E40: A Bestie gets COVID, Delta breakthrough, Billionaire Space Race & more

The prediction was that, from mid‑2021 onward, essentially all unvaccinated people would eventually be infected with the COVID‑19 Delta variant.

Evidence from seroprevalence and variant‑specific surveillance indicates:

• By late 2021 and early 2022, Delta did become the dominant variant globally for a time and infected a large share of unvaccinated people, but not all. Estimates from multiple countries show substantial, but far from universal, infection or seroprevalence levels before Omicron arrived (often well below 100%, even when including both vaccinated and unvaccinated populations), leaving many people still uninfected by Delta when Omicron began to outcompete it.
• Omicron (and its sublineages) rapidly displaced Delta in late 2021 and 2022, becoming the overwhelmingly dominant variant worldwide. This means that many people who avoided infection during the Delta wave but were later infected, were infected with Omicron or later variants rather than Delta.
• There is no credible epidemiological evidence that nearly all unvaccinated individuals (let alone literally all) were ever infected with Delta specifically; instead, data consistently show a mix of: people infected with pre‑Delta strains, people infected with Delta, people infected only with Omicron or later variants, and some who (as of 2025) still show no evidence of infection.

Because the statement is strong and specific to the Delta variant ("if you're not vaccinated, you're going to get the delta variant"), and available data show large numbers of unvaccinated people who either remained uninfected through the Delta period or were first infected by later variants, the prediction is best classified as wrong.

The episode aired on 16 July 2021. On 19 November 2021, less than five months later, the FDA amended the Emergency Use Authorizations for the Pfizer-BioNTech and Moderna COVID-19 vaccines so that a single booster dose was authorized for all adults 18 and older, rather than just high‑risk groups. (biopharminternational.com) That same day, the CDC director endorsed ACIP’s expanded recommendations, explicitly expanding eligibility and recommending booster shots for all adults ages 18 and older who had completed an mRNA primary series at least six months earlier. (archive.cdc.gov) This represented a clear shift in U.S. policy and practice from targeted boosters to a general‑population booster recommendation within a short period after July 2021, matching Friedberg’s prediction that “we’re going to need to do a booster very soon for everyone.”

The prediction was that mRNA COVID‑19 vaccine effectiveness against infection would decline to roughly two‑thirds (about 60–70%) around six months after the initial 2‑dose series.

Real‑world data show substantially lower effectiveness against infection by about 5–6 months, especially once Delta became dominant:

• A large Kaiser Permanente / Pfizer cohort study of BNT162b2 found effectiveness against any SARS‑CoV‑2 infection fell from 88% in the first month after full vaccination to 47% at six months overall, with Delta-specific effectiveness around 53% by four months and non‑Delta effectiveness about 67% at 4–5 months – already below or barely at the lower edge of the predicted 60–70% band and still declining. (pubmed.ncbi.nlm.nih.gov)
• A nationwide Qatar study found BNT162b2 effectiveness against infection peaked early then dropped rapidly, reaching roughly 20% in months 5–7 after the second dose, far below two‑thirds. (pubmed.ncbi.nlm.nih.gov)
• In England’s OpenSAFELY cohort, by weeks 23–26 (about six months) after the second dose, rates of positive SARS‑CoV‑2 tests in vaccinated people were similar to or higher than in unvaccinated people, implying vaccine effectiveness against infection near zero or even negative at that time. (pubmed.ncbi.nlm.nih.gov)

Earlier phase 3 trial follow‑up for BNT162b2 reported high efficacy (around 84–91%) against symptomatic COVID‑19 through six months, but those data were mostly pre‑Delta and not focused specifically on the six‑month point in a real‑world, variant‑mixed setting. (pubmed.ncbi.nlm.nih.gov) Subsequent observational studies across multiple countries consistently show that by roughly six months post‑series, protection against infection had typically fallen below 60%, often much lower, although protection against severe disease remained high.

