MDCS.AI ads

If your AI is strategic, does it belong in a rented and shared environment? Harbour gives you a sovereign AI Foundation. Best tokenomics. Simple to run. Predictable in cost. Start with a board discussion.

Helping doctors see cancer sooner. Fons van der Sommen develops AI that helps doctors detect cancer at an earlier stage. He opens the GTC Viewing Party on 26 October. AI at its most human. 📅 Monday 26 October. 14:30 to 20:30. TU Eindhoven Full program and registration: gtcviewingparty.com

Win-lose agents take from us. Win-win agents build with us. AI agents will act on our behalf. So how should they behave? Dr. Anneloes Smitsman has a clear answer. Not grab the short-term win. Protect what we share. Choose the outcome where everyone wins. Her EARTHwise AI Arena trains agents to do exactly that. On 26 October, she brings that lens to the GTC keynote. 📅 Monday 26 October. 14:30 to 20:30. TU Eindhoven Full program and registration: gtcviewingparty.com

Boris Vermaas from Cisco raised something in our conversation that most teams skip past too quickly. Most infrastructure discussions start with hardware. Which GPUs. How much storage. What networking. But for many organizations, the harder question comes before that. What's the use case? AI is everywhere. The fear of missing out is real. Teams want to move, but they're not sure where to start. Boris's advice: find a partner who helps you define the use case first. What can AI actually do for your organization? What do you need to make it work? Answer that before the infrastructure conversation. 🎧 Up for an inspiring lunch break today? Listen to episode 2 of the Your AI Your Way Podcast now: https://lnkd.in/eKYh7cny

Data moves to the GPU and back to storage. In cloud, every one of those hops is a separate charge on your invoice. I see teams discover this after the first production sprint. Sander ten Hoedt from Cisco maps this out in under two minutes. Translate a reference architecture to cloud, and every sub-element becomes its own cost. Networking, storage tiering, data transfer fees, all layered on top of the GPU price you thought you were paying. On-premise with a validated architecture, those charges disappear. The data sits in your storage. No fees per hop. Cost planning becomes predictable from day one. Full podcast covers where this complexity builds up and how reference architectures cut through it. 🎧 https://lnkd.in/e4ku973B

Most AI pilots never reach production. Not because the technology fails. Because the cloud invoice arrives. It starts small. A few GPUs in the cloud. Reasonable invoices. Then the project scales. Storage costs appear. Data transfer fees stack up. GPU utilization drops below thirty percent while the invoices keep climbing. In this podcast episode, Guy D'Hauwers (Automation Group) and Sander ten Hoedt (Cisco) break down what actually drives AI infrastructure costs, and when it makes sense to move from cloud to owned infrastructure. What this episode covers: - Why AI infrastructure costs appear where teams least expect them - What happens to utilization when storage and network can't keep up with compute - How to think about the tipping point between cloud and owned infrastructure Listen to Episode 3 of the Your AI Your Way Podcast, by Cisco & MDCS.AI.