Broadcom Frames Platform Engineering 2.0 for AI

Alan Shimel and Luca Galante talk with Pankaj Gupta of Broadcom about why Platform Engineering 2.0 is emerging as the next evolution of internal platforms as AI changes how software is built, deployed, secured and governed. Gupta explains that platform engineering is no longer optional for modern application delivery, but the model is expanding beyond developer productivity and internal developer portals into a broader operating framework for AI-native enterprises. Rather than describing this shift as a rip-and-replace reset, the discussion frames Platform Engineering 2.0 as an evolution built on proven foundations such as platform as a product, golden paths, self-service, standardization and reduced cognitive load.

The conversation explores how AI is changing the role of developers from writing code to validating, governing and moving code safely into production. As AI agents become new platform users, organizations will need platforms that support both human and non-human personas with guardrails, bounded permissions, observability and governance. Gupta, Shimel and Galante also examine why the platform audience is expanding to include security teams, compliance leaders, FinOps stakeholders, data scientists, machine learning engineers, business leaders and AI agents that may eventually outnumber human developers.

A major theme is that AI creates revolutionary pressure, but IT teams still need evolutionary paths that protect existing investments. Platform Engineering 2.0 is presented around five pillars: AI-native platforms, multi-persona experiences, embedded FinOps, security shifting down into the platform and composable design. The episode also highlights why security must become more native, immutable and invisible to developers as AI increases vulnerability discovery, shadow IT, prompt injection, model poisoning and inference data leakage risks. For CIOs and engineering leaders, the takeaway is clear: Platform Engineering 2.0 gives organizations a practical way to stay in control of AI adoption, data, costs, security and delivery speed while still enabling experimentation, productivity and measurable business value.