Harry Stebbings@HarryStebbings2d ago

I get more messages about this podcast every week than anything else I do. We all want to look and sound smart in the investment committee or all hands meeting. This is the only show in tech you have to listen to every week. No egos, no politics, just tech. AGENDA: - Venture is Broken: Trillion or Bust - Coreweave Sell-Off Creates Panic - Town at $1B, Anduril at $90B: Are Revenue Multiples Out of Control? - General Atlantic Eyes IPO - Modal Raises $750M at $15.75B My notes with @jasonlk and @rodriscoll 1. Why Fundraising Is Back to Being a Numbers Game Founders with top-decile numbers now have to meet 100 VCs, as in the old days. It will not be Benchmark, Index, Andreessen or Bessemer, but other growth investors will take the bet at a reasonable but still elevated valuation because it is the best deal they have seen that quarter. 2. Why the Big Funds Have to Have Billion-Dollar Positions to Make It Worthwhile? A $5B plus fund needs $BN positions in everything. Without 51% ownership, which most founders will not accept, a $2B Listen Labs is not good enough, while 1% of Anthropic could still get there even with dilution. Love to hear your thoughts on this @venkyganesan @DavidGeorge83 3. Why Are CoreWeave and Oracle B-Tier and What Does That Mean for Their Price Volatility? With massive debt loads, CoreWeave and Oracle should be hypersensitive to any bump at Anthropic or OpenAI. CoreWeave exists because NVIDIA needs it for now, and Oracle rents out excess capacity because it cannot sell enough enterprise software. 4. Why the Best Markets for App Layer Companies Compete Directly With the Models The fastest-growing AI apps, such as code generation, code review and legal, sit closest to the foundation models. That brings a moment of terror every six months, but it is also why companies like CodeRabbit are growing faster than ever. Love to hear your thoughts on this @spdholakia @mmurph 5. Why Code Review Is the Single Most Important Category in Software AI-generated code is pretty good but never production-ready, and reviewing it takes an order of magnitude more time than building it. Getting code into production matters more than generating it, so the best code review teams would be worth 10 bets. Love to hear your thoughts on this @reidRMC @evrishria @Joshuabrowder 6. How Big Is the Market for Inference Truly? The key inference providers add up to roughly $5B to $6B in run-rate revenue, under 5% of OpenAI and Anthropic combined. If tokens are roughly the same, Replit would rather pay Modal a 30% gross margin than give Claude Code 70%. How do you analyse the TAM for inference today? What do people not see that they should see? @sarahcat21 @bernhardsson 7. Why Bending Spoons Is a Berkshire Hathaway of Software Bending Spoons buys software to hold, the way Warren Buffett wanted to hold See's Candy forever. It made Vimeo and Evernote work with price increases, and with Miro and Airtable the quality of its assets is improving. Love to hear your thoughts on this as an analogy @ncsh (links in comments)