Edge AI chip startup SiMa.ai has raised a $70 million extension to its Series B round, as reported by TechCrunch, to develop a new generation of chip designed specifically for multimodal generative AI workloads. As Jagmeet Singh outlines, SiMa.ai aims to capitalize on the growing market for AI processing at the network edge rather than centralized clouds.
Founded in 2017, SiMa.ai currently offers an AI system-on-chip (SoC) and no-code software platform that has seen adoption by over 50 companies. The startup’s first-gen chip achieved best-in-class performance on the MLPerf benchmarks for edge and power-constrained inference. However, as demand increases for more advanced generative AI capabilities, SiMa.ai is now developing a second-gen chip optimized for multimodal inputs like audio, video and text.
CEO and founder Krishna Rangasayee notes the new “GenAI” SoC will leverage the same software-centric architecture and compatibility features as the first chip. However, it will introduce “architectural tunings” to specifically handle emerging workloads involving multiple data types and generative models like GPT-3. Rangasayee views this as the right strategic “vector” given AI’s direction.
The new SoC is slated for Q1 2025 using TSMC’s 6nm process and Synopsys vision processors. It will aim to deliver higher performance and efficiency than competitors for multimodal edge AI scenarios. Rangasayee cites edge leader Nvidia as the primary competition but argues its solutions lack adequate power optimization and software support for embedded applications.
Among other AI chip startups, Rangasayee acknowledges Hailo’s accomplishments but asserts SiMa.ai offers better performance and efficiency through its holistic system approach. With $270M raised to date, the fresh funds will support GenAI chip development and global commercial expansion.
This new funding round was led by investment firms Maverick Capital and Point72, demonstrating continued faith in SiMa.ai’s strategy and ability to capture the sizable and growing edge AI chip market. As Andrew Homan of Maverick Capital stated, SiMa.ai is well positioned to drive transformation at the network edge just as data centers have evolved to enable generative AI capabilities.
By introducing a purpose-built multimodal chip, SiMa.ai aims to strengthen its leadership delivering high-performance and power-efficient AI processing solutions where data is created. If successful, this new product could allow the startup to maintain differentiation and better serve customers across industries adopting advanced AI technologies. SiMa.ai’s continued momentum reflects edge AI’s promising outlook as a major computing paradigm of the future.