Elon Musk on the Joe Rogan Experience #1169, fact-checked

What the evidence shows: Tunneling and earthquake engineering literature broadly supports the premise that underground structures, including tunnels, generally suffer less seismic damage than surface buildings, because they move with the surrounding ground rather than oscillating independently, avoiding the resonant amplification effects that damage above-ground structures; this is the accepted engineering basis for calling tunnels comparatively low-risk in earthquakes. However, "earthquakes are essentially a surface phenomenon" overstates the case: seismic shaking and its damage potential are not confined to the surface, and tunnels and underground stations have suffered severe, sometimes catastrophic, damage in major earthquakes, most notably the 1995 Kobe (Hyogoken-Nanbu) earthquake, when the Daikai subway station's central columns collapsed, causing up to 2.5 meters of surface subsidence, with additional tunnel and station damage occurring elsewhere in the system. Similarly significant tunnel damage occurred in the 1999 Chi-Chi (Taiwan) and 2008 Wenchuan (China) earthquakes, particularly where tunnels were shallow, crossed an active fault rupture, or sat in poor or liquefiable soil. Current engineering consensus therefore treats tunnels as generally lower-risk than surface structures rather than immune, with fault-crossing, shallow depth, and portal/entrance zones identified as continuing points of vulnerability requiring specific seismic design. The claim is best characterized as a misleading oversimplification: broadly consistent with real engineering advantages of underground construction, but its blanket framing elides well-documented cases of serious tunnel and subway damage in large earthquakes.

What the evidence shows: Musk's description of flexible, self-sealing tunnel segments rated to five atmospheres of pressure is a company engineering specification that has not been independently tested or certified by a third party in public reporting; no peer-reviewed or regulatory engineering assessment of the segment joint design or pressure rating could be located. Since this 2018-era claim, the Boring Company's operating tunnels (notably the Las Vegas Convention Center Loop and expanding Vegas Loop) have drawn sustained scrutiny from Nevada lawmakers and regulators over safety and environmental compliance, including nearly 800 alleged environmental violations reported by ProPublica and 17 complaints filed with Nevada OSHA between 2020 and 2026, though these relate to operational and environmental practices rather than a direct test of the segment/seal engineering Musk described. The company's Nashville Music City Loop project has separately drawn a metro council resolution citing safety, transparency, and geological concerns (the area's porous limestone and flood history), again without directly confirming or refuting the segment engineering claims. No source was found confirming or refuting the specific "five atmospheres" pressure rating or the "underground snake" flexible-joint description through independent structural testing. Status: unsupported/unverified, resting solely on the company's own unaudited claims.

What the evidence shows: In context, Musk describes this as a permit for a pit dug in the company's own parking lot ("it was a pit in our parking lot"), i.e. an informal prototype pit on SpaceX/Boring Company property around late 2016, distinct from the Boring Company's formal city permit for its 1.14-mile Hawthorne test tunnel, which the Hawthorne City Council did not approve until August 2017 after a months-long review. No independent news report, government record, or other third-party source documents the exact hours or days it took to dig this specific parking-lot pit; the only account of that timeframe is Musk's own retelling, which is internally inconsistent even within this same podcast answer (two to three days, then '48 hours, something like that'). Independent reporting does corroborate that the Boring Company has a documented pattern of moving unusually fast and securing expedited or minimal government approval compared with typical infrastructure projects, including more recent tunnel projects that began construction before completing standard permitting and environmental review. Because the specific claim rests solely on the speaker's own uncorroborated recollection of an informal, unrecorded event, it cannot be independently verified, though it is broadly consistent with the company's well-documented operating pattern. Status: unsupported by independent sourcing, though plausible given the company's track record.

