OpenAI's Departure: A New Chapter

The episode opens with a rapid-fire overview of four significant developments in the AI and tech space before diving into each in detail.

First, Stuttgart-based robotics startup Secreact raised $110 million in a Series B round led by Headline, with participation from Bullhound, Felix Capital, Airstreet, and others. The funding supports the opening of a Boston office and the scaling of Cortex 2.0, their proprietary 'robot brain.' Unlike conventional vision-language-action systems that observe and immediately act, Cortex 2.0 incorporates a world model that simulates the physical consequences of potential actions before executing them. This allows robots to reason about outcomes — for example, predicting what happens if an arm moves too fast near a stack of objects — improving safety and reducing errors in industrial settings. Customers include BMW, Daimler Truck, and PepsiCo, and the system has completed over one billion real production picks. The host draws a parallel to Jeff Bezos's stealth AI lab, which Bloomberg reported is pursuing a similar world model thesis with a rumored $10 billion round.

Second, David Silver — creator of AlphaGo, AlphaZero, and AlphaStar at DeepMind — announced a $1.1 billion seed round for his new startup, Ineffable Intelligence, at a $5.1 billion valuation. The round, co-led by Sequoia and Lightspeed with participation from Nvidia, DST, Index, Google, and the UK Sovereign AI Fund, is described as the largest seed round in European history. Silver's thesis is that the most important aspects of intelligence are 'ineffable' — literally impossible to capture in language — which he argues means LLMs will hit a fundamental ceiling. His approach instead focuses on large-scale reinforcement learning agents, world models, and what he and researcher Rich Sutton termed 'the era of experience.' The host expresses personal conviction in this direction, citing firsthand experience building production applications on top of LLMs and encountering their limitations. A Sequoia partner's quote — 'Language is a beautiful compression of the world. It is not the world' — is highlighted as particularly apt.

Third, the long-running legal battle between Elon Musk and Sam Altman officially went to trial at the Oakland Federal Courthouse, with Judge Yvonne Gonzalez Rogers presiding. Jury selection began, with opening arguments scheduled for the following day and a four-week trial ahead. The witness list includes Musk, Altman, Greg Brockman, and Satya Nadella. Of the original 26 claims Musk filed in 2024, only two survived: unjust enrichment and breach of charitable trust. The core argument is that Musk donated tens of millions to OpenAI as a nonprofit, and Altman allegedly defrauded him by converting it into a for-profit entity now valued near a trillion dollars. The host argues the case has implications far beyond a personal feud — Judge Gonzalez Rogers could potentially pause OpenAI's for-profit conversion or reshape the rules governing how nonprofit AI labs commercialize, with potential downstream effects on Anthropic and future AI ventures.

Fourth and most extensively, the host covers the restructured OpenAI-Microsoft partnership. Microsoft loses its exclusive license to OpenAI's models, meaning OpenAI can now sell directly on AWS, Google Cloud, and other platforms. Microsoft will no longer pay a revenue share to OpenAI on resold products, while OpenAI's revenue share to Microsoft is capped at 20% through 2030. Microsoft retains a non-exclusive IP license through 2032 and keeps its approximately 27% equity stake, currently valued around $135 billion. Microsoft's stock dropped roughly 5% on the news before partially recovering. The host argues this is strategically bad for Microsoft for several reasons: the announcement lands two days before Microsoft's Q3 earnings call, where Azure growth is already decelerating; it formalizes what an internal OpenAI memo from April 13th had already signaled — that the Microsoft partnership had actively limited OpenAI's enterprise reach; and it reduces antitrust exposure by eliminating the exclusive arrangement. The host contends that Azure's outperformance of AWS over the past two years was largely driven by the OpenAI lock-in, and that AWS can now compete on equal footing with the same models, which could significantly shift enterprise cloud demand.