A Singapore-based deep tech startup referred to as SixSense has developed an AI-powered platform that helps semiconductor producers predict and detect potential chip defects on manufacturing traces in actual time.

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

Based in 2018 by engineers Akanksha Jagwani (CTO) and Avni Agarwal (CEO), SixSense goals to deal with a elementary problem in semiconductor manufacturing: changing uncooked manufacturing knowledge, from defect photographs to gear indicators, into real-time insights that assist factories stop high quality points and enhance yield.

Regardless of the sheer quantity of information generated on the fab ground, what stood out to the co-founders was a shocking lack of real-time intelligence.

Akanksha brings a deep understanding of producing, high quality management, and software program automation by way of her expertise constructing automation options for producers like Hyundai Motors and GE and led product growth at startups like Embibe. Agarwal provides technical expertise from her time at Visa, the place she constructed large-scale knowledge analytics techniques, a few of which had been later protected as commerce secrets and techniques. A talented coder with a powerful background in arithmetic, she had lengthy been curious about making use of AI to conventional industries past fintech.

Collectively, the duo evaluated sectors from aviation to automotive earlier than touchdown on semiconductors. Regardless of the semiconductor trade’s popularity for precision, inspection processes stay largely handbook and fragmented, Agarwal informed TechCrunch. After talking with greater than 50 engineers, it grew to become clear there’s important room to modernize how high quality checks are completed, she added.

Fabs at this time are full of dashboards, SPC charts, and inline inspection techniques, however most solely show knowledge with out additional evaluation, Agarwal stated. “The burden of utilizing it for decision-making nonetheless falls on engineers: [they must] spot patterns, examine anomalies, and hint root causes. That’s time-consuming, subjective, and doesn’t scale properly with growing course of complexity.”

SixSense supplies engineers with early warnings to deal with potential points earlier than they escalate with capabilities corresponding to defect detection, root trigger evaluation, and failure prediction.

SixSense’s platform can also be particularly designed for use by course of engineers relatively than knowledge scientists, Agarwal stated. “Course of engineers can fine-tune fashions utilizing their very own fab knowledge, deploy them in beneath two days, and belief the outcomes — all with out writing a single line of code. That’s what makes the platform each highly effective and sensible.”

The aggressive panorama consists of in-house engineering groups utilizing instruments like Cognex and Halcon, inspection gear makers integrating AI into their techniques, and startups together with Touchdown.ai and Robovision.

SixSense’s AI platform is already in use at main semiconductor producers like GlobalFoundries and JCET, with greater than 100 million chips processed thus far. Prospects have reported as much as 30% quicker manufacturing cycles, a 1-2% enhance in yield, and a 90% discount in handbook inspection work, the founders stated. The system is suitable with inspection gear that covers over 60% of the worldwide market.

“Our goal prospects are large-scale chipmakers — together with foundries, outsourced semiconductor meeting and take a look at suppliers (OSATs), and built-in gadget producers (IDMs),” Agarwal stated. “We’re already working with fabs in Singapore, Malaysia, Taiwan, and Israel, and at the moment are increasing into the U.S.”

Geopolitical tensions, particularly between the U.S. and China, are reshaping the place chips are made, driving new manufacturing investments throughout the globe.

“We’re seeing fabs and OSATs develop aggressively in Malaysia, Singapore, Vietnam, India, and the U.S. — and that’s a tailwind for us. Why? As a result of we’re already primarily based within the area, and plenty of of those new services are beginning recent — with out legacy techniques weighing them down. That makes them way more open to AI-native approaches like ours from day one,” Agarwal informed TechCrunch.