Bharat Semi Raises $24M Seed for Single-Chip Programmable AI Image Sensors
New Delhi fabless chip startup Bharat Semi raised $24 million in seed funding and won design contracts with General Atomics and Kaynes Semicon.

By Marcus Vance, Semiconductors & Hardware
3 min read
New Delhi semiconductor startup Bharat Semi has raised $24 million in an all-equity seed round to bring programmable AI image sensors to commercial production, backed by early design wins with US defence contractor General Atomics and Indian packaging firm Kaynes Semicon.
Info Edge Group, South Park Commons, and Indus Bridge Ventures led the round, with participation from Valour Capital, Micromax, and Gaingels, according to TechCrunch.
Collapsing three chips into one sensor
Conventional vision hardware relies on three separate components: an image sensor to capture raw light, an image signal processor to clean the frame, and a system-on-chip to run computer vision models. Moving raw video across those chips consumes power and drives up bill-of-materials costs.
Bharat Semi designs fabless silicon that merges capture, processing, and programmable AI inference onto a single die. Chief executive Vrinda Kapoor told TechCrunch that integrating those stages cuts total system costs by at least 70% compared with multi-chip vision pipelines.
"Our aim here is to move from sensors to sense-making solutions," Kapoor said.
Leadership and 2027 production roadmap
Founded in late 2021 by Kapoor and managing director Vinayak Dalmia, the company recruited former OmniVision chief technology officer Boyd Fowler as its CTO. Fowler brings deep image-sensor architecture experience to a market where power budgets are tight.
First-generation sensors are currently in prototyping. A second-generation design with broader on-chip AI capabilities is scheduled to reach prototype stage by mid-2027, with initial customer shipments targeted before the end of 2027.
Defence, industrial, and appliance targets
Despite remaining pre-revenue, Bharat Semi has secured design wins with General Atomics and Kaynes Semicon, alongside interest across smart home appliances, factory automation, surveillance, and drones.
For edge hardware makers, on-sensor inference avoids streaming sensitive video to external processors or cloud servers. It also positions an Indian fabless designer to compete for US and Asia-Pacific supply chain diversification contracts as governments push for alternatives to Chinese sensor suppliers.
Sources
- 1TechCrunch reporttechcrunch.com