Is AI Really That Hard: A Real Talk about ML Infrastructure with Alex Patrushev

That Tech Pod

On this episode of That Tech Pod, Laura and Kevin sit down with Alexander Patrushev, Head of Product for AI/ML at Nebius, a European cloud platform focused on machine learning infrastructure. Kicking things off, we unpack the buzz around China’s fully AI-powered “Agent Hospital,” raising questions about what role AI should play in medicine. Laura also shares a personal rant about navigating the healthcare system in search of real answers.

Then, Alex joins from Luxembourg to explain what Nebius does, who it's built for, and how it helps companies tackle ML workloads. He breaks down the common mistakes teams make when launching ML projects and why success depends more on people and processes than just tech. The conversation also goes into whether companies should host their own LLMs or use pay-per-token APIs, and how to approach that decision. Looking ahead, Alex challenges the idea that AI is inherently hard and shares what businesses should really be focusing on, from smarter software to evolving hardware needs, as AI infrastructure continues to grow. It's a thoughtful, practical look at the future of AI from someone who builds the tools to power it.

Alexander Patrushev is an experienced IT leader with over 15 years in the industry, including the past six focused on artificial intelligence. He currently serves as Head of Product for AI/ML at Nebius, a European cloud platform designed for building and deploying machine learning workloads. Based in Luxembourg, Alexander is known for his practical approach to solving real-world challenges faced by AI practitioners. He’s especially focused on delivering user-centric, innovative solutions, and believes strongly in the importance of in-house R&D to drive product excellence and stay ahead in a fast-moving space.