AI at edge to be next battleground: SiMa.ai’s Krishna Rangasayee

While the cloud market will remain large, the embedded edge is poised to be the next gold rush, said SiMa.ai’s founder and chief executive officer Krishna Rangasayee. Founded in 2020, Maverick Capital-backed SiMa.ai is an artificial intelligence (AI) startup that focuses on developing semiconductor designs and software solutions for taking AI to the edge. “We feel like we have earned our seat as one of the key leaders in the market,” added Rangasayee.

In embedded systems, edge computing refers to processing data more locally than centralised cloud systems. As per MarketAndMarket Research firm’s data, the embedded AI market alone was valued at over $9 billion in 2023 and is expected to grow to $18 billion by 2028. Rangasayee told TechCircle that factors such as application-specific demand, security and privacy, and low total cost of ownership are driving the market's growth. 

SiMa.ai tapped into the now burgeoning market at the right time, Rangasayee believes. “If we had started the company in 2015 it would've been too early and also certainly if we had started the company now, it would’ve been too late -- I think now everybody sees it. From day one till today nothing has changed, including our strategy, market outlook, and product fit. There was enough confidence that AI at the edge was going to be the next battleground. When we started our investors believed in us, and till now we have raised $270 million which is a lot of money for the space.”

To be sure, the company raised $70 million in an extended Series B funding round in April, with the new round led by Maverick Capital, along with Point72, Jericho, and others. At the same time, the company also announced the second generation of its Machine Learning System-on-Chip (MLSoC) platform. 

MLSoC is a computing platform that helps users in running legacy code at first and then deploy these capabilities into a machine learning component in phases. The platform also consists of a software component called Palette which works similar to a cloud environment, for developers to build and deploy on multiple devices. The first generation of MLSoC was released in 2022. The second generation is slated for Q1 2025 release, which will represent ‘one software-centric platform for all edge AI’, helping customers build in domains ranging from computer vision to multi-modal generative AI, the company had then said. “One platform for edge AI — that's really the tagline that we are positioning MLSoC for,” said Rangasayee.

TSMC is the manufacturing partner for SiMa.ai, which has also built MLSoC. Not just a manufacturing partner, but TSMC is also an investor in the AI on edge company, a first for the Taiwan-based chipmaker. 

SiMa.ai in India

SiMa.ai has over 140 employees, almost equally divided between its California and Bengaluru offices. “Bengaluru is our larger R&D site, where our employees work on silicon design and software development. The talent pool here is amazing and I am constantly amazed. People here also want to join an exciting startup, which has helped us make an attractive place of work,” said Rangasayee. 

That said, frugality is a mantra Rangasayee swears by. “We are well staffed for what we're doing. I have been a big student of being frugal. I think I see too many startups rushing to hire too quickly. While we fully expect our India office to grow, we will hire on a need basis.”

However, on the customer side, SiMa.ai has grand plans. “We've done fairly well with customers in North America and Europe. And we just started scaling our sales activities in India, South Korea, and Japan.” The company plans to grow its customer size in India from 10 to over a thousand in the next four years. Globally, SiMa.ai is planning to tap into over 200 clients by the end of 2024. SiMa.ai has partnered with Taiwan-based electronics distributor WPG to reach out to more clients. Going forward, the company will be targeting sectors such as robotics, industrial automation, and drones and unmanned aircraft.

To be sure, in the AI chip business, NVIDIA leads with over 90% market share, with competitors trailer far behind. Not just chips, NVIDIA also offers software solutions to support AI projects.  For its first-quarter earnings announcement on May 22, NVIDIA reported revenue of $26.0 billion, up 18% from the previous quarter and up 262% from a year ago. So while NVIDIA bests competitors in the data centre and GPU business, SiMa.ai hopes to fill in the void of AI in the embedded edge.