SiMa.ai has launched MLSoC Modalix, a multi-modal, software-centric Edge AI chip for its embedded Edge machine learning system-on-chip (MLSoC) offering.

MLSoC Modalix is the second generation of SiMa’s MLSoC, first released in 2022, and which, unlike other accelerators that only perform the neural net function of machine learning, allows customers to run entire applications on a chip.

According to SiMa, the upgraded platform supports CNNs, transformers, LLMs, LMMs, and generative AI at the Edge, reportedly delivering more than 10x the performance per watt of alternatives. The company has not detailed the comparative offerings.

MLSoC Modalix is offered in 25, 50, 100, or 200 TOPS configurations, in multiple form factors, and has been purpose-built to deploy generative AI at the embedded Edge. The platform is fully software-compatible with the previous MLSoC generation.

SiMa is also offering new hardware components with MLSoC Modalix, including an integrated ISP model, four lanes of PCIe Gen 5, and eight Arm Cortex-A65 CPUs built on TSMC’s N6 technology.

Samples of MLSoC Modalix will be available to customers in Q4, 2024.

“SiMa.ai is leading the Edge AI market and building increasing momentum as customers around the world continue to adopt our gen 1 MLSoC. We are excited to introduce the MLSoC Modalix family to give customers even greater choice and flexibility to embrace multi-modal AI,” said Krishna Rangasayee, founder and CEO at SiMa.ai.

“With the introduction of MLSoC Modalix, SiMa.ai’s One Platform for Edge AI now provides coverage from CNNs to Gen AI and everything in between with industry-leading performance and power efficiency.”

Founded in 2018 by former Groq COO, Krishna Rangasayee, the San Jose, California-based company targets the market segment between 5W and 25W of energy usage.

In April 2024, SiMa.ai closed a $70 million funding round, bringing the total raised by the company to date to $270m.