In The Arena by TechArena - Startup Innovation Key to Address Sustainable Infrastructure with Qarnot CEO Paul Benoit

Episode Date: April 25, 2024

TechArena host Allyson Klein chats with Qarnot CEO Paul Benoit about how VC backed startups are an essential element of vibrant industry innovation, and how the AI era has placed requirements on innovative approaches to sustainable IT.

Welcome to the Tech Arena, featuring authentic discussions between tech's leading innovators and our host, Alison Klein. Now let's step into the arena. Welcome to the Tech Arena. My name is Alison Klein. We're recording this week from OCP Lisbon, and I am really excited to be joined by Paul Beno, CEO of Carno. Welcome to the program, Paul. Thank you, Alison. So Carno is a new name for the tech arena, and I have learned a bit about the company. I'm really excited to share the story with my listeners. Why don't you go ahead and introduce yourself, your background, and why it was that you decided to found Carnot? Okay, that's a large question.

So Carnot, I founded Carnot 14 years ago. I'm an engineer. I worked in the finance industry first in a trading room where we had a very large data center and discovered this IT infrastructure. I was fascinated. I saw that they will need every year more energy, more space. I felt this trend that will last for years. And I really saw that the problem of energy, and that's why I decided to create Carno. So Carno is a cloud service provider. We are specialized in HPC, so high performance computing.

And the specificity of Carno is that we have our own hardware that focused on retrieving the waste heat of the servers. Because really, for us, why we do that is because we really think that valuing the waste heat of data centers is maybe the best way to provide more IT infrastructure in the next decade. There are many solutions to reduce the consumption of processors, but there are margins compared to the fact that we consume so much power in the data center and it generates so much heat that we are able to use heat twice. It opens up a lot of space for the next AP. So the R&D of Carno is really focused on the performance of weight retrieval. I've heard about you and have had your story shared to me by others.

But this is my first time actually talking to you. Before we get into the depth of your solution, I just want to ask you about the panel discussion that you participated in this morning. It was about the role of startups in the data center arena and how VC funding is playing a role in that space. How do you see in 2024 the startup industry accelerating to address the growing demands in data centers? That's very difficult because when you are a startup, for instance, Karna, we are 100 people. There are really blorfs in the world of data center because take place in this arena. We are really passionate with what we do and we really think that the future of data centers have to redesign themselves. And today it's the right moment because the industry is discovering

the pain of energy and resources in a context where there is a huge business but they are stuck with the problem of energy today they want to build a lot of data centers because of ai but they cannot they cannot be able faster they are limited in terms of resources. So I think this is a good pivotal moment for startups to propose new solutions that will enable giants to grow. I think we'll be growing more, I think, during the next year, because without new solutions,

it's very difficult. Peter, you talked about how you focus on high-performance computing solutions, and there's obviously a lot of parallels between HPC and AI training clusters. So you obviously made some advancements in this arena that could help others as they try to tackle the power and coin. What would you say is unique about the solutions that you've delivered and what can others learn from it? When I was talking about redesigning a data center,

it's what they have in their mindset. It could be really redesigned the way to create a rack, a thought out of the box because a rack and a computer design was, everything was designed 40 years ago, in design, but the power increased, and so it's very difficult now to design. In fact, the computer is a CPU and a motherboard, a box with fans.

Now the processor can consume up to a kilowatt. So it's very difficult to take out the pixel. And nobody thought to rethink. Everybody has habits. They don't want to change. And now the power is so huge that you have to think another way. And that's why OCP is very interesting because you talk about the design.

And you see that in the community, everybody has the same kind of problems. All the industry is understanding that you need water cooling instead of air cooling. You have a little bit about immersion. It's another kind of solution. It's not so new, but it's another kind of solution. So it's very interesting to

discuss with the community and to see that even if you are competitors, you have different solutions, you have common problems, and that the community can help you frankly. Obviously, OCP announced a major initiative around sustainability a couple years back,

and I know that there's a work group that is working on this. When you talk to customers and you talk to them about what their challenges are, what is their mindset? And how does that map with what you see in the industry? Okay, so I think it's a part of the problem. We just think that customers always look for performance.