Because the actual effectiveness against infection at about six months was generally well under the predicted 60–70% (two‑thirds) level, the prediction overestimated how much protection would remain and is best classified as wrong.

FDA in the United States authorized a Pfizer-BioNTech COVID‑19 booster dose for certain populations on September 22, 2021, with CDC advisory panel backing and CDC director sign‑off on September 23–24, 2021, after which booster shots began being administered nationally.

Sources:

• The FDA issued an Emergency Use Authorization (EUA) for a single Pfizer booster dose for high‑risk groups on Sept. 22, 2021.
• CDC’s ACIP recommended the booster on Sept. 23, 2021, and CDC’s director endorsed and slightly expanded those recommendations shortly thereafter, allowing boosters to begin for eligible groups in late September 2021.

Since boosters were both authorized and available in the U.S. market in September 2021, Friedberg’s prediction that “we’re going to get boosters and … see them hit the market next month in September” is correct for the United States and timeframe specified.

Two key facts:

1. Antecedent didn’t occur (much less than 40% unvaccinated).
   CDC data summarized in August 2023 reporting shows that by May 11, 2023, about 81.4% of the U.S. population had received at least one COVID-19 vaccine dose, i.e., only ~18.6% remained entirely unvaccinated. (frequentbusinesstraveler.com) Later reporting notes that by 2025 about 70% of Americans had completed the initial vaccine series, implying ~30% were unvaccinated or incompletely vaccinated—still well below 40%. (nypost.com) So the condition “if ~40% of Americans remain unvaccinated 2–3 years from mid‑2021” was not met.

2. No new “killer strain” emerged in 2023–2024.
   The dominant variants in that period were Omicron sublineages such as JN.1, KP.3, LB.1, and later NB.1.8.1. Multiple analyses from WHO, CDC, and academic/medical sources consistently report that these subvariants are more transmissible and more immune‑evasive but do not cause more severe disease or higher mortality than prior Omicron strains:

   • JN.1: WHO and CDC data show no evidence of increased severity vs other Omicron variants. (yalemedicine.org)
   • KP.3: Described as more contagious but not associated with more severe illness than recent strains. (health.com)
   • LB.1 and NB.1.8.1: Monitored variants with no evidence of more severe disease or higher hospitalization/death rates than contemporaneous variants. (cbsnews.com)
     Broader reviews also note that post‑Omicron, COVID‑19 has generally become more transmissible but less severe, with lower case‑fatality rates compared with earlier waves like Delta. (lemonde.fr) There was no distinct “highly lethal” new U.S. variant in 2023–2024 that caused catastrophic mortality specifically among the unvaccinated.

Why this is scored as “ambiguous”:
Chamath made a conditional forecast: if ~40% of Americans stayed unvaccinated 2–3 years out, then a “killer strain” would emerge with essentially 100% probability. In reality, the vaccination rate rose high enough that this 40%‑unvaccinated scenario never happened, and pandemic evolution took place under different immunity conditions. Because the stated condition failed, the prediction’s implied causal claim about what would have happened under that counterfactual cannot be directly tested.

If one ignores the conditional and treats his statement as an unconditional claim that a killer strain would appear by ~2023–2024, available evidence would make it wrong; but under the literal conditional phrasing, the fairest scoring is that the prediction’s truth cannot be determined from realized data, hence “ambiguous.”

Multiple data sources show that in the weeks after mid‑July 2021, a substantial share of Americans voluntarily pulled back from travel and dining as the Delta wave and breakthrough reports grew, even though there were no new broad stay‑at‑home lockdowns.

