In this episode of AI Unchained, Jack and Corr dive into the accelerating AI race and the surprising moves shaping it. They unpack Meta’s scramble for compute—including massive “tent” data centers built at record speed—Airbnb CEO Brian Chesky’s plan to launch a new AI lab, and a provocative research thread questioning whether the classic QKV structure at the heart of transformer models is even necessary.
Along the way, they debate the real bottlenecks in AI development, the economics of massive infrastructure spending, and whether the next big breakthroughs will come from better models—or better interfaces. They also explore Apple’s quiet approval of a startup AI agent inside iMessage and what it could signal about the future of agent-based apps.
Is this the beginning of a more efficient generation of AI systems and agent-powered interfaces—or another step in a fast-moving hype cycle? Jack is bullish. Corr remains skeptical.
