AI Daily Podcast: Cheaper AI, Smarter Workflows

In this episode of AI Daily Podcast, we explore two of the biggest shifts redefining artificial intelligence innovation: the race to make AI infrastructure more efficient and affordable, and the rise of AI systems that behave less like tools and more like coworkers. The episode begins with a look at the changing economics of AI. As attention moves beyond model size and benchmark wins, the spotlight is turning to infrastructure efficiency. A key example is OpenAI’s reported custom chip effort with Broadcom, code-named Jalapeño, which reflects a growing industry belief that the future of AI depends not only on more compute, but on cheaper and more optimized compute. We also break down new revenue data showing that global AI revenues outside China reached $25 billion in Q1 2026, topping estimated depreciation costs of $21 billion for the second straight quarter. The signal is important: demand is real, but the economics remain tight. From there, we examine what this means for the next phase of innovation. AI is increasingly entering an industrial optimization era, where custom silicon, networking, memory, power efficiency, thermal design, and software optimization may matter as much as model intelligence itself. The conversation also highlights why vertical integration is becoming more strategic, as leading AI companies seek deeper control over chips, cloud systems, and deployment costs. We connect these infrastructure trends to practical enterprise use cases like supply chain planning, where AI can deliver measurable business value and help justify the enormous cost of the ecosystem. The second part of the episode turns to a different but equally important frontier: the growing tendency for people to treat AI like a teammate. As software shifts from command-based interfaces to agentic systems that can take goals and act on them, human-computer interaction is changing dramatically. AI assistants are becoming more conversational, more persistent, and more socially present through innovations like voice mode, memory, multimodal interaction, and conversational continuity. These features improve usability, but they also increase personification, making it easier for users to project trust, empathy, and authority onto systems that do not actually possess those traits. We also explore why this makes governance, oversight, and workflow design one of the most important innovation areas in AI today. If AI is influencing approvals, feedback, hiring, or employee well-being, organizations need auditability, escalation paths, and human-in-the-loop controls. In that world, the most valuable human skill becomes judgment: setting goals, defining limits, evaluating outputs, and recognizing when the AI is wrong. The episode argues that the next major breakthroughs in AI may come not only from smarter models, but from the systems that help organizations manage AI as an active participant in work. Tune in to AI Daily Podcast for a deeper look at how the future of artificial intelligence is being shaped by infrastructure economics, enterprise adoption, human attachment to AI, and the redesign of work itself. This is a conversation about where AI innovation is really heading—and why the most important changes may be happening far beyond the benchmark charts.Links:Broadcom, OpenAI deal hit as infrastructure costs take center stageKI-Nachfrage rechtfertigt Kosten: Umsätze decken erstmals Abschreibungen, zeigt StudieBest Practices for Using AI in Supply Chain PlanningUnsettling Relationships Developing Between Workers And AI Coworkers