Investor Jeremy Giffon on Private Markets, the Death of SaaS, and AI's Capex Era
This episode does not feature representatives from major AI labs speaking in an official capacity. Instead, it is a wide-ranging conversation between host Patrick O'Shaughnessy and guest Jeremy Giffon of Atreides Management, an investor with broad exposure to private technology markets. Their discussion of AI is filtered through an allocator's lens rather than a builder's. They argue that the current AI cycle is fundamentally reshaping the economics of software: the old SaaS model of selling a copy of a string at near-zero marginal cost is being displaced by a compute-heavy model where the marginal cost is paid on every call. Giffon describes this as a shift from high-gross-margin software to thinner margins rewarded by massive scale, comparing it to a 'Walmart effect' in software. He notes that roughly 60% of Atreides' portfolio by market value is now 'not pure bits,' a striking departure from a decade of pure software venture investing. The hosts frame the rise of high-capex AI businesses as partly a release valve for an oversupplied venture capital market that had run out of places to deploy incremental dollars before the AI buildout. They also discuss how AI-driven timelines (X, YouTube) increasingly set the narratives that move public markets, and express skepticism about the 'death of software' consensus while acknowledging that per-seat SaaS pricing specifically looks vulnerable. On labor, Giffon argues most white-collar work is 'fake and made up' and that any job which can be automated should be, while accepting that short-to-medium-term displacement could be painful. The conversation also covers Giffon's 'billion dollar PDF' thesis, the feudal economics of SPV allocations in late-stage private companies, why founders and capital are subservient to the 'poster class' on X, and the hidden philosophical traditions (neo-Buddhist utilitarianism, effective altruism, Curtis Yarvin-adjacent thought) he believes underpin Silicon Valley's worldview. For an AI-labs tracker, the useful signal is Giffon's market-level read: AI is redrawing software economics, the capital cycle is rewarding compute-heavy and real-world-asset businesses over traditional SaaS, and the loudest voices shaping both public and private market AI valuations now live on the timeline.
Original sources
Podcast Invest Like the Best
Episode:Jeremy Giffon - The Billion Dollar PDF - [Invest Like the Best, EP.481] ↗
AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.