What the evidence shows: The regulatory history is more layered than the claim suggests: federal law required lap belts in front seats of new US cars starting with the 1968 model year, but mandatory seatbelt-use laws (requiring occupants to actually buckle up) did not begin until New York's 1984 statute, decades after the mortality benefits of belts were being documented; about 30 states had such laws by 2010. A related and longer fight concerned passive restraints (airbags), which automakers resisted for years even after mortality benefits were identified in research as early as 1976; a federal mandate requiring front airbags in all new vehicles did not take effect until 1998, over two decades after that evidence emerged. Crash data reviewed since then shows seatbelts reduce front-occupant mortality by roughly 51-72%, and by up to about 80% combined with airbags, underscoring that the safety case was strong well before full regulatory adoption. The claim is directionally accurate about a long delay between clear evidence and regulation, but "more than a decade" understates the passive-restraint/airbag fight (which ran over 20 years), while the initial seatbelt-installation mandate itself arrived comparatively quickly (by 1968); the precise timeline depends on which specific regulatory fight is meant.

What the evidence shows: Musk framed Neuralink's mission as building a high-bandwidth brain-computer interface (BCI) for human-AI symbiosis, and in this September 2018 podcast said Neuralink would have a major announcement 'in a few months.' Neuralink's first human implant did not occur until January 29, 2024, more than five years after this claim, well beyond the near-term timeline Musk implied. Outside scientists quoted after the first human case described it as an early, promising step but cautioned that Neuralink had not released performance data, so it was not yet a demonstrated technical leap. The long-run vision of BCIs enabling human-AI symbiosis remains an active but unproven research goal rather than an established capability. Neuralink has made real, verifiable progress toward implanted BCIs since 2018, but the specific near-term timeline Musk described was significantly exaggerated relative to what materialized.

What the evidence shows: Musk did publicly warn state leaders about AI risk at the National Governors Association's July 2017 summer meeting in Providence, Rhode Island, telling governors AI poses a "fundamental risk to the existence of human civilization." However, NGA's own accounting of that meeting states it brought together more than 30 governors, not all 50, so describing it as "a meeting of all 50 governors" overstates actual attendance. As for meeting "with Congress," the best-documented instance of Musk appearing before US lawmakers on AI is the September 2023 closed-door Senate AI Insight Forum, which occurred five years after this September 2018 podcast episode aired and therefore cannot be what Musk was referring to at the time of this claim; no independently reported Musk meeting with Congress on AI predating this episode was found. The 2017 governors' warning itself is well documented and accurately described in substance, but the claims of meeting "all 50 governors" and meeting "with Congress" (as of 2018) are unsupported or overstated based on available reporting.

What the evidence shows: Musk is restating philosopher Nick Bostrom's 2003 "simulation argument," which is a trilemma, not a proof that a simulation is near-certain: it holds that at least one of three propositions must be true, (1) civilizations almost always go extinct before becoming capable of running ancestor simulations, (2) advanced civilizations are unlikely to bother running such simulations, or (3) we are almost certainly in a simulation. Musk's framing, both in this 2019 conversation and in earlier public remarks ("the odds that we are in base reality is one in billions"), assumes the first two propositions are false without justifying that assumption, which is the weakest link in the argument according to philosophers and astronomers who have examined it. A 2024 Bayesian analysis by Columbia astronomer David Kipping formalizing the trilemma found the probability that we are living in base (non-simulated) reality to be roughly 50-50, not the near-certainty Musk asserts, and noted that Musk's confident claim only follows if propositions one and two are assumed false, an assumption he does not defend. The underlying premise remains a widely discussed but empirically untestable philosophical thought experiment rather than an established scientific finding, and no consensus exists among physicists or philosophers that we are, or are likely to be, living in a simulation.

What the evidence shows: Musk stated flatly that a large, building-sized machine to remove carbon dioxide from the atmosphere is "not possible." Direct air capture (DAC) technology of exactly this kind already existed at the time of this 2018 episode (Climeworks had operated commercial DAC plants in Switzerland since 2017) and has since been built and scaled substantially: Climeworks' Mammoth plant in Iceland, rated at roughly 36,000 metric tons of CO2 per year, began operating in 2024, and the U.S. Department of Energy has funded regional "DAC Hubs" intended to demonstrate commercial-scale facilities capable of capturing at least 1 million metric tons of CO2 annually, including projects in Texas and Louisiana backed by $3.5 billion in DOE funding announced in December 2022 and up to an additional $1.8 billion announced in December 2024. Independent peer-reviewed modeling published in Nature Communications treats DAC as a technically viable but still nascent industry, noting that near-term deployment is likely to remain at the megaton scale absent aggressive early capacity expansion, with gigaton-scale deployment considered plausible only under rapid growth and strong policy support. The claim that such a filter is technically impossible is contradicted by operating facilities and government-funded commercial-scale projects; the more accurate characterization is that DAC is technically possible but currently expensive, energy-intensive, and limited in deployed scale relative to global emissions.

