Anthropic's Wet Lab, Artificial Wombs in 24 Months, and Scalable Gene Editing | MOONSHOTS Live #297
Colossal Biosciences CEO Ben Lamb joins Peter Diamandis, Alex Wissner-Gross, and Dave Blundin to unpack how AI and gene editing are compounding inside a de-extinction company now valued over $10 billion. Lamb reveals Colossal's multiplex base-editing output has grown from about 20 edits a year ago to 300+ at 90%+ efficiency (now testing 1,000-edit batches), and predicts the company will birth a mammal fully ex utero within 24 months. The panel pushes on the philosophy and ethics of de-extinction -- from Russian cosmism's dream of reviving every organism to who decides which species get brought back -- and on Colossal's new data-centric spinout Astromech. Lamb also calls Anthropic's newly announced CRISPR-like wet lab "massively validating" rather than threatening, and names unchecked biodiversity loss (half of it projected gone within 25 years) as the biggest moonshot nobody is funding at government scale.
Ben Lamb explains why having no synthetic-biology background helped him build Colossal into a $10B+ de-extinction company, leaning on experts like George Church.
LLMs are useful today mostly as 'middleware' bridging inconsistent biology literature and nomenclature; simulation-driven lab science is still a few years out and wet-lab validation remains required.
Global biovault / genome sequencing initiativeBiotech▶ 8:12
Colossal is building a 'Noah's Ark 2.0' network of government-backed bio vaults doing tip-to-tip genome sequencing, believing the resulting dataset is more valuable than any AI model.
Alex Wissner-Gross frames de-extinction against 19th-century Russian cosmism's goal of reviving every organism that ever lived; Ben Lamb says full resurrection is implausible, and the real ethical question -- which species to revive, and who decides -- weighs ecological role, cause of extinction, and indigenous spiritual ties.
Gene-editing and DNA-synthesis scaling curveBiotech▶ 20:28
Colossal's multiplex base-editing output jumped from ~20 edits a year ago to 300+ at 90%+ efficiency, now testing 1,000-edit deliveries; Ben Lamb expects DNA synthesis to eventually overtake multiplex editing.
Artificial womb / ex utero gestationBiotech▶ 22:05
Ben Lamb predicts Colossal will birth a mammal fully ex utero -- gestation through delivery, never in a surrogate -- within 24 months, while a human artificial womb remains an ethics/regulatory question, not a technical one.
Genotype-to-phenotype prediction across cladesBiotech▶ 24:51
Using the Tasmanian tiger/fat-tailed dunnart size gap (1500x) as a model, Colossal is trying to extrapolate trait-engineering rules (size, coat, stripes, hair) across whole animal clades rather than one species at a time.
Ex utero rearing ethics & elephant conservation workBiotech▶ 28:48
Responding to an audience question about the emotional needs of ex utero-born mammals, Ben Lamb describes Colossal's open-source conservation foundation and its Elephant Havens work rearing orphaned elephants into social herds.
Mammoth rewilding and climate impactBiotech▶ 31:29
Dave Blundin recounts George Church's thesis that mammoth-driven grassland (vs. forest) sequesters more carbon and raises albedo, potentially cooling permafrost 6-8 degrees -- though Colossal itself is split on how big that climate effect really is.
Colossal's new spinout Astromech builds narrow 'inflection models' on evolutionary genome data meant to plug into and direct giant foundation models rather than compete with them.
Biodiversity-loss crisis and biobankingBiotech▶ 36:19
Ben Lamb calls the projected loss of half of Earth's biodiversity in 25 years the biggest unaddressed moonshot, pushing for federal biobanking infrastructure, and cites Colossal's LifeBank USA placental stem-cell storage and new US Fish & Wildlife partnership.
Asked whether Anthropic's newly announced CRISPR-like wet lab threatens Colossal, Ben Lamb calls it 'massively validating' for biotech rather than a competitive threat, since the field has barely begun.
Predictions made
openBen Lamb: Colossal will birth a mammal fully ex utero -- gestation through delivery -- that never spent time in a surrogate.
“I think that world is less intense. Less than 10 years out.”
Your call:
openBen Lamb: Engineering key phenotypes in base organisms across different clades, synthesizing or multiplex-editing them, and growing them ex utero will be achievable.
“the technologies of being able to engineer key phenotypes on base level organisms in different clades... and then grow them ex utero is within a decade.”
Your call:
openBen Lamb: Colossal will publicly reveal another de-extincted species beyond the dire wolves and woolly mammoth.
“we are going to lose half of biodiversity in the next 25 years.”
Your call:
Numbers that matter
Colossal is now valued at over $10 billionDespite founder Ben Lamb having no prior synthetic-biology background.
40-60% of published biology experiments fail to replicateDue to inconsistent gene naming/classification across papers, especially in mouse studies.
Fewer than 100 T. rex fossils have ever been foundIllustrates how few extinct species left a usable fossil record, limiting true de-extinction.
Base-edit delivery rose from ~20 edits a year ago to 300+ edits at 90%+ efficiency, now testing 1,000-edit batchesColossal's multiplex gene-editing scaling curve, described as tripling year over year.
1500x size difference between a marsupial mouse (fat-tailed dunnart) and the Tasmanian tiger model speciesUsed as Colossal's test case for extrapolating size-engineering rules across animal clades.
Colossal's woolly mice required just 8 gene edits in a single deliveryFirst-generation public demo of multiplex base editing.
Colossal has 75 global conservation partners using its tech for freeVia Colossal's open-source conservation foundation.
Colossal is running 16 parallel genetic lines aiming to eventually bring back tens of thousands of mammothsBecause mammoths are highly social animals, Colossal is engineering herds, not individuals.
Grassland sequesters carbon about 6x more efficiently than forest and has a 2-3x albedo effectGeorge Church's rationale for why mammoth-driven deforestation of the tundra could help cool the climate.
Modeled mammoth density could lower Arctic permafrost temperature by 6-8 degreesColossal's internal extrapolation of George Church's mammoth-rewilding climate thesis; disputed internally.
Half of global biodiversity projected to be lost in the next 25 yearsBen Lamb's answer to 'biggest moonshot nobody's working on.'
Worth digging into
🕳️ Russian cosmism and Nikolai Fyodorov's 'common task'
Ties today's de-extinction/biobanking ambitions to a 19th-century Russian philosophical movement arguing humanity's ultimate purpose is to revive every being who ever lived -- worth understanding as the intellectual lineage behind Colossal's framing.
Colossal's 1500x marsupial-mouse-to-Tasmanian-tiger sizing model hints at a generalizable 'CAD for biology' approach to trait engineering across whole animal groups, not just species-by-species.
🕳️ Base-editing delivery scaling curve (20 to 300+ to testing 1,000 edits)
If Colossal's tripling-year-over-year curve holds, large-scale multiplex genome engineering compounds similarly to compute, which would make ambitious de-extinction and ex utero gestation timelines far more credible sooner.
🕳️ Anthropic's CRISPR-like wet lab
A major AI lab running physical biology experiments blurs the AI-company/biotech-company line and could reshape competitive dynamics in gene editing and drug discovery.
🕳️ Mammoth rewilding as a climate-engineering lever
George Church's thesis -- that restoring megafauna grazing could cool Arctic permafrost 6-8 degrees via grassland albedo and carbon sequestration -- is a contested but potentially consequential geoengineering idea that Colossal itself is internally split on.