Bengaluru: Semiconductor start-up SandLogic is in early talks to raise $30-40 million in Series A funding at a valuation of $200 million, two people aware of the matter said.
Bengaluru-based SandLogic is among India’s few chip builders targeting commercial-grade artificial intelligence (AI) applications. The company has already appointed an investment banker, and domestic and foreign venture capitalists have been approached, one of the two people said on the condition of anonymity.
Founded in 2018 by Kamalakar Devaki, Jesudas Fernandes, Radhika Kanigiri and Ravi Kumar Rayana, the Texas- and Bengaluru-based company started as a low-code/no-code platform, before entering enterprise AI and designing its own chip prototypes. Its low-power Krsna chip aims to facilitate on-device AI and has already achieved 22 trillion of operations per second (TOPS).
Devaki declined to comment on the funding round.
AI chip start-ups have seen increasing investor interest this year. While Kerala-based Netrasemi raised about ₹107 crore in Series A funding led by Zoho and Unicorn India Ventures, Bengaluru’s Maieutic Semiconductors secured roughly $4.15 million seed money from Endiya Partners and Exfinity. SandLogic has previously raised around $3.5 million from high-net-worth individuals and a few angel investors, including Google and AT&T executives.
ExSLerate V2 is SandLogic’s underlying chip design that can be reused and scaled. It is meant to run AI tasks faster while using less than 2W of power, and the company says it will be ready for commercial use by December 2026. Krsna chip is built on that design and is now in prototyping, with test chips planned for the third quarter of FY27, Devaki said. It is aimed at AI applications like image, language and audio processing, and is designed for use in small devices without heating up.
“These chips can give more battery life for small devices and then still run AI algorithms on those devices such as smart watches, infotainment devices in the cars, routers at home, TVs, and phones,” Devaki said.
Semiconductor bets rise
Investor interest spans deep-tech specialists such as Speciale Invest, which has closed a ₹600 crore fund to back early-stage deep-tech start-ups including AI and semiconductors, and global investors like Celesta Capital that have been active across semiconductor and AI infrastructure bets.
Other investors include venture capital (VC) funds like Endiya Partners, Unicorn India Ventures and 3one4 Capital that are actively writing seed-to-Series A cheques in AI-led start-ups. Among foreign VCs, Celesta Capital remains active across semiconductor and AI infrastructure, while larger multi-stage VCs such as Peak XV Partners, Accel, Lightspeed and Elevation Capital have been leading or participating in new AI and agentic AI rounds this year.
Shakti is SandLogic’s in-house large language model (LLM) built for enterprise use, with versions released till 4 billion parameters and 8 billion parameters under progress, according to Devaki. In simple terms, parameters are the dials inside an AI model that it learns to tune; more parameters generally allow the model to handle more complex tasks.
SandLogic has been selling an “agent AI” layer on top of these models to businesses. “We started giving it to banks and healthcare companies and manufacturing companies,” Devaki said.