Travel & flying behavior
• A Destination Analysts survey (summarized in a Montana tourism bulletin dated August 13, 2021) reported that 54% of American travelers said the Delta variant made them less interested in traveling right now, with notable increases in trip postponements (27%) and cancellations (23%) compared with late July. Nearly half (48%) agreed that media coverage of COVID had them doubting it was safe to travel, and fewer than half considered flying on an airplane safe. (content.govdelivery.com)
• The same research series found that by early August, optimism about the near‑term pandemic trajectory had crashed about 40% since early June, and travel intent had dropped from its early‑summer highs, which analysts explicitly linked to the Delta surge. (qatar-tribune.com)
• Southwest Airlines, in an August 11, 2021 regulatory update, said that Delta‑related developments had caused negative effects on August and September revenue trends, undermining its earlier expectation of a profitable Q3. The article discussing this cites survey data that more than 54% of Americans said the Delta variant made them less interested in traveling right now, reinforcing that there was a demand‑side, traveler‑driven pullback rather than only regulatory limits. (dallasnews.com)

Restaurant dining behavior
• An early‑August 2021 OpenTable/Booking Holdings release noted that U.S. dining demand was down 13% compared with just one month earlier, with significant declines in cities like Minneapolis, Seattle, and New York. This compares July to early August—precisely the “weeks following mid‑July” window. (bookingholdings.com)
• A QSR Magazine analysis, drawing on National Restaurant Association survey work, reported that, amid Delta’s rise, 19% of adults said they had stopped going out to restaurants and 9% said they had cancelled existing plans in recent weeks because of COVID trends. The same piece cited OpenTable data showing dining down 16% in San Francisco compared with July and much further below 2019 levels—clear evidence of a voluntary pullback. (qsrmagazine.com)

Lack of new formal lockdowns, but rising Delta/breakthrough fears
• By early August 2021, the Delta variant accounted for roughly 99% of U.S. cases and drove a large summer wave, including rising concern about breakthrough infections, but the U.S. did not re‑enter nationwide economic lockdowns. Restrictions were more targeted—e.g., New York City’s proof‑of‑vaccination rule for indoor dining announced August 3 and effective August 16, with enforcement only starting in September. (en.wikipedia.org)

Putting this together: within a few weeks of mid‑July 2021, measurable shares of Americans reported being more hesitant to travel or fly, showed increased postponements/cancellations of trips, and reduced in‑person restaurant dining, in response to the Delta surge and associated fears. These shifts occurred largely through voluntary behavior changes and sentiment‑driven pullbacks rather than broad re‑imposed lockdowns. That matches Friedberg’s prediction that people would “get scared again,” go into a kind of social lockdown, and become nervous about traveling, flying, and going to restaurants. Hence, the prediction is best classified as right.

Multiple surveys from late July and early August 2021 show a clear, rapid increase in hesitancy around discretionary activities like travel and public leisure, especially among the vaccinated, within roughly four weeks of mid‑July 2021.

• Morning Consult’s “Return to Normal” tracking found that Americans’ average comfort with more than 40 leisure activities (shopping malls, vacations, sporting events, movie theaters, etc.) peaked at 55% around July 8–10, then fell every week in July, reaching 49% by the Aug 2–3 survey. Comfort among vaccinated adults dropped more sharply, from a record 55% in early July to 47% by early August, with noted declines in going to malls, restaurants, movie theaters and sporting events.(pro.morningconsult.com) This is squarely inside Friedberg’s “next four weeks” window after mid‑July.
• A Gallup poll conducted July 19–26 found concern about catching COVID-19 rising again (from 17% in June to 29% in July) and a growing share saying the better advice for healthy people was to stay home as much as possible (up from 35% to 41%), indicating more caution about normal out‑of‑home activities as the Delta wave hit.(news.gallup.com)
• For travel specifically, an August Fortune/SurveyMonkey poll reported that 61% of Americans said they were more concerned about traveling because of the Delta variant, with concern especially high among the vaccinated (72% of vaccinated vs. 40% of unvaccinated).(surveymonkey.com)
• A YouGov survey for The Points Guy (fielded Aug 18–20) found that, among people who travel at least occasionally, 49% of fully vaccinated respondents now felt less comfortable taking a domestic flight, 53% less comfortable with an international flight, and 35% less comfortable just traveling to a different city or town, specifically “in light of the delta variant’s spread.”(thepointsguy.com) The same piece cites a Cars.com mid‑August survey where 20% of respondents were canceling flights and driving instead for Labor Day, and 24% were changing their destination—direct evidence of altered discretionary travel plans.(thepointsguy.com)
• Morning Consult also reported that concern about the Delta variant was very widespread and rising between late July and late August, with 82% of U.S. adults saying they were very or somewhat concerned, and vaccinated adults notably more worried than the unvaccinated.(pro.morningconsult.com) This matches Friedberg’s intuition that vaccinated people, hearing about more breakthrough infections, would become especially uneasy.

