GPU Powerhouse: Scaling an AI Cloud in the Heart of Europe

The AI revolution needs more than models - it needs massive infrastructure. And Julien Gauthier is building it.
In this episode of Uplink, Julien, CEO of Arkane Cloud, joins host Michael Reid to unpack how his company scaled from 3D rendering and gaming to delivering GPU cloud services for AI workloads across the globe.
We explore how Arkane built a 1,000-GPU cluster in Paris (with capacity for 6,000), the rise of inference workloads in Europe, and the real-world engineering and business challenges of deploying high-density infrastructure - including cutting-edge liquid cooling handling 135kW per cabinet.

This episode covers:
🤖 How Arkane pivoted from game servers to AI infrastructure
🌍 Why most of their customers are American companies deploying in Europe
⚙️ The evolution of NVIDIA GPUs, from P100 to B200
🚰 The liquid cooling systems that make high-density GPU clusters possible
💰 The financial risk of scaling before customer contracts are secured
🧠 What’s next for AI infrastructure heading into 2025

🎧 Listen to the audio version on Spotify, Apple Podcasts, and more:
🔗 https://www.uplinkpod.com/

🚀 Uplink explores the future of connectivity, cloud, and AI—with the people shaping it. Hosted by Michael Reid, we dive into cutting-edge trends with top industry experts.

👍 Like, comment, and subscribe to stay in the loop.

Chapters:

0:57 Company Origins During the Pandemic

2:15 Evolution from 3D Rendering to AI

5:00 NVIDIA GPU Generations and Clusters

9:18 GPU-as-a-Service Business Model

12:18 Liquid Cooling and Infrastructure Challenges

17:40 Global Strategy

26:31 Future Growth and Industry Outlook