And for planet and the governments, they look for performance. For planet and the governments, they look for efficiency. I say, for instance, maybe the most efficient car will be very slow. But people want cars that go fast. So that's really the problem today. Customer want performance. Hyperscale provide performance, but consuming a lot

of resources. And it's very difficult to change the game. And the best way, I think, to change the game is to be able to provide a solution that is both performance and efficient. It's really what we work on at Kano, keep the performance. Because the mindset of Kano is not about reducing the amount of computation we really develop this solution to provide more IT infrastructure with less resources thank you so many solutions make um virtuous solutions emerge

mindset we have at carno so it's really difficult to manage the demand of the customers the problem we have is not sustainable but what I like at the OCP is that the people here are really passionate about IT. There are no people saying because sometimes you get some people who are ecologists and to say oh, IT is bad. That's not the point. Here is number 30. It's really how to make more IT, get more money. It's not a taboo idea.

When you look at that displaced heat and you have so much heat coming out of a data center, what is the approach of Carnot on what to do with that heat in terms of secondary uses? So, at Carnot, we really design the hardware and software to be able to deploy IT infrastructure differently. Instead of building a dedicated data center, we will deploy the IT in a place that will use and buy the heat. So in fact, we create a small data center in place like district heating facilities or large swimming pools. And here you can deploy data center around 1 megawatt or 2 megawatts, not 100 megawatts like Google or

. But we also really think that the future of IT will be a mix of very large data centers and many small data centers because when you have, say, 1,000 1 megawatt data centers, you have 1 gigawatt. And 1 megawatt is not so small. It's quite large. You can handle many, many workloads this way.

A lot of customers don't move more than one minute what our particular vision we have is to deploy different many small data centers right place instead of we use the right place are the place that will be able to use it so we'll I think this pre-seating for us are large winning booths that before so in the industry, or for instance, in the agro-food industry, they need a lot of hot water. So, in fact, in places like this, there are tens of thousands, for instance, just in France. So, it's a different model for hygiene procedures. That's very cool, and it's very innovative compared to what has been done in the past.

When you are talking, you talked about the customer challenge of they want performance, getting sustainability prioritized with them may be a challenge. What do you see with the rest of the industry in terms of the approach? And are you are you seeing the right level of prioritization from the industry on this topic? What would you like to see from others in the industry to deliver innovations that might help? Yeah, I see the customer look for performance and their provider want to deliver the performance. And there are also habits they don't want to change. Habits, for instance, they want traditional data center like this, but one day they will hit the wall. So I think that both the customer and the large scale try alternative approach

to prepare the future. This way, they can see solution will be able to scale for them in the future. For instance, we talk a lot about edge computing. Some of the hyperscaling are starting, but the others are still very large. I see also large posters. They are mostly air-cooled. For them to bring water-cooled solution is very difficult, because it's difficult to integrate

very different customers and from partner in a water-cooled system. They never start. They will never get there, right. Never get there. So I think the old, I don't know, person, or person of their infrastructure, that is alternative to test the filter.

That's fantastic. Paul, I know that OCP is the highlight of the year for many in the industry, but what can we look forward to from Quarno in 2024? Yeah, today we are really working on scaling or large sites so we deployed several sites in the northern europe that is very uh in advance in terms of uh carbon footprint in 2024 we are really focused on the deploying a large site in the um we have a project in France, Italy, Switzerland, Belgium, and the UK. So I hope we can settle that in 2024.

When you talk about infrastructure, it's always a long way to deploy. But we focus on running for large customers this kind of HPC service. Thank you so much for spending time with us today. It was a real pleasure to get to know you, and I know that this is a busy week. So thanks for spending some of it on the Tech Arena. Thank you, Alison. Thanks for joining the Tech Arena. Subscribe and engage at our website, thetecharena.net.

All content is copyright by The Tech Arena.