for decision-making still falls on engineers: [they must] spot patterns, investigate anomalies, and trace root causes. That’s time-consuming, subjective, and doesn’t scale well with increasing process complexity.”

SixSense provides engineers with early warnings to address potential issues before they escalate with capabilities such as defect detection, root cause analysis, and failure prediction.

SixSense’s platform is also specifically designed to be used by process engineers rather than data scientists, Agarwal said. “Process engineers can fine-tune models using their own fab data, deploy them in under two days, and trust the results — all without writing a single line of code. That’s what makes the platform both powerful and practical.”

The competitive landscape includes in-house engineering teams using tools like Cognex and Halcon, inspection equipment makers integrating AI into their systems, and startups including Landing.ai and Robovision.

SixSense’s AI platform is already in use at major semiconductor manufacturers like GlobalFoundries and JCET, with more than 100 million chips processed to date. Customers have reported up to 30% faster production cycles, a 1-2% boost in yield, and a 90% reduction in manual inspection work, the founders said. The system is compatible with inspection equipment that covers over 60% of the global market.

“Our target customers are large-scale chipmakers — including foundries, outsourced semiconductor assembly and test providers (OSATs), and integrated device manufacturers (IDMs),” Agarwal said. “We’re already working with fabs in Singapore, Malaysia, Taiwan, and Israel, and are now expanding into the U.S.”

Geopolitical tensions, especially between the U.S. and China, are reshaping where chips are made, driving new manufacturing investments across the globe.

“We’re seeing fabs and OSATs expand aggressively in Malaysia, Singapore, Vietnam, India, and the U.S. — and that’s a tailwind for us. Why? Because we’re already based in the region, and many of these new facilities are starting fresh — without legacy systems weighing them down. That makes them far more open to AI-native approaches like ours from day one,” Agarwal told TechCrunch.