Marcus Richards
We create tensor processing unit cores for edge applications.
Essentially, we allow any chip to become AI-enabled by incorporating low-power, low-latency, silicon-proven cores for machine learning inference into their design.
Our mission is to make the world a smarter place by bringing the power of AI to any device, anywhere.
Tell us about yourself?
I am a serial entrepreneur and have been working in the high tech industry for over 15 years, in domains ranging from nuclear power to aerospace. In 2016 I started an IT company, one of the lines of business of which was semiconductors.
It was around the time machine learning was just starting to take off, and I was amazed by the potential of this technology.
We started adding machine learning capabilities to our products, and at some point realized that there was a big opportunity in spinning off a company that would focus exclusively on this domain. And so Edged.ai was born.
If you could go back in time a year or two, what piece of advice would you give yourself?
I would tell myself to focus on the product and the technology, and not get too caught up in the business side of things.
It is easy to let the day-to-day grind of running a business take over and lose sight of what is really important.
What problem does your business solve?
Our cores allow any chip to become AI-enabled, which opens up a whole range of new possibilities for both traditional chip manufacturers and up-and-coming startups.
There are so many talented people out there with bright ideas on how to use AI to make the world a better place, but they are often limited by the lack of AI-enabled hardware.
Our cores make it possible for them to bring their ideas to life without having to design their own AI hardware from scratch.
What is the inspiration behind your business?
Believe it or not, the idea for adding AI capabilities to chips actually came when I was in the countryside, watching a harvester machine working in a field. I imagined how much more productive the harvester would have been if it could “learn” and adapt to the changing conditions on the field.
That got me thinking about how AI could be used to improve all sorts of machines and devices. (Ironically, one of our first customers was actually a company that makes agricultural machines.)
I think that is illustrative of how misguided we can be when we think about “AI.” It’s not just about self-driving cars or super-intelligent robots — it’s about all sorts of things that surround us in our everyday lives.
Microwaves were first researched in the 1930s as a way to treat certain medical conditions, and now we use them to cook quick meals. I believe AI will have a similar journey, and eventually become ubiquitous in all sorts of devices that we use in our everyday lives.
What is your magic sauce?
Our cores are low-power, low-latency, and silicon-proven. This last part is very important: A lot of AI hardware startups have great ideas that work great in simulations. But when it comes to actually putting them into silicon, they run into all sorts of problems.
We have a team of experienced semiconductor engineers who have been working in the industry for many years, and we have a proven track record of being able to take our cores from concept to production.
Another common issue with specifically AI cores is the lack of documentation and software tools to support them. We have a team of software engineers who are constantly working on developing new tools and documentation to make it easy for our customers to use our cores in their designs.
What is the plan for the next 5 years? What do you want to achieve?
There are so many uncharted territories when it comes to AI that it is hard to predict exactly what will happen in the next five years. Some areas that excite me most are federated learning and training on the edge.
I believe that, however convenient the “good old Cloud” is, the future of AI is in distributed intelligence, where devices are able to learn and collaborate with each other without relying on a central server.
As for our five-year plans as a company, we want to assert ourselves as the leading provider of AI cores for edge applications.
We want to be known as the company that makes it easy for anyone to add AI capabilities to their products. I know this might sound overly ambitious, but we have a great team and I am confident that we are able to achieve our goals.
What is the biggest challenge you’ve faced so far?
The biggest challenge we’ve faced so far is the same challenge that all startups face: making people aware of our company and what we do. It is hard to stand out in such a crowded market.
Another challenge is that AI is still a very new and emerging technology, and it is constantly evolving. Keeping up with the latest trends and developments is a full-time job in itself.
Finally, there are mindset issues to consider. A lot of people are still very skeptical about AI, and there is a lot of misinformation out there. It can be hard to convince people that AI is not just a fad, and that it is here to stay.
How do people get involved/buy into your vision?
If you are a chip manufacturer or a startup working on an AI-enabled product, we would love to hear from you. You can contact me personally via LinkedIn (https://linkedin.com/in/eterentev/), and I will be happy to discuss how our cores can be integrated into your design.
This is an exciting time for both edge and AI, and I can’t wait to see what amazing new products our customers will come up with, how they will change the world, and how we can play our little but important part in that.
