Summary
Harry Stebbings and the 20VC panel dig into a packed week of AI industry news, anchored by the close of SpaceX’s $60 billion all-stock acquisition of Cursor — a remarkable outcome for a product that nearly seemed finished when Claude Code launched. The discussion covers why SpaceX was uniquely positioned to make the deal (compute infrastructure without a clear software business on top), why Meta passed despite similar strategic logic, and what Cursor’s trajectory from near-dead to $60B says about the importance of multi-model agility in AI product development.
Anthropics first-ever profitable quarter — on $11.5 billion of Q2 revenue — gets a detailed financial breakdown. The hosts explain why the massive stock-based compensation figures that will appear on Anthropic’s books are economically misleading in a hypergrowth context, why off-balance-sheet compute commitments are rational bets rather than liabilities if revenue continues compounding, and why the only number investors will care about is the 2027–2028 revenue projection.
The episode rounds out with a discussion of consumer AI valuations: Higgsfield raised at $5.5 billion and Lovable at $13.3 billion. The panel debates how to apply traditional SaaS metrics to AI-native consumer products and what gross margin conversations mean when the category is still in land-grab mode.
📺 Source: 20VC with Harry Stebbings · Published August 20, 2026
🏷️ Format: Podcast
