CharlieKey takeaways from your favorite podcasts
← 3 November 2026TBPN

Anthropic Q3 Revenue, Europe’s Freezer Hack Scare, Northwood Raises $375M | Diet TBPN

Joe Eisenthal and the hosts discuss midterms, prediction markets, and whether AI needs a capabilities jump to justify its spending, then cover Anthropic's Q3 growth, hacked European freezers, and Northwood's raise.

Watch on YouTube ↗

Charlie's summary

  1. Markets are good “conventional wisdom distillers,” and prediction markets deserve to be taken seriously

    Joe still trusts markets, prediction markets included. He argues that calling them conventional wisdom is not a sneer, because they turn a vague sense that a candidate is favored into a number, such as 55% versus 80%. Few people can confidently say a market is wrong, so a 40% chance of a Democrat winning Texas has to be taken seriously.

    ▶ 3:03
  2. No one can agree who profits from AI, so the AI trade has no clear winners yet

    Joe says Wall Street talks about the AI trade, yet even if everything goes well there is no consensus on who makes the money. It could be the labs, the chip makers, the Microsoft-plus-Palantir diffusion players, or the rest of the economy. He also notes that past technologies like the internet and mobile phones left productivity effects that are hard to see in the charts.

    ▶ 7:37
  3. The AI trade needs a big capabilities jump soon, because spending is concentrated in tech

    One host argues the overall trade only works with a massive capabilities jump soon. Torsten Slok's view, as relayed, is that the Mag 7 see positive ROI while tire makers, foresters and oil and gas firms struggle to improve revenue or replace jobs with AI. Goldman is clamping down on non-technical staff, including traders, vibe coding, partly over productivity. Few non-engineers can justify $10,000 a month on AI.

    ▶ 12:12
  4. Today's agents lack true agency and are clones of each other

    Joe leans on the “messy jobs thesis”: jobs bundle about 30 tasks, so AI doing one task does not replace the job. He says the next leg up needs real agency, such as giving an agent a budget cap and letting it spend it to grow the business. Today's agents are sub agents that only do what they're told. A hundred identical PhDs would not work as a team. Multi-agent teams need genuinely different points of view, and nobody has cracked that orchestration.

    ▶ 14:17

Give your email to unlock the rest of the key points — for every episode.

Do more with Charlie

Free — sign in with your email, no password.