Netrasemi, a Zoho-backed startup designing chips for surveillance, robotics and drones, has reduced script development time to two-three hours from a week in the post-design stage.
Mindgrove Technologies is using AI to parse lengthy technical specifications running hundreds of pages, cutting down time spent on documentation, a critical element in chip designing business.
For a sector which has a very long product incubation period and is cost-intensive, innovations around AI and its wide-scale adoption have become a boon for frugal startups. Industry experts said AI can help early-stage startups like Maieutic Semiconductors cut overall design cycles by three-four months.
Chip designing entails a long gestation period, requiring engineers to work for hours to build the product, test it across multiple iterations and then push it into production.
While the likes of Google and Qualcomm can afford such processes, early-stage startups with a fraction of that budget found the entry barriers daunting. But with AI and advanced computing, the cost and time barriers are going down.
Agnit Semiconductors, which makes gallium nitride chips for fast-charging applications, has found that AI not only speeds up chip design processes but could also reduce team sizes. AI could potentially reduce team sizes to 70 designers from 100, according to the company. Maieutic, a fabless semiconductor startup, has saved 20-25% of its time using AI.
AI to the rescue
“AI has become a tool for the architect or engineer to get direction in building designs,” said Jyothis Indirabhai, founder of Netrasemi, which has raised nearly $15 million in funding.
“With AI we can develop a script in two to three hours compared to a week earlier, especially in the post-design stage,” Indirabhai said.
Post-design is the phase where the chip goes through intensive testing. Engineers use scripting languages such as python and Tcl to run diagnostic tests and capture data from the physical chip over multiple iterations.
However, intellectual property (IP) concerns such as losing software and design specifications to AI models and limited data availability continue to be major challenges.
Faster development, smaller teams
Netrasemi has used AI since last year to automate productivity scripts, code that handles repetitive tasks throughout the chip design process.
The technology is also reducing workforce needs. “A lot of the design software we use already integrates AI,” said Hareesh Chandrasekar, CEO of 3one4-backed Agnit Semiconductors. “If you had 100 designers on a project, AI could bring that down to maybe 70.”
Major design software providers such as Cadence and Synopsys have embedded AI and machine learning into their platforms, which many startups are using extensively to speed up the process.
It comes at a time when there is a broader policy push from the Indian government towards building more semiconductor startups in the country. The government launched the India Semiconductor Mission in December 2021 with a budget of Rs 76,000 crore. In July, the government approved design-linked incentives for 23 chip companies, including startups like Peak XV Partners-backed Incore Semiconductors.
Beyond faster design cycles, startups are using AI to read lengthy technical specifications, documents that can run hundreds of pages and contain detailed instructions for building specific chip components.
“With AI we can now read instruction documents with details of everything from currents to voltages,” said Shashwat T R, founder of Mindgrove Technologies, a government incentive recipient.
Documentation is a critical element of chip designing where AI is playing a pivotal role. Industry experts said that leading firms globally are actively using AI and that it helps improve their efficiency. But for startups it is also about cost and giving founders a realistic shot to be competitive with such large players.
Maieutic Semiconductors founder Ashish Lachhwani said, “Easily we can tackle a good 20-25% of the design time and then in the future look at getting to 50% as well.”
Lachhwani is a second time founder, whose first semiconductor startup, Steradian, was acquired by Japanese multinational Renesas in 2022.
Maieutic Semiconductors has raised nearly $6 million since its inception in the beginning of 2025.
Adoption barriers
Despite these gains, high manufacturing costs limit experimentation. “Chip business is expensive. Manufacturing at a foundry costs around $3 million on average, and reaching production quality takes at least $20 million,” said Indirabhai, who uses Taiwan-based TSMC’s facilities.
The lack of data availability compounds the challenge. “For chip design, there isn’t as much data available,” Shashwat said.
Security concerns also persist, as engineers risk exposing proprietary code when inputting it into AI systems, compromising IPs, said Indirabhai.