Why This Billionaire Is Bringing Back the Dire Wolf, Woolly Mammoth & More w/ Ben Lamm & Salim | 165
Summary
- The dire-wolf launch is best read as a platform proof, not a settled taxonomy verdict. Colossal reconstructed a nearly complete genome from 72,000-year-old skull DNA and 13,000-year-old tooth DNA, versus just 0.15% coverage previously, then introduced 20 edits—including 15 ancient variants—into gray-wolf cells. Lamm accepts “dire wolf,” “Colossal dire wolf,” or “genetically modified gray wolf,” but argues the irreducible achievement is producing healthy animals with phenotypes driven by “genes that have been lost for 12,000 years.”
- Colossal claims its moat is an integrated software-biology stack that makes increasingly complex edits repeatable. Lamm describes the company as roughly half software and half biology: its mouse demonstration made eight edits across seven genes simultaneously with reported 100% delivery, 100% efficiency, and zero off-target effects, while one nucleotide change gave dunnarts 5,000-fold resistance to cane-toad toxin. “We just read the code and then we rewrite the code.”
- Artificial wombs could become the largest conservation unlock in the portfolio. A 17-person team is targeting the first fully ex-utero mammal birth—a small elephant—by the end of 2026, with a longer-term vision of producing 200 genetically diverse northern white rhinos without burdening living surrogates. Lamm’s deliberately hedged conclusion: “I think we could save every species on the planet.”
- The $10.2 billion valuation rests on three prospective businesses rather than animal spectacle alone. Colossal expects technology spinouts, biodiversity or “nature credits” that turn rewilded populations into multiplying annuities, and paid government recovery programs built atop tools it open-sources for conservation. Lamm cited one prospective recovery effort costing a government about $300 million over 23 years that Colossal believes it could execute for roughly $70 million in under three years.
- The next catalysts form a “three-horse race” among the mammoth, thylacine, and dodo. The mammoth remains targeted for the end of 2028, but the thylacine’s 13.5-day gestation and rapidly improving editing—Colossal has made 300 edits in a dunnart cell—could move it ahead; solving primordial germ cells could similarly accelerate the egg-based dodo program. Genuine dinosaurs remain outside reach because no dinosaur DNA survives, though Lamm speculates that an ancestral archosaur-like reconstruction might become technically conceivable in eight to 10 years at well above $100 million.
- Commercialization is intended to follow rewilding, not conventional zoo exhibition. Lamm expects animals to return through preserves, governments, Indigenous partners, and private landowners, potentially supported by ecotourism and remote viewing; his preferred formulation is to “put the science on display, not animals on display.” Four red wolves were also cloned with a new, less-invasive technique, a conservation result he says was largely buried beneath the dire-wolf naming fight.
- The broader optionality is a “CAD software of biology” spanning agriculture, materials, medicine, and habitats. Examples discussed include hornless cattle, drought-resistant organisms, a microbe engineered to break plastic’s chemical bonds into biomass, p53-related cancer work, epigenetic resetting, self-healing structures, and even trees directed to grow into houses. Lamm estimates editable biology could become a five-year problem with concentrated funding, versus perhaps 10 years otherwise, and calls AI “the biggest accelerant.”
- Execution and governance risk were as visible as the science during launch week. A broken media embargo forced Colossal to publish an unfinished rollout, blindsided advisers, and allowed taxonomy and politics to dominate before its scientific package was digested; Diamandis’s enthusiastic framing should also be read alongside his disclosure that he is a seed investor and adviser. Lamm’s operating answer is a 50-year horizon, strict exclusions around humans and non-human primates, and engagement across administrations: “There’s a difference between empowering and educating.”
Deep dive
1. Colossal’s scientific claim survived a semantic fight
Lamm accepted multiple descriptions—dire wolf, Colossal dire wolf, or genetically modified gray wolf—because his non-negotiable claim concerns function: healthy animals now express physical attributes driven by ancient variants absent for roughly 12,000 years.
The team extracted DNA from a 72,000-year-old skull and a 13,000-year-old tooth, moved from approximately 0.15% prior genome coverage to a nearly complete reconstruction, and compared it with the gray wolf, described as the closest living relative, 99.5% genetically similar, and a non-model species for this work.
Academic criticism focused heavily on what constitutes a species; Lamm noted that roughly 31 classification frameworks can produce inconsistent answers. His frustration was that the naming dispute eclipsed “the incredible work by the women and men at Colossal and our academic contributors.”
A second overlooked result was four cloned red wolves, produced through a new, less-invasive cloning technique. Lamm said only 15 remain in the wild, yet the conservation achievement received little attention beside the three headline-making dire-wolf pups.
2. A non-biologist CEO built the company around teams and time
Ismail noted that de-extinction had been discussed for 10 or 12 years in their circles. Lamm framed execution as rebuilding the entire operating stack, while Diamandis emphasized Lamm’s lack of formal biology training as evidence for “a beginner’s mind” paired with accelerating technology.
