Podwise Weekly Picks, Week 45: Who Pays for Growth?
Ten English-language episodes from Nov 2–8: the $4T AI sell-off, Anthropic's S-1, vibe coding, Mercado Libre, and the money behind the buildout.
Weekly picks6 min read
The through-line this week is cost. Not the sticker price of a subscription, but the question underneath almost every conversation: who pays for growth, and who carries the risk when it stops? The midterms turned on affordability, the AI rally turned on debt, and even the longevity advice came down to where the money goes before people get sick.
These are the ten episodes from the Podwise Weekly Picks shelf for the week of November 2. The shelf covers English-language shows only; the Chinese picks are published separately as Podwise 每周精选. Each entry says what the episode is about and who it is for, and the show name links to the episode in Podwise if you want the summary and transcript.
The ten
Midterm Blood Bath, $4T AI Sell Off, Data Center Freeze
All-In with Chamath, Jason, Sacks & Friedberg · 7 Nov · 1h 33m
Four trillion dollars of market value came off the AI trade, and the panel argues the sell-off is less about earnings than about who can still finance the buildout. Their sharper point is political: data centers have become totems in the split between asset owners and wage earners, and local moratoriums could push AI capacity offshore. For listeners who want the bear case and the political read in one sitting.
Anthropic's S-1, read as an operating business
Latent Space · 7 Nov · 1h 42m
Altimeter's Jamin Ball goes through the filing line by line. The number that matters is efficiency: inference revenue per unit of compute rose about 7x in a year, gross margin sits near 85%, and the company is already free-cash-flow positive. The reported net loss is largely an accounting effect from convertible notes that revalue as the valuation rises. The open question is the $855 billion compute commitment, which only works if enterprise demand keeps compounding.
What it actually takes to build frontier models
Lenny's Podcast · 8 Nov · 59 min
Tulsee Doshi runs product for Gemini, and her framing is that frontier models are grown rather than built, so the PM's job is to find the magic that emerges and turn it into something users can trust. She offers a 60-20-20 split — shipping what works today, improving what is close, and reserving room for experiments — plus a warning that evaluation sets need to be small and high-signal. For product people working with models they cannot fully predict.
How To Build The Future: Paul Graham & Amjad Masad
Y Combinator · 7 Nov · 44 min
Paul Graham and Amjad Masad trace vibe coding back to a decade of watching people patch the gap between a model and a working app. Masad's line is that the visible English prompt is only the tip of the iceberg; sandboxing, deployment and file systems are the moat. The episode is also a compact argument for why cheaper programming expands the number of programmers super-linearly. For founders and anyone building internal tools.
Inside Salesforce's $100 Billion M&A Playbook
The a16z Show · 6 Nov · 49 min
John Somorjai has overseen nearly $100 billion of acquisitions at Salesforce, and he explains the build-versus-buy test: distribution, talent, and a product that fills a real enterprise gap. His most useful warning is that ARR can mask attrition, and that pricing is shifting from seats to outcomes. For founders thinking about selling, and operators who benchmark on ARR alone.
Mercado Libre's fintech flywheel
The Investor's Podcast · 8 Nov · 1h 04m
A two-part bull case for Mercado Libre that spends most of its time on the flywheel, not the headline growth. Fintech is now 43% of revenue and growing faster than commerce, while the margin drop from 12% to 7% reflects credit provisioning and logistics investment rather than a price war. For investors who want the mechanics of a Latin American compounder, including the credit risk.
The health interventions actually worth your time
The a16z Show · 8 Nov · 26 min
Justin Mares makes a simple, uncomfortable point: about 90% of US healthcare spending goes to chronic conditions, and less than 3% goes to prevention. His answer is not a new supplement stack but sourcing, sleep, sunlight, strength, and using HSA/FSA money to pay for them. For listeners who want practical prevention without the biohacker theatre.
Controlling biology with virtual cells
Latent Space · 6 Nov · 51 min
Fabian Theis describes a shift in virtual cell models: from static atlases that describe cell types to world models that predict how a cell responds to a genetic or environmental perturbation. That makes them experiment-design tools rather than replacements for the bench, and the episode stays close to what the models can predict today. For listeners tracking where AI in biology has moved past the demo.
An endowment CIO on SpaceX, AI and concentration
The Twenty Minute VC · 7 Nov · 58 min
Scott Wilson runs Washington University's endowment, and he explains the shift from fund-of-funds to concentrated direct positions: the portfolio went from $8 billion to $19 billion, with SpaceX alone above 20% of assets. His read on AI is unsentimental. Most of these companies will not exist in ten years, and for a large share of tasks the frontier model is interchangeable, which puts the moat somewhere other than the model. For allocators, and for founders who want to hear how the money on the other side of the table is thinking.
How the Democrats Did It
The Ezra Klein Show · 7 Nov · 1h 07m
Michelle Goldberg spent a year in Texas, Michigan and Maine, and she and Ezra Klein reconstruct how Democrats won in places they were supposed to lose. The common thread is economic populism aimed at concentrated corporate power, paired with a moral vocabulary that does not sound like a focus group. For anyone who wants to understand the coalition that may emerge from this midterm, not just the seat count.
How these are picked
The shelf is curated by Podwise from the week's episodes, and the order here follows the collection. This English edition covers English-language shows only; the Chinese edition, Podwise 每周精选, is picked separately from Chinese-language shows. Summaries and transcripts for all ten are already processed, so opening any of them costs no credits.
Browse the full shelf, including past weeks and themed collections, on Explore.
