Acquired's Ben Gilbert and David Rosenthal live from Google Cloud Next

Ben Gilbert and David Rosenthal from the Acquired podcast join Jason Davenport on the Google livestream stage. Together, they analyze the massive shift from the "Era of Search" to the "Era of Agents" and what it means for the next decade of business. In this high-energy conversation, the trio explores the rapid evolution of AI infrastructure and why the "boring" parts of the stack—like inference chips—are actually the most exciting part of the story. Key Highlights: TPU v8 & The Inference Shift: Why inference (running models) is officially eclipsing training as the dominant workload, and how Google's bet on custom silicon years ago is finally paying off. The "Internet but Faster": Comparing the adoption of AI to the early web crawl. Training is the "one-time crawl," while inference is the "ongoing query" that powers daily life. The "Pseudo" Barrier: Why the world needs a "UI for agents" to move AI out of the developer terminal and into the hands of 99% of users. Google Cloud’s "Refounding": How Thomas Kurian shifted Google's DNA to become a customer-focused, enterprise-first organization (and why having a 1000x larger sales force matters). Killer Native Apps: Why we are currently in the "efficiency phase" of AI (digitizing the newspaper) and what the true "native" AI business model might look like. YouTube: The Giant in Plain Sight: Why AI is the key to unlocking the research potential of the world’s largest media platform. Watch more from Google Cloud Next '26 and see how the integrated stack—chips, cloud, and models—is closing the gap between hype and value. Book on Inference Engineering → https://goo.gle/4e4l05p NVIDIA Nemotron Super model API → https://goo.gle/3QTSb1S Google Gemma 4 E2B dedicated deployment → https://goo.gle/4tDafvC #AcquiredPodcast #GoogleCloudNext #AIAgents #TPUv8 #BusinessStrategy #GenerativeAI