From Arkane Cloud’s genesis to AIaaS

During the month of June, our CEO, Julien Gauthier, was invited in Paris by Megaport to meet Michael Reid, the CEO of Megaport. They discussed innovation, digital infrastructure, AI and GPU as a service, as well as how Arkane Cloud came to be. From Arkane Cloud’s genesis to AIaaS, here’s an extract from the exchange:

Michael Reid: “Welcome to the Arkane Cloud’s CEO and founder Julien Gauthier. Could you just quickly explain how did you start this company? What’s happening? Give us some perspectives, I’m really curious on how things are playing out from this space, what’s your service, what’s growing, what’s happening that you didn’t expect? So let’s just chat through it so give me a little bit about your company?”

Julien Gauthier: “Everything began in 2020 during the pandemic, I decided to build few servers for customers, I was reselling on Amazon hardware components because I’m passionate in technology in general.“

Michael Reid: “Did you sign some compute in 2020 during covid to some customers?”.

Julien Gauthier: “Absolutely, because everyone wanted to be on streaming or be on crypto as well. I saw a good opportunity to do that. We were GPU cloud provider at the beginning. Before AI, because we did cloud gaming, HPC and 3D rendering. Then, more and more AI because it was the booming in 2023 – 2024. It was the release of ChatGPT. So it was a good way to go more and more on this area. So we got at the moment something between 80 to 90% of our demand for AI rather than HPC or 3D rendering what we got at the beginning.

Michael Reid: “Okay, so you started with the 3D rendering which is more around gaming and video as well?

Julien Gauthier: “ Yes, more or less because we decided to aggregate compute for 3D rendering like you can do CAD (Computer Aided Design).

Michael Reid: “And what’s that? Who uses that? An example of users?

Julien Gauthier: “Users like companies who needs to build video games.

Michael Reid: “Actually building the game, not running the game, they design the content? ”

Julien Gauthier: “Rendering, everything like characters, etc. You need a lot of compute to do that. Compute and GPUs could be more or less the same. But here I’m talking about GPUs. GPUs are more parallized compute. So it’s a very efficient for that type of workload. At the beginning, NVIDIA was involved in this area. Based on that, they not pivot, but they saw very interesting demand for that. And in 2020, 2021, they built a little new architecture and a new design called VOLTA. They released something called V100. Before that, you got a P100. A full lineup of new products for AI. You got V100, A100, H100 right now. And it’s a cluster we are building in Paris: 1,000 GPUs. So it’s called more or less large clusters, close to a mega cluster. Because when you need 1,000 GPUs, it’s to build a foundation model.

Michael Reid: “So you’re actually training a model when you’re building that? So before we get to that, so you’re bouncing along, creating NVIDIA server for some rendering companies. AI appears 2022, 2023, somewhere around that. Is it 23? I can’t remember now. I think it’s something like this with GPT-3, because of course they got GPT-1, GPT-2, but it wasn’t spread like a massive deployment for any person. Not customers were really figuring out what to do with it. So a year later, probably GPT-3, what have you come out of it? Then you start to see demand for GPUs and then GPU as a service. So you’re sitting there ready to roll because you’ve actually been building this for a while. You’ve got your relationships in place. And so all of a sudden people turn up and they’re like, time to scale. Is that it?

Julien Gauthier: “Absolutely. We decided to add new features for that. Something we call “AI as a service” with API endpoints, capability to deploy any AI models from Hugging Face for example. So every open source models are hosted on Hugging Face, and you would enable that to be loaded on your GPUs.

Would you like to know the rest of the interview?

Follow the link to watch the full interview from Uplink studio: GPU Powerhouse: Scaling an AI Cloud in the Heart of Europe

Follow the link to listen the full podcast: GPU Powerhouse: Scaling an AI Cloud in the Heart of Europe – Uplink: AI, Data Center, and Cloud Innovation Podcast