Taken together, these data show that by early to mid‑August 2021—a roughly four‑week period after the mid‑July podcast—large shares of Americans, particularly the vaccinated, had become less comfortable and more hesitant about travel and other discretionary public activities, consistent with Friedberg’s prediction. While the surveys don’t directly measure “hearing about friends getting COVID,” they do show the predicted behavioral shift and heightened caution in nearly the exact timeframe he specified, so the forecast is best judged as right.

Evidence shows that California did not return to a broad COVID lockdown or stay‑at‑home order around September 2021.

• On June 15, 2021, California fully reopened, with Governor Newsom terminating the statewide Stay‑at‑Home Order and ending physical distancing, capacity limits, and the color‑tier system.(gov.ca.gov) There is no record of these statewide closures being reimposed later in 2021.
• Subsequent statewide changes before and around September 2021 were targeted measures, not lockdowns: e.g., an August 11 order requiring teachers and school staff to be vaccinated or tested weekly, and an August 19 rule that, effective September 20, large indoor events (1,000+ attendees) must verify vaccination or a recent negative test.(en.wikipedia.org) These rules allowed events and businesses to continue operating under conditions, rather than “locking back up.”
• Local governments (e.g., San Francisco, Contra Costa County, Los Angeles City/County, West Hollywood, Berkeley) added their own vaccine‑check and mask rules for certain indoor businesses and large events in August–October 2021, but businesses generally remained open with restrictions, not closed under a stay‑at‑home or broad shutdown order.(en.wikipedia.org)
• A broader statewide indoor mask mandate was reintroduced later, from December 15, 2021 to January 15, 2022, again as a mitigation measure rather than a stay‑at‑home or lockdown order.(en.wikipedia.org)

Given that the prediction was that Newsom would be “locking it back up in September” (i.e., reimposing major lockdown‑style restrictions), and the actual policy changes were limited, targeted measures without a renewed statewide lockdown or stay‑at‑home order, the prediction did not come true.

The prediction’s time horizon is within a decade from July 2021, i.e., by around 2031. As of today (November 29, 2025), only about four years have passed since the prediction was made, so we are not yet at (or close to) the end of the stated period.

Because the claim concerns what will happen by 2031, and not what must already have happened by 2025, it is too early to determine whether:

• An additional 1–2 billion people who previously lacked broadband will have gained access, and
• This change will have the described transformational impact on the planet’s economy and society.

Even though we can look at interim broadband expansion trends today, those do not definitively confirm or refute what will be true by 2031. Therefore the status of this prediction, judged as of 2025, is inconclusive (too early to tell).

The prediction’s horizon is within five years of July 2021, i.e., up to roughly mid‑2026. As of today (29 November 2025), that deadline has not yet passed, so we cannot definitively say whether it will be true by the stated time.

What we can say so far:

• As of July 2025, about 68–69% of the global population used the internet (around 5.6 billion people), meaning over 30% of people are still offline.(statista.com)
• The UN/ITU estimates roughly 2.6 billion people remained offline in 2024, most in rural and low‑income areas, with affordability and infrastructure still major barriers.(social.desa.un.org)
• Mobile broadband networks (3G/4G/5G) do not yet cover all rural areas; for example, only about 82% of rural populations have 4G coverage, and some rural regions—especially in low‑income countries—still lack any mobile broadband at all.(itu.int)
• Satellite constellations like Starlink have expanded rapidly (service in ~130–150 countries/territories by 2025), but they still face regulatory gaps, cost constraints, and do not provide practical, affordable access to everyone, especially in the poorest regions.(bez-kabli.pl)

So as of late 2025, the world clearly does not have “pervasive internet access everywhere on Earth” in the sense of effectively complete global coverage. However, because the prediction allows until mid‑2026, it is technically too early to say with certainty that it has failed, hence the classification as inconclusive (too early) rather than definitively wrong.

