Chips are the new oil

The MLSoC (machine learning system-on-chip) Modalix...

Our Generation 1 product has been in production now for 2.5 years. It would be fair to say that during this time we’re technologically the best solution out there in the Edge market. We’ve had the benefit of building something from scratch. It’s a very software centric architecture. It does two things really well, maybe better than anybody else. One is from an AI-ML ease of use and customer experience. The second is performance per watt, which really matters at the Edge. We refer to ourselves as one platform for Edge-AI. 

Modalix to us is an extension of what we’ve already done. If you look at the industry in the last 5-7 years, we’ve gone from classic computer vision to CNN-based AI models, to vision transformers. We are seeing use cases for LLMs and GenAI in the Edge. We wanted to create a platform that comprehends all these different variants. 

It is also multi-modal. We have taken on computer vision, text and audio. So both the input or the output could be audio, video or text. We are also introducing a family of devices: From a 25 Tera Ops device to a 200 Tera Ops one.

Have chips become “sexy” in the 2020s?

There was a drought of semiconductor investments for a long time. From late 2000s all the way to 2015 or 2018, very few semiconductor companies would get invested in, primarily because the expense is pretty high. The investment cycle is long. It is much easier to invest in software or SaaS companies. 

What we have now learned is that chips are the new oil. What we’ve also learnt is AI-enablement happens because of chips. Chips have become mission-critical. They are probably the front and central outlook for every government worldwide. It is amazing to see how many startups have come globally on semiconductors.  

The India growth story

I see an enormous infrastructure deployment that’s going to be happening in India. You may be in the early innings, but I think it’s not about how you start but how you finish. I think India has all the potential and the capacity to step in. The 1.4 billion population is amazing and going to work for you because they are young, and they are all about deploying and delivering technology and building India’s future.

Getting other industries to catch up

If you look at AI and how it has scaled in the last decade, we’ve done a really good job as a society in scaling AI in the cloud. We’ve also done a good job in scaling AI in consumer experience, on the phone and other appliances. What’s been left behind is the physical or industrialized world. When I say Edge, what it really comprehends is robotics, industrial automation, automotive, medical, agricultural tech, construction tech. If you dig deep into the technology that has been used in all these industries, it’s 40-50 years old. These industries haven’t really been brought forward. What I wanted to do was move these industries from classic compute to AI based compute.

Check out the complete video interview…

Have chips become sexy?

GenAI: “Nervousness” nothing new

Still 40-50 years behind