dots dots dots arrow THE ROUNDTABLE

ML will impact every piece of software we know – will make this a reality

May 12, 2020 · 2  min
SiMa Logo

Machine Learning has the potential to impact every piece of software we know – will make this a reality

Today we are announcing that Dell Technologies Capital is leading the Series A financing in, with seed investors Amplify and Wing also participating. I will be joining the board.

It is rare in my world that you run across the perfect team, solving a problem in exactly the right way, at just the right time. In a world filled with ML chip companies, is building their MLSOC with a software-first mentality to seamlessly support customer’s existing system software, while delivering 30x improvement in key figures of merit like frames per second per watt (FPS/W) as well as overall TOPS and TOPS/W performance.

CEO and co-founder Krishna Rangasayee helped Xilinx start and build their embedded system business to over $3B in the last 12 years and is a respected leader that knows this space as well as anyone on the planet.

We (DTC) made our first machine learning investment nearly six years ago, in 2014. I would love to claim we were incredibly insightful. The reality is we saw enough amazing founders talking about using newly available “big data” and powerful distributed processing engines to disrupt the world, that it became obvious machine learning was a trend.

We now have about a dozen ML/AI (machine learning / artificial intelligence) companies in our portfolio, ranging from advanced ML processing semiconductor chips to AI application builders to ML data governance to ML enabled infrastructure software to ML automation. Our focus continues to be on investing in companies with the capacity to cause real disruption. fits that bill by delivering breakthrough ML performance for existing embedded systems software while meeting stringent power constraints.

In the six years since our first ML investment, the proliferation of machine learning is obvious if not pervasive. Machine learning drives a meaningful fraction of the compute workloads in data centers, ranging from training of models and networks to inference. And all of us are likely experiencing first hand the use of machine learning in our handheld and home devices, where I have basically been trained not to think. “Alexa, how many tablespoons in a quarter cup?” or “Siri, what time is it in Pune right now?”

We fundamentally believe that this trend will continue for the next decade or more, such that every software application will evolve (or be replaced) to leverage machine learning to be better … better efficiency, better accuracy, better predictability, better optimization, faster, smarter, safer, etc.

One still emerging area for machine learning adoption is embedded system software. This predominantly purpose-built compute is typically in power-constrained and/or form factor-constrained environments, with a wide range of use cases and functionality. It is a $50B+ market in aggregate, and the use of ML will be truly disruptive.

These markets are typically long lifetime, where devices and software are deployed and expected to run for 5-10 years or longer. Supporting existing software stacks and working in existing hardware platforms is required. Serving these markets takes a different approach, with a focus on supporting customer needs over the lifetime of the system.

Embedded systems for surveillance, security, autonomous vehicles, industrial robotics, communications, smart cities, medical imaging, and more are on the cusp of being disrupted by ML functionality. has the right approach, the right technology, and the right leadership to serve this market. We are excited to join their journey! And we are thrilled to work with other early stage investors Mike Dauber and Jake Flomenberg and industry titan Moshe Gavrielov to help build a great company.