Chamath made the prediction in mid‑July 2021 that Virgin Galactic would be “starting commercial ops” in the “next 2 or 3 quarters,” implying a start roughly by mid‑2022.

In reality, Virgin Galactic had not begun commercial passenger operations by mid‑2022. In October 2021 the company updated guidance to target commercial service in Q4 2022, already outside the 2–3 quarter window. (space.com) In May 2022 it further delayed commercial passenger service to the first quarter of 2023. (space.com)

Virgin Galactic ultimately announced the start of commercial spaceflight service on June 15, 2023, stating that its first commercial revenue‑generating mission, Galactic 01, would fly in a June 27–30 window, and this mission flew on June 29, 2023 as the company’s inaugural commercial spaceflight. (virgin.com) Since commercial operations began about two years after the prediction—well beyond the forecasted 2–3 fiscal quarters—the prediction did not come true.

Jason’s prediction has two parts:

1. Time‑bound claim: Within 1–2 years from July 2021 (i.e., by roughly July 2023), commercial space tourism flights would have a lower per‑trip risk of death or serious injury than riding a motorcycle.
2. Open‑ended claim: Eventually that risk would fall below that of driving a car.

Only the first part is time‑bounded and testable now; if that is false, the overall prediction fails regardless of the second, open‑ended clause.

Baseline risk for motorcycles and cars

U.S. National Highway Traffic Safety Administration (NHTSA) data show that motorcyclists have a very high fatality rate relative to cars, but the absolute risk per mile is still small:

• In 2023 there were 31.39 motorcyclist fatalities per 100 million vehicle miles traveled, versus 1.13 per 100 million miles for passenger car occupants—about a 28× higher rate for motorcycles. (nhtsa.gov)

Even if motorcycles are much more dangerous than cars, this still corresponds to a fatality probability on the order of 10⁻⁷–10⁻⁶ per mile. A typical trip of tens of miles yields a per‑trip fatality risk of only a few in a million or less.

Best estimates of human spaceflight / space tourism risk

Historical and contemporary analyses consistently place human spaceflight orders of magnitude riskier than everyday transport:

• As of early 2021, the in‑flight astronaut fatality rate was about 1 death per 31 boardings (~3.2%). (newspaceeconomy.ca)
• A 2023 safety overview notes that about 3% of astronauts have died during spaceflight, and concludes that spaceflight “still is and probably will remain a dangerous industry,” expecting more deaths as tourism scales. (spacegeneration.org)
• For modern orbital crewed flights, NASA‑linked risk assessments during the era of SpaceX Crew Dragon put the probability of a catastrophic failure on a mission like Inspiration4 at roughly 1 in 300 (≈0.3%). (space.com)
• An Aviation Week analysis, using the X‑15 program as an analogue, assumes a ~0.5% per‑flight fatal crash probability (about 1 in 200) for Virgin Galactic–style commercial suborbital flights when projecting program‑level accident odds. (aviationweek.com)

These are engineering or historical risk estimates—not just counting the small number of recent tourist flights. Even the most optimistic of these (≈0.3% per mission) is thousands of times higher than the per‑trip fatality risk for a typical motorcycle ride.

Regulatory posture is consistent with this high‑risk view. For U.S. space tourism, the FAA does not certify vehicles as safe for human passengers; instead it requires informed consent, explicitly telling customers that the vehicle is not safety‑certified and that they fly “entirely at their own risk,” with disclosure of human‑spaceflight accident history and vehicle safety record. (newspaceeconomy.ca)

What actually happened between July 2021 and July 2023?

