In this episode, Shilpa, Kevin, and Jared unpack the turbulent sequence of events around Windsurf, first the reported collapse of an anticipated OpenAI deal amid Microsoft data-access sensitivities, then Google’s selective leadership and IP licensing move, and finally Cognition’s rapid-fire acquisition of the remaining team. We explore what this “strategic decapitation” style of transaction signals for employee expectations, option value, and emerging term sheet protections, while carefully distinguishing between reported facts and interpretation.

That segues into a broader look at the tightening AI talent market: how corporate constraints, antitrust scrutiny, and a narrowed IPO window reshape exit math; why Section 174’s restoration (allowing immediate expensing of U.S. R&D again) may advantage deep-pocketed incumbents; and the downstream implications for early-stage hiring, compensation, and founder signaling.

We then analyze the milestone of OpenAI’s and Google’s models achieving International Math Olympiad gold-level performance, what’s genuinely new (end-to-end natural language reasoning, multi-hour coherence, emergent refusal when out of depth), and why selective abstention may matter as much as raw accuracy for hallucination reduction. The team examines long-horizon “think + tool” workflows, shrinking core model footprints via externalized fact retrieval, and how product interfaces must evolve beyond chat metaphors as reasoning spans hours.

Finally, we reflect on what Olympiad-grade reasoning might mean for education, assessment integrity, and the division of cognitive labor, plus a lighter close: Millennium Force at Cedar Point, a reclaimed planer from a Rotary sale, and a standout Seattle donut run. A dense week of AI strategy, economics, and capability inflection, packaged for builders calibrating what’s signal versus noise. Tune in and tell us where you think the next structural shift lands.