01
0:00–4:29Cold OpenINTERLUDE
The Teacup Mammoth Request
Ben Lamm, CEO of Colossal Biosciences (the de-extinction company), jokes about pet mammoths and credits outsider curiosity, mentored by geneticist George Church.
▸ Colossal is "now a company worth over $10 billion"; the "number two request we get is teacup mammoths."
⚠ The valuation checks out: Colossal raised $200M at a $10.2B valuation in January 2025, and 2026 coverage calls it a $10-12B company, so 'over $10 billion' understates if anything. One title fix carried from the tape: George Church is Professor of Genetics at Harvard Medical School and a Colossal co-founder, not head of the department. (techcrunch.com)
Peter Diamandis, Ben Lamm
02
4:30–9:48AI Meets the LabVERIFIED
The Dataset Beats the Model
Why AI chatbots can't do biology yet: published experiments often fail to repeat, so Colossal bets its genome library matters more than any model.
▸ Rerunning published mouse experiments gives "somewhere between 40 and 60% failure rate on published work."
✓ Verified, and if anything conservative: Amgen reproduced only 6 of 53 landmark cancer papers (Begley & Ellis, Nature 2012), and the Reproducibility Project: Cancer Biology found roughly half of experiments failed to replicate. The 40-60% band sits inside the published range. (nature.com)
Dave Blundin, Ben Lamm
03
9:49–14:40Resurrection LimitsVERIFIED
Fewer Than 100 T-Rexes
De-extinction's hard truth: almost nothing fossilizes. Lamm says engineering new life is likelier than reviving every creature — or every human — that ever lived.
▸ "There's less than 100 T-Rexes that have ever been found," though billions allegedly lived on Earth.
✓ Both halves verified: documented T. rex specimens number roughly 50-100, and Marshall et al. (Science, 2021) estimated about 2.5 billion T. rex individuals ever lived. (science.org)
Alex Wissner-Gross, Ben Lamm, Peter Diamandis
04
14:40–19:21Playing GodsHOST TAKE
Who Decides What Returns
From Pikachu requests to Colossal's dire wolf pups — gray wolves carrying 20 edits, per independent scientists: Colossal weighs ecology, indigenous ties, and pop culture — wielding what Lamm calls history's most powerful technology.
▸ Lamm: synthetic biology intersecting with compute, AI, and quantum "will be more powerful than any weapon system that's ever been created."
⚠ The weapon-system comparison is the guest's own forecast, not a checkable claim. The dire wolf context carries a correction: the pups (Romulus, Remus, Khaleesi, announced April 2025) are real, but independent scientists describe them as gray wolves carrying 20 edits across 14 genes — look-alikes, not an uncontested de-extinction, and no peer-reviewed paper accompanied the announcement. (science.org)
Peter Diamandis, Alex Wissner-Gross, Ben Lamm, Moonshot co-hosts
05
▶ ANIMATED
19:21–24:11The Biological SingularityBOLD CLAIM
A Thousand Edits Coming
The curve goes vertical, by Lamm's own count: gene edits jumped from 20 to 300-plus, now testing 1,000 — and he predicts mammals born entirely in artificial wombs within two years.
▸ "We are delivering 300 plus [edits] at 90 plus percent efficiency, consistently... We're now testing 1,000"; Colossal will birth mammals fully ex utero "within 24 months."
⚠ Self-reported: the 20-edit baseline is anchored (the dire wolf record, April 2025), but the 300+ edits at 90%+ efficiency and the 1,000-edit tests appear nowhere outside Lamm's own statements — no paper, no preprint, no independent replication. The ex utero birth is a dated founder prediction, not a capability: no mammal has ever been gestated birth-to-delivery outside a womb. Lamm himself adds the caveat that the curve 'doesn't mean it's going to continue.' (science.org)
Peter Diamandis, Ben Lamm, Dave Blundin
06
24:12–27:50Genotype to PhenotypeVERIFIED
Mouse Into Wolf
Mapping genes to traits: a marsupial mouse must scale 1,500-fold toward the Tasmanian tiger, while Colossal's woolly mice needed only eight edits.