Lamm’s first principle is that “entrepreneurship, it’s a team sport”: lacking the scientific background himself, he must trust specialist teams, empower them, and accept that one or two public figures will absorb both the glory and criticism.
His second principle is to be “a little more persistent than time.” Press cycles, including the classic “trough of disillusionment,” matter less when every outcome is measured against Colossal’s intended 50-year horizon: two chaotic launch days are merely “two days in a 50-year journey.”
Founded in September 2021, Colossal reached a stated valuation of $10.2 billion in under four years. Diamandis praised that acceleration while disclosing that he is both a seed investor and an adviser to the company.
3. Ancient DNA is a probabilistic reconstruction problem
Lamm divided de-extinction into three inputs: ancient DNA, the closest living relative, and the engineering tools connecting them. Colossal holds roughly 59 mammoth genomes; its oldest specimen is a 1.2-million-year-old steppe mammoth, near the present practical limit.
More sequencing coverage raises confidence that each genomic position contains, for example, a C rather than a G. The dire-wolf work achieved roughly 13× coverage; Lamm said functional reconstruction might work at 5–6×, while north of 10×—and preferably 20×—provides better results.
Ancient sequencing is destructive, which Lamm compared to a claw machine: “If you don’t get the teddy bear, you still lost your money.” Cold, dry environments preserve material best, while heat, acidification, bacteria, predation, later animals, and contamination complicate identifying endogenous DNA.
This is functional de-extinction, not cloning an extinct animal from a surviving cell: there are no living cells to clone. Once reconstructed, the extinct genome is compared with its living relative to locate coding and regulatory regions likely to drive target phenotypes. Those variants enter donor cells, followed by somatic cell nuclear transfer—modern cloning using robotics and lasers—and implantation into a host.
4. Multiplex editing turns biology into a coding problem
Lamm rejected the popular image of moving ancient DNA with “tiny tweezers.” Colossal reads a biological coding language, then changes individual letters, knocks sequences in or out, or synthesizes a replacement block: “We just read the code and then we rewrite the code.”
In the mouse demonstration, the team simultaneously made eight edits across seven genes, reporting 100% delivery, 100% efficiency, and zero off-target effects. Lamm contrasted that with prior work making comparable edits sequentially over eight generations.
The dire-wolf work extended the approach to 20 edits, 15 involving ancient variants. Where many adjacent changes are required, Colossal may synthesize a whole DNA block, potentially limiting off-target opportunities to its ends instead of creating a risk around every individual edit.
The sharpest conservation example was one nucleotide among roughly 3.5 billion base pairs: matching a variant found in cane-toad-eating species produced dunnarts with 5,000-fold greater toxin resistance. “One letter change could change the entire animal.”
5. Editing scales faster than embryology
Computational biology, sequencing, monoclonal screening, and tool-selection models are already improving across projects. Colossal performs full genotyping on embryos before implantation, spending additional time and money to identify healthy candidates and feed outcomes back into its models.
Embryology remains more species-specific. Lamm stressed that Colossal has not yet created a universal donor egg, but it is exploring one with matched mitochondria that might serve multiple species without separately using stem-cell gametogenesis to produce eggs.
He expects full chromosome synthesis eventually, though “it’s a ways away.” The larger ambition is to replace custom biological architectures with reusable components, just as common computing layers let developers stop redesigning every chip and operating system.
Some organisms expose transferable rules and others do not: Chihuahuas, Great Danes, and wolves scale unusually well, whereas a goldfish cannot simply scale into a killer-whale-sized body. Cell-growth media, gene pathways, placental types, and bird germ cells remain stubbornly bespoke.
6. Artificial wombs could industrialize species recovery
Colossal’s 17-person artificial-womb team is targeting a fully ex-utero mammal birth by the end of 2026. Lamm said the first would be a small elephant before the system scales, making the forecast unusually near-term rather than a distant aspiration.
Colossal says it will not build human artificial wombs, but patents technologies with possible human applications and spins them out. The same model applies to work around p53, which Peter linked to the extra copies found in elephants and whales.
His conservation end state is the production of perhaps 200 genetically diverse northern white rhinos in a laboratory, without having to interfere with another animal or rhino, followed by release through rewilding partners.
Asked when DNA becomes as editable as a word-processing document, Lamm called it “a focus and funding problem” rather than a knowledge barrier: five years with concentrated effort, perhaps 10 without it. AI, compute, and eventually quantum systems would accelerate the work.
7. Synthetic biology expands far beyond extinct animals
Lamm defined synthetic biology broadly as using data, AI, and engineering to change or direct life. Nearer-term applications include drought-resistant plants and animals and hornless cattle that avoid the dehorning required when tightly packed animals injure one another.
Colossal’s spinout Breaking uses synthetic biology to supercharge a microbe that breaks chemical bonds in plastic. Lamm distinguished it from merely eating plastic or generating microplastics: the intended output is biomass.
His longer-range “CAD software of biology” includes curing most disease states, epigenetically resetting cells, and enabling youthful lives for as long as people choose, excepting accidents or “acts of God.” These were presented as visions, not completed capabilities.