What the evidence shows: Musk claimed that even if 100% of new car and truck production were instantly switched to electric, replacing the existing global transport fleet would take about 25 years. Peer-reviewed fleet-turnover research supports the general shape of this claim while suggesting the number is a rough approximation rather than a precisely sourced figure. An empirical survival-rate study of 31 European countries finds mean car lifespans of 18.1 years in Western Europe to 28.4 years in Eastern Europe, and separate U.S. fleet-emissions modeling shows that even an aggressive 50%-EV-sales target by 2030 only reduces light-duty fleet emissions by roughly 25% by 2030 and 45% by 2035, underscoring that full fleet turnover genuinely spans decades. No study cited here pins the global average at exactly 25 years; commonly cited average vehicle lifespans (roughly 15-20 years in wealthier markets, longer via secondhand export in developing markets) suggest the true global average could be somewhat lower or higher depending on region and vehicle-export patterns. Musk's underlying point, that annual new-vehicle production of roughly 100 million units is small relative to the global fleet of over 1 billion vehicles, making an instant fleet-wide transition to electric impossible in the near term, is well supported by the arithmetic and by fleet-turnover literature. The specific "25 years" figure is directionally reasonable and in the right order of magnitude but should be treated as an illustrative estimate rather than a rigorously documented statistic.

What the evidence shows: Multiple independent studies of the peer-reviewed climate literature and surveys of climate scientists find that roughly 97% to over 99% of publishing climate scientists agree that recent warming is human-caused and poses significant risk, a level NASA and other scientific bodies describe as an overwhelming consensus. That figure means a small minority of credentialed scientists have published dissenting or skeptical views on the severity, causes, or framing of climate risk, so the claim that "literally zero" serious scientists disagree overstates the actual unanimity, even though the consensus itself is real and very strong. The core substance of Musk's statement, that mainstream science recognizes climate change as a serious risk, is well-supported; the absolute-zero framing of dissent is an exaggeration of otherwise accurate consensus figures.

What the evidence shows: Tesla unveiled the second-generation Roadster in November 2017 with a claimed 620-mile range from a 200-kWh battery pack and a stated production start of 2020. As of Tesla's most recent annual reports on file with the SEC, the Roadster has not entered production: Tesla's 10-K for fiscal year 2023 (filed January 2024) lists 'Tesla Roadster' status as 'In development' in its vehicle-capacity table, and the 10-K for fiscal year 2024 (filed January 2025) again lists it as 'In development,' with no production capacity assigned. No production vehicle has shipped to customers, so the 600-620-mile range figure remains an unverified manufacturer projection rather than a confirmed, tested specification. The claim is therefore best characterized as unresolved and substantially delayed rather than false or confirmed: Tesla has repeatedly pushed back the Roadster's production and unveiling timeline for multiple years past the original 2020 target with no independently verified range figure to date.

What the evidence shows: Musk's claim echoes Tesla's own October 2018 blog post, which used NHTSA's published crash-test data to calculate a 'Vehicle Safety Score' and argued the Model S, X, and 3 held the three lowest probability-of-injury scores of any vehicle NHTSA had tested. NHTSA disputed the framing: its Chief Counsel sent Tesla a cease-and-desist letter calling the claim misleading, noting NHTSA's public rating system stops at 5 stars and does not itself rank vehicles beyond that, and referred the matter to the FTC's Bureau of Consumer Protection to assess whether the statements were an unfair or deceptive practice. Tesla did not retract the claim and maintained its statements were based on 'actual test results and NHTSA's own calculations for determining relative risk of injury and probability of injury.' NHTSA did not allege the underlying arithmetic was fabricated, only that presenting a derived, unofficial ranking this way could mislead consumers since NHTSA does not endorse cross-model probability-of-injury comparisons as an official ranking. So the claim is a real (Tesla-derived, NHTSA-data-based) calculation that the regulator itself publicly called misleading and referred for further review, rather than a claim with either clean confirmation or outright debunking on the numbers.

