SiMa.ai secures $150 million Series C for on-device AI chips
SiMa.ai, a developer of chips and software for running artificial intelligence directly on devices, has raised a $150 million Series C at a $1.45 billion valuation. Fidelity Management & Research Company and Amplify co-led the financing, joined by Alter Venture Partners, Dell Technologies Capital and StepStone Group.
The round takes the startup's total capital raised to more than $500 million. SiMa.ai was founded in 2018 by Krishna Rangasayee, formerly chief operating officer of chipmaker Groq.
Funding backs physical AI hardware
SiMa.ai develops technology intended for robots, drones, cameras and other connected equipment that must process AI workloads locally. Its approach avoids repeatedly transmitting data to and from the cloud for inference, placing computing capability on the device itself.
The company is positioning its chips around energy efficiency, low latency and lower cost relative to Nvidia GPUs. It hopes those characteristics will help it compete for deployments in the expanding physical AI market, a category that includes humanoid robots.
A sharp increase from the prior round
The new valuation is up from the $960 million assigned after SiMa.ai's $85 million Series B in July 2025, as reported by PitchBook. The latest financing gives the company additional capital as developers of physical AI systems look for computing hardware suited to devices operating outside traditional data centres.
For enterprises assessing AI in cameras, drones, robots or similar equipment, the funding illustrates why hardware selection is becoming an operational decision as well as a model decision. Local processing can shift the balance among latency, energy use, chip cost and reliance on cloud data transfers, so teams should evaluate those trade-offs against the requirements of each deployment.
What businesses should assess
Organisations planning on-device AI deployments should define where inference must run, how quickly the device must respond and what power budget it can sustain. They should also compare the economics of local compute with the ongoing need to move data to the cloud, especially where a physical device must make decisions in real time.