Indian chip startup Bharat Semi raises a $24 million seed round, claiming it can cut vision system costs by 70 percent
A pre-revenue Indian company whose sensors are still in development just raised $24 million on the claim that it can remove two chips from the bill of materials.
Three key facts
Bharat Semi, based in New Delhi, closed a $24 million all-equity seed round led by Info Edge Group, South Park Commons and Indus Bridge Ventures, with participation from Valour Capital, Micromax, Gaingels and individual investors. The nearly five-year-old startup has no revenue yet. It says it has secured design wins with U.S. defense technology company General Atomics and India's Kaynes Semicon, while its image sensors remain under development and initial commercial production is targeted for 2027.
Technically, it builds programmable image sensors that combine image capture, processing and AI on a single chip, so visual data no longer has to be sent to a separate processor. Co-founder Vrinda Kapoor said the approach can cut system costs by at least 70 percent compared with conventional setups using a separate image sensor, image signal processor and system-on-chip. Her framing is to move from sensors to sense-making solutions, with target markets in smart homes, industrial automation, surveillance and aerospace.
The company has about 75 employees across India and the U.S., including 10 to 12 in engineering and business development in Santa Clara. It plans to grow U.S. headcount to around 20 by mid-2027 and has begun hiring engineers in Europe. Its first-generation sensors are in prototyping, while second-generation sensors with more advanced AI are expected to reach prototype stage by mid-2027 and reach customers by the end of that year.
Wang Dou's Take
Break down the 70 percent claim. A conventional vision system uses three chips: an image sensor, an image signal processor and a system-on-chip. Bharat Semi wants one chip to do all three. What gets saved is not just the cost of two chips but board space, power and the coordination costs of an entire supply chain. In hardware, a figure in the 70 percent range is plausible, because hardware costs really do stack up one chip at a time.
The real barrier is not design. It is tape-out and volume manufacturing. A sensor in prototyping and a sensor shipping millions of units reliably are separated by yield, packaging, testing and about five years. The company says production in 2027, with second-generation parts reaching customers by the end of 2027. That is a clear timeline with zero slack in it.
Then look at its resources: 75 people and $24 million. In semiconductors that is enough to do the work but not enough to be comfortable. Its investor list includes Micromax, a phone maker rather than a pure financial backer. Industrial capital usually shows up when a specific shipping channel is waiting.
One more thing. An Indian company listing a U.S. defense contractor as a design-win customer says more about where it wants to stand than the $24 million does.
Source: TechCrunch