The deliberately wild examples were trees engineered to grow into houses, filter water, and support bioluminescent fungi, followed by self-healing underwater cities. Lamm floated the latter as a possible 2090 project, helped by stable temperatures and lower per-kilogram deployment costs than space.
8. Specific dinosaur recovery remains out of reach, but reconstruction might create an analogue
Lamm’s categorical present-tense answer was that no dinosaur DNA exists; amber is porous, and heat, degradation, and fossilization prevent usable genetic material from surviving. Birds preserve evolutionary inheritance, but not a recoverable genome for cloning a specific extinct dinosaur. He also avoids declaring things impossible in principle.
Pressed by Diamandis to propose a theoretical route, Lamm suggested that perhaps eight to 10 years from now—or over some other period—researchers might sequence broadly across birds and reptiles, reconstruct an ancestral state near the base of the tree—an archosaur-like animal—and engineer selected dinosaur-associated phenotypes.
He was equally clear about scale: such an attempt would be “quite a lot more” than a $100 million project. He said he did not know whether the result would be less of a dinosaur than a dinosaur that existed. Amino acids recovered from demineralized bones could help identify a fossil, but not supply instructions for rebuilding its animal.
Colossal has discussed roughly 20–25 extinct species. Active priorities remain the mammoth, thylacine, and dodo, while Lamm’s favorite prospective project, Steller’s sea cow, cannot currently be grown in available systems; artificial-womb work would be needed.
9. The mammoth’s 2028 target faces faster-moving rivals
Colossal remains “on target for the mammoth by the end of 2028,” according to Lamm, but an elephant’s 22-month gestation makes iteration slower even when editing succeeds.
The thylacine could arrive first if editing continues at its current rate: Colossal has already made 300 edits in a dunnart cell, and Lamm cited a thylacine gestation of only 13.5 days.
The dodo offers a different shortcut because development occurs in a self-contained egg, using genetically modified chickens as surrogates with roughly 30-day cycles. The bottleneck is culturing pigeon primordial germ cells, something Lamm said has not yet been achieved.
His honest non-answer was therefore a “three-horse race.” The mammoth has the public target date, while breakthroughs in editing or bird germ-cell media could allow the thylacine or dodo to overtake it.
10. Three revenue engines underpin the valuation
Lamm recalled the original pitch as: George Church believes he can restore a mammoth, and “we’re pretty sure that we make money somewhere in there, but we don’t know.” The model has since separated into technology, nature-linked finance, and implementation services.
Public spinouts include computational-biology company Form Bio and plastic-focused Breaking. A third, still-undisclosed company was already valued above $100 million at seed, while another embryology project could eventually serve IVF clinics.
The second engine combines biodiversity, nature, and carbon markets into what Lamm said will probably be called “nature credits.” He imagines rewilded animals as annuities whose ecological contribution carries measurable value and whose populations can multiply, with extraction companies buying credits for compliance or social commitments.
The third engine keeps conservation tools open-source while charging institutions that want Colossal to implement them. Alongside a $50 million foundation, Lamm cited a government program that might fall from $300 million and 23 years to about $70 million and under three years.
11. Rewilding takes precedence over putting animals on display
Asked when visitors might see a dire wolf or mammoth at the Los Angeles Zoo, Lamm said an ecological preserve in the species’ natural environment is more likely. Releases would involve governments, Indigenous groups, private landowners, and ecosystem-restoration planning.
He conceded Diamandis’s point that good zoos inspire children and can increase concern for animals; Colossal has therefore not closed the exhibition question entirely. Its default, however, is habitat restoration rather than conventional zoo exhibition.
Possible compromises include Tasmanian ecotourism after a genetically diverse thylacine population is established, remote wildlife cameras, documentaries, and educational programming. The guiding distinction is to “put the science on display, not animals on display.”
12. Ethical boundaries and fewer edits define the risk framework
Colossal does not work on humans and has also drawn a line around non-human primates, rejecting Neanderthal, Gigantopithecus, and “King Kong” requests. Lamm finds smarter companion animals intellectually interesting but warned that uplift becomes darker when extended to livestock.
For invasive feral cats in Australia, he favors gene drives that allow animals to live and reproduce once, but render their offspring unable to produce further offspring, shrinking the population across generations without poisoning or shooting it.
Toxin-resistant northern quolls represent the inverse intervention: protected quolls can keep eating invasive cane toads, recover their own population, and reduce pressure on other marsupials. Lamm’s defense was blunt: humans introduced the disruption, so “we’re geoengineering by default anyway.”
Quantum computing could eventually simulate genotype-to-phenotype effects and identify the fewest necessary interventions, but Lamm considers it too early; sensing is present and communications are “here-ish,” while useful computation remains “two years every two years.” His safety rule is to use as few edits as possible to achieve the desired outcome.
Colossal’s ethicists review each species and increasingly focus on conservation and how to rise above political polarization. After criticism that engaging the current administration conferred legitimacy, Lamm answered, “There’s a difference between empowering and educating”—a 50-year mission must outlast four- and eight-year political cycles.