▸ The thylacine's model species, the fat-tailed dunnart, is "a 1,500x fold from a marsupial mouse to a marsupial wolf"; the woolly mice took eight edits in one delivery.
✓ Verified on both counts: the fat-tailed dunnart is Colossal's documented thylacine model species, and a ~15g dunnart to a ~15-30kg thylacine makes 1,500x a fair middle. The Woolly Mouse (announced March 2025) carried eight edits across seven genes in one delivery — though released by press release and preprint, not peer review, and it demonstrates editing capability, not mammoth biology. (biospace.com)
Ben Lamm, Dave Blundin, Peter Diamandis
07
▶ ANIMATED
27:51–31:30Herds, Not SpecimensBOLD CLAIM
Tens of Thousands of Mammoths
On the ethics of lab-born animals: Lamm says 16 mammoth lines run in parallel, herds are the goal, and elephant orphanages teach Colossal how to raise them.
▸ Colossal has "16 different lines being worked on at the same time" and wants to bring back "tens of thousands" of mammoths.
⚠ Self-reported: no outside record of 16 engineered mammoth lines exists — the number first surfaces on this tape — and 'tens of thousands' is aspiration, not plan-of-record. The Colossal Foundation behind the conservation framing is real ($50M launch, announced doubling to $100M), but its partner count and free-licensing terms are also the company's own framing. (company-reported; no independent source)
Peter Diamandis, Ben Lamm, audience questions
08
31:31–36:10Mammoth Climate EngineEARLY SIGNAL
Grass, Albedo, Permafrost
George Church's contested climate case: mammoths fell trees, grassland stores more carbon and reflects sunlight, cooling permafrost — the tape's cooling figures are an internal model; the published model says ~2-4°C. Plus confirmed AI spinout Astromech ($20M at a $3.8B mark).
▸ The tape says grass is "about six times more efficient" at carbon with "a two to three X albedo effect" and claims a larger summer permafrost cooling — the record shows the published model (Beer et al., 2020) at roughly 2-4°C soil cooling in high-density scenarios, mostly via winter snow trampling; the tape's multiples match no published number.
⚠ The underlying mammoth-steppe hypothesis is real, published science (Zimov; Beer et al., Scientific Reports 2020), and grassland albedo does exceed boreal forest. But the 6x carbon figure, the 2-3x albedo multiple, and the tape's summer-cooling number are an internal Colossal/Church modeling exercise — Lamm himself notes scientists inside Colossal dispute the magnitude. Screen the published ~2-4°C figure, labeled as a model. (nature.com)
Dave Blundin, Ben Lamm, Peter Diamandis
09
▶ ANIMATED
36:11–42:41Back Up LifeHOST TAKE
Backing Up the Biosphere
The unfunded moonshot: the tape says half of species in 25 years — the record shows IPBES's ~1 million species threatened within decades. Colossal is building open biovaults with governments, including a confirmed US Fish and Wildlife partnership — and stays optimistic on AI.
▸ The tape says "we are going to lose half of biodiversity in the next 25 years"; the record shows IPBES (2019) estimating ~1 million species threatened with extinction within decades — the 'half' projections in the literature run to 2100. Colossal announced a BioVault biobanking partnership with US Fish and Wildlife (DOI MOU, June 2026).
⚠ The headline number is dramatized: no published consensus says half of biodiversity in 25 years — IPBES puts ~1 million species at threat 'within decades,' and the 'half' figures in the literature run to 2100. The USFWS partnership is real on a federal primary (DOI press release, June 25, 2026, a biobanking MOU for ESA-listed species), though it was announced ~3 months before taping, not 'just.' (ipbes.net)
Peter Diamandis, Ben Lamm, audience