Synthesis

Exits, write-downs and split landings

How companies end or are sold in the AI cycle: write-downs, acquisitions at far below peak value, split landings, pivots and acquihires, and what early investors get.

Type
Synthesis
Confidence
Medium
Sources
7 videos
First published
Updated

Nine episodes mention exits, with Airtable the one worked example (August). Jessica put Airtable’s peak valuation at $11 billion and its sale to Bending Spoons at about $1.5 billion; Sam, an early Slow investor who put $3 million in at a $90 million post around 2012, calls the roughly 10x write-down brutal and says it is good to be a seed investor. Dave names the ‘split landing’: the old product goes to a good acquirer, the core team keeps going on the new bet (here HyperAgent, which Peter Deng says was carved out), and investors get upside in both. Sam adds that big investors prefer this because it lets them avoid the write-down on their books, and that Slow, as a seed investor, learned of the sale from a tweet. Other cases are reported: Allbirds (over $4 billion to about $40 million; Slow sold earlier), Intercom/Finn at about $3 billion, World Labs at $8 billion after a $5 billion raise (unattributed), the OpenRouter deal with $6 billion of $7.5 billion to investors, and Boom Supersonic’s turbine pivot. Sam admits he has not read the Airtable documents.

Key points

  • Big peak-to-exit gaps are reported: Airtable $11B to about $1.5B, Allbirds $4B to about $40M, with early seed investors protected by low entry prices. — 2 episodes. (Jess · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:42)
  • Split landing (Dave): the legacy product lands with an acquirer, the core team stays on the new opportunity, investors get a piece of both; Sam says larger investors favour it to avoid a write-down. — 2 episodes. (Dave · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 23:20) (Peter Deng · Peter Deng & Rachel Holt (Aug 2026) ▶ 49:39)
  • Sell when the numbers require an unlikely outcome: Sam says Slow sold Allbirds because holding meant believing it was the next Nike. — 1 episode. (Sam · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 09:39)
  • Seed investors have little information: Slow had no information rights and found out about the Airtable sale from a tweet. — 1 episode. (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 24:57)
  • Other exit patterns are mentioned: deals where most proceeds go to investors (OpenRouter), pivots to a different market (Boom to data-centre turbines; Allbirds to AI compute), and a reported high-priced sale (World Labs, Intercom/Finn). — 4 episodes. (Jess · Stripe Buys OpenRouter, OpenAI (Aug 2026) ▶ 23:59)
  • Teamshares’ Mike Brown argues against exits altogether (permanent ownership; a sale at about 6x EBITDA destroys value). — 1 episode. (Mike Brown · Mike Brown (Jun 2026) ▶ 47:03)

Frameworks

  • Split landing — Dave’s name for an exit where the old product lands with a good acquirer, the core team stays together on the new opportunity, and investors get upside in both. (Dave · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 23:20)
    • Land the former product in the best available home
    • Keep the core creative team together
    • Give investors a piece of the new venture (HyperAgent)

Stories & examples

  • Slow’s Airtable investment: Sam put a $3 million check into Airtable at a $90 million post around 2012 after using the product while building Finn; Dave supported it, though it was too far along for Slow, and Slow took money off the table through later rounds. Lesson: Taking a bunch off the table through the rounds let the seed investor make money despite the later write-down. (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:56)
  • Slow and Allbirds: Sam says Slow was an investor and sold at a later round because the only reason to hold was believing it was the next Nike. Lesson: Sell when the numbers require an unlikely outcome. (Sam · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 09:39)
  • Finn / Intercom as a boring outcome: Sam notes Intercom, renamed Finn AI, was reported acquired by Benioff for about $3 billion and calls it unbelievably boring beside SpaceX; he asks how an engineer is recruited to such a company. Lesson: Mid-tier outcomes are hard to recruit for when wild narratives are available. (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 18:02)
  • Boom Supersonic’s turbine: Sam says Boom could not get jet engines and now has a turbine to power data centres, so early investors may profit anyway. Lesson: A pivot onto the narrative can reward early investors for the wrong reasons. (Sam · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 13:18)

Numbers

  • Slow’s cheque into Airtable: 3000000 dollars (host’s framing) (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:56)
  • Airtable valuation at the time of Slow’s investment (post-money): 90000000 currency unstated (post-money) (host’s framing, currency unstated) (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:58)
  • Airtable sale price to Bending Spoons, as put by Jess: 1500000000 dollars (host’s framing, approx.) (Jess · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 19:42)
  • Allbirds peak valuation: 4 billion (currency unstated) (host’s framing, approx., currency unstated) (Jess · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 09:55)
  • Allbirds sale price: 40 million (currency unstated) (host’s framing, approx., currency unstated) (Jess · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 09:55)
  • Intercom/Finn AI acquisition price: 3 billion dollars (host’s framing, approx.) (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 17:22)
  • World Labs’ previous raise: 5000000000 currency unstated (self-reported, currency unstated) (one of the hosts · Adrian (Oct 2026) ▶ 13:01)
  • World Labs sale price: 8000000000 currency unstated (self-reported, currency unstated) (one of the hosts · Adrian (Oct 2026) ▶ 13:03)
  • Share of the OpenRouter deal that went to investors: 6 billion of 7.5 billion, currency unstated (host’s framing, currency unstated) (Jess · Stripe Buys OpenRouter, OpenAI (Aug 2026) ▶ 23:59)

Caveats & warnings

  • Host has a stake: Slow was an Allbirds investor and sold earlier. (Sam · Allbirds Pivoted to AI Data Ce (Apr 2026) ▶ 09:08)
  • Host has a stake: Sam says he was an investor in Airtable. (Sam · Peter Deng & Rachel Holt (Aug 2026) ▶ 04:59)
  • Sam says he has not read the Airtable deal documents and has not spoken to Howie. (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 24:25)
  • Sam admits he does not know the details of the Airtable deal. (Sam · Jeff Dean Leaves Google, Airta (Aug 2026) ▶ 21:50)

How much to trust this

  • Evidence: 9 episodes, 4 independent speakers (the four hosts and unidentified host speech count as one voice, the show; guests counted individually).

Blueprints that use this

Sources

  1. (01)

    Teamshares IPO: $60M EBITDA, 92 Companies, Zero Exits

    YouTube

    47:03 into “Teamshares IPO: $60M EBITDA, 92 Companies, Zero Exits”

  2. (02)

    Allbirds Pivoted to AI Data Centers | SpaceX $1.5T IPO, Anthropic $800B, $175M Seed Rounds

    YouTube

    9:08 into “Allbirds Pivoted to AI Data Centers | SpaceX $1.5T IPO, Anthropic $800B, $175M Seed Rounds”

  3. (03)

    Stripe Buys OpenRouter, OpenAI vs. Anthropic & the AI Backlash

    YouTube

    23:59 into “Stripe Buys OpenRouter, OpenAI vs. Anthropic & the AI Backlash”

  4. (04)

    The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition

    YouTube

    4:59 into “The Risk Of Securitizing GPUs Like Long-Lived Infrastructure & More - A MOL Special Team Edition”

  5. (05)

    SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.

    YouTube

    17:22 into “SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.”

  6. (06)

    Jeff Dean Leaves Google, Airtable Sells, and the Bag-Securing Era

    YouTube

    19:42 into “Jeff Dean Leaves Google, Airtable Sells, and the Bag-Securing Era”

  7. (07)

    Meta Muse vs Instinct vs OpenAI Dots: Do AI Agents Have Any Moat?

    YouTube

    13:01 into “Meta Muse vs Instinct vs OpenAI Dots: Do AI Agents Have Any Moat?”

(By email)

Compounding Insights