Into The Madverse

AI performance is not decided by GPUs alone because the network now determines how effectively thousands of accelerators can operate as one machine. Clayton Wagar, Leader of AI and High Performance Networking at Nokia, explains why congestion, resilience, optics and fabric design could matter more than adding another rack of compute.

Mark Venables speaks with Kevin Roof from LiquidStack about why heat is becoming one of the defining constraints of the AI economy. As GPU densities surge towards megawatt-scale racks, this episode explores how liquid cooling, modular infrastructure and radically new data centre architectures are reshaping the future of AI infrastructure before the industry melts under its own success.

Mark Venables speaks to Pablo Ruiz Escribano from Schneider Electric about why the AI boom is fundamentally rewriting the assumptions behind Europe’s data centre industry. From megawatt racks and liquid cooling to power accessibility, grid integration and the race to deploy AI infrastructure faster than engineering culture can adapt, this episode explores why the future of AI may ultimately be constrained not by models or chips, but by energy, cooling and physical infrastructure.

In this episode of Into The Madverse, Mark Venables speaks with David Weinstein, CEO at KayOS, and an AI strategist and systems thinker, about why the next phase of AI is not about replacing people, but scaling human intuition, interpretation, and curiosity. From encoding organisational bias into AI agents to building infrastructure that prioritises context over models, the conversation explores why companies obsessed with automation may be missing the real competitive advantage.

In this episode of Into The Madverse, Mark Venables speaks with Daniel Meyer, Chief Technology Officer at Camunda, about why AI in regulated industries fails at scale when enterprises cannot prove what the system did, why it did it, and who stayed accountable. They unpack agentic orchestration as the pragmatic route to autonomy, blending deterministic controls with dynamic AI, keeping humans in the loop, and turning “start small” into something that can survive compliance, audit, and legacy complexity.

In this episode of Into The Madverse, Mark Venables speaks with Matt Penton, Director of Data & AI at BIP Group, about why relying on a single large language model is a fundamentally flawed strategy for enterprise AI. They explore multi-LLM consensus, agentic supervision, bias, hallucination, and why no model has a monopoly on truth when real decisions are at stake.

Mark Venables is joined by Sean Grimm, US Country Manager at Ipsotek, to explore why security control rooms remain stuck in passive monitoring despite massive investment in surveillance. They discuss how AI is turning video into proactive intelligence, and why the real opportunity lies in using data organisations already have.

Mark Venables speaks with Dr Ian Robertson, Interim Director at Tandem Health, about how AI is reshaping healthcare not through sci-fi diagnosis but by eliminating the administrative burden that consumes 40 percent of clinicians’ time. They explore how ambient AI, automation and intelligent workflows can reduce burnout, improve care and transform the patient experience.

Mark Venables grills Justin Sharrocks, Managing Director UK/EU, Trusted Tech Team, on why so many AI projects stall at pilot and how to break through to real impact. They tackle hype versus outcomes, data readiness, realistic timelines, and the incentive and leadership shifts needed to turn adoption into measurable value.

Most companies think they are ready for AI because they have a data lake, a chatbot, or a Co-Pilot licence, but Lori MacVittie reveals why those assumptions are dangerously wrong. In this episode, she dismantles the myths of AI maturity and explains what it really takes to scale, secure, and trust AI before the entire enterprise feels the impact.