CharlieKey takeaways from your favorite podcasts
← 1 October 202620VC with Harry Stebbings

Instinct Raises $1B at $10B Valuation | AMD Buys World Labs for $8.2B | Bessemer Raises $5.75B

AI is producing extreme valuations, huge early rounds, a talent war and active acquirers, so investors face high variance, size bets carefully, and watch inference, open-weight models and IPO signals.

Watch on YouTube ↗

Charlie's summary

  1. Anthropic's leaked S1 told investors little; Q3 numbers and market structure matter, and the risk disclosures may fuel political backlash

    Rory says 2025 numbers are long spent and the leak held almost nothing new beyond two customers making up 25% of revenue. The Q3 revenue figure is “90% of the data required” to price Anthropic. Jason worries the negatives, such as existential risk, could hurt retail sentiment after the IPO. Rory adds that a Florida suit shows the disclosures could invite AGs to sue.

    ▶ 1:31
  2. Consumer agents like Instinct are a new paradigm, and Benchmark treated a $10B round as an early-stage bet

    Jack sees agents that act across the third-party internet as a shift on the level of chat and coding, where “a lot of things can win.” Benchmark first invested at $2.5B, then at $10B, and saw it as early stage because the paradigm keeps changing. Jason's open question is whether people will use these agents eight hours a day, as they do coding tools. Jack notes Harvey and Legora once barely worked and now are used all day.

    ▶ 7:42
  3. Size huge early bets by shots on goal and Kelly logic, since variance is the highest it has been

    Jack says sizing is bounded below by needing enough chances at something great and above by needing to matter in the round. Harry would put 5% of a fund in Instinct, citing downside protection from likely acquirers. Rory says full Kelly would imply about 30%, but his LPs would balk. Jack says valuations and traction are both extreme, so “the odds of having it right are zero.” Tyler Cowen's line: variance goes up, and many will fail.

    ▶ 17:07
  4. Acquirers are now plentiful, so AMD's $8.2B World Labs deal may be the first of many

    Harry thinks foundation-model companies will want to buy robotics or world-model teams. Jack notes “there are like 10 companies that can do $10 billion acquisitions and want to,” and a sale is quicker than going public. Jason argues that for world models, proven technical credentials matter. AMD, up 279% this year, paid about 8% of its market cap to stay in the race.

    ▶ 25:59

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.