What the evidence shows: Rogan is describing the real 2008-2009 'Kids for Cash' scandal in Luzerne County, Pennsylvania, in which judges Mark Ciavarella and Michael Conahan received kickbacks from the builder and co-owner of two private, for-profit juvenile detention facilities in exchange for closing the county-run juvenile center and steering youth offenders into the private facilities. NPR reported at the time of Ciavarella's 2011 sentencing that he 'was convicted of taking $1 million in bribes from developers of juvenile detention centers' and then presiding over cases that sent juveniles to those same centers, closely matching Rogan's 'a million dollar payment' framing. Federal prosecutors and the DOJ separately described the scheme, across both judges, as involving 'millions in bribes and kickbacks,' with other reporting citing a combined total closer to $2.6-2.8 million; Ciavarella's own court-ordered restitution came to roughly $1.17 million ($965,930 plus $207,861 in tax-related restitution). Conahan pleaded guilty to racketeering conspiracy and received a 17-year sentence; Ciavarella was convicted at trial on 12 of 39 counts and sentenced to 28 years, a sentence upheld on appeal and again after a 2020 resentencing challenge. The core claim, that judges were paid to sentence juveniles to private detention centers, is well-supported, and the 'million dollar payment' figure Rogan cites matches at least one contemporaneous NPR characterization of Ciavarella's individual take almost exactly, even though other sources describe a larger combined bribe total across both judges.

What the evidence shows: Musk has repeated this claim publicly since at least 2018, typically framing it around major, high-volume American automakers. Among the "Big Three" legacy automakers, General Motors filed for Chapter 11 bankruptcy protection on June 1, 2009, and Chrysler filed for Chapter 11 on April 30, 2009, both restructuring with U.S. government backing before emerging as new entities; Ford avoided bankruptcy during that crisis by having mortgaged its assets in 2006 to raise cash. This directly contradicts the literal claim that "only" Ford and Tesla have avoided bankruptcy, since it ignores GM and Chrysler as counterexamples of major American automakers, and it further omits smaller historical manufacturers (such as American Motors, Nash, Hudson, and Packard) that ceased independent operations through mergers or wind-downs rather than a formal bankruptcy filing. The claim contains a kernel of truth if narrowly restricted to comparing Ford and Tesla against GM and Chrysler specifically, but as stated it is an overbroad generalization about American car-company history. Current status: misleading.

What the evidence shows: The commonly cited historical figure for Manhattan's horse population is much lower than Musk's claim. The Henry Ford museum states that 'as late as 1900, Manhattan had 130,000 horses,' compared to 74,000 in Chicago, 51,000 in Philadelphia, and 32,000 in St. Louis, roughly 2.3 times fewer than the 300,000 Musk cites. The same source notes that in the 1880s the New York City Sanitation Department was removing about 15,000 dead horses from the street each year, a real historical mortality figure in the same order of magnitude as, but below, the 20,000-per-year estimate Musk derives from his population figure. The broader point, that a large urban horse population created a substantial ongoing disposal problem before automobiles replaced horse-drawn transport, is accurate and well documented; only the specific 300,000 population figure (and the derived annual death estimate built on it) appears overstated.

What the evidence shows: Musk speculated, without citing any study, that deleting social media apps produces roughly a 10% boost in happiness. No published research establishes a specific percentage gain in happiness or wellbeing from quitting or deactivating social media. A 2025 systematic review and meta-analysis of ten social-media-abstinence studies (N=4,674 participants, 38 effect sizes) found no significant effect of abstinence on positive affect, negative affect, or life satisfaction, and no relationship between abstinence duration and these outcomes. A separate systematic review of 16 studies on social media and mental health describes mixed, methodologically limited findings (cross-sectional designs, sampling limitations) rather than a settled quantitative effect. The claim's specific "10%" figure appears to be an off-the-cuff estimate rather than a finding from any known study.