During this 1–2 year window, Blue Origin, Virgin Galactic, and SpaceX flew multiple paying or privately funded passengers without any passenger deaths or serious publicized injuries. However:

• The overall human‑spaceflight fatality rate and expert risk estimates remained at the levels above (~0.3–3% per mission), far higher than motorcycles.
• In September 2022, Blue Origin’s NS‑23 mission suffered a booster failure; the capsule’s abort system fired and landed safely, but the booster was destroyed. Although this was an uncrewed research flight, the same vehicle family is used for human tourism, and the FAA grounded New Shepard while the failure was investigated. (en.wikipedia.org) This highlighted that catastrophic‑type failures were still a real possibility in exactly the regime being sold to tourists.

So, while the observed number of tourist fatalities between 2021 and 2023 was zero, the best‑estimate underlying probability of a fatal or serious‑injury event per spaceflight remained orders of magnitude higher than the risk per motorcycle trip.

Comparison and verdict

• Motorcycles: per‑mile fatality risk ≈ 3×10⁻⁷, implying per‑trip risks in the few‑in‑a‑million range for typical rides. (nhtsa.gov)
• Cars: roughly 25–30× safer per mile than motorcycles, so per‑trip risk is even lower. (nhtsa.gov)
• Commercial human spaceflight / space tourism (2021–2023): credible estimates in the 0.3–3% per‑mission range, i.e. one chance in hundreds to low dozens, far higher than road transport.

Given these orders of magnitude, there is no plausible basis on which, by July 2023, commercial space tourism had genuinely become safer per trip than riding a motorcycle, let alone safer than driving a car. Regulatory treatment and expert commentary in this period explicitly describe space tourism as a significantly higher‑risk activity than common transport modes.

Because the time‑bounded part of Jason’s prediction (“in the next year or two”) clearly did not come true, the prediction as a whole is best classified as wrong, even though the longer‑term “eventually safer than cars” clause remains untestable at this time.

Available biographical sources indicate that Chamath Palihapitiya has three children from his first marriage, born in 2008, 2010, and 2012, plus two younger children with his current wife.(stg.scsonline.ca) That means his oldest child will turn 18 in 2026, and as of the evaluation date (November 29, 2025) none of his children has yet reached 18.

The prediction he made on the July 16, 2021 All‑In podcast episode (E40) was explicitly about a future point when his children are of age—either buying each one a ticket to space when they turn 18 or taking a full‑cabin family spaceflight.

Public reporting on Palihapitiya focuses on his role in taking Virgin Galactic public, later selling his personal stake, and eventually resigning from the company’s board.(cnbc.com) These sources mention his desire to one day fly to space but do not document him or his family already having gone, nor do they report him buying spaceflight tickets for his children.

Because (1) the triggering condition (his children being 18+) has not yet occurred, and (2) there is no contrary evidence that definitively shows he won’t follow through later, it is too early to judge whether this long‑dated promise will ultimately be kept or broken.

Publicly available material confirms that in E40 the Besties talk about an upcoming/ongoing Italy trip, Jason’s heavy gelato and pasta consumption, and a “one meal a day” challenge, which matches the context of Chamath joking that Jason will gain significant weight on that trip.(podcasts.apple.com)

Later, E44 is explicitly labeled as including an “Intro & Italy recap,” indicating that the trip had finished by August 28, 2021, and was debriefed on-air.(podcasts.apple.com) However, the accessible summaries and transcript stubs for E44 (and E40) from sites like TranscriptForest and Glasp do not include any concrete statement about how many pounds Jason actually gained or lost, nor any explicit callback that resolves Chamath’s “+15 lbs” prediction.(glasp.co)

Broader searches of Jason Calacanis–related content (including later All-In episodes where they discuss weight and obesity more generally) likewise do not surface any reliable, specific number tied to that 2021 Italy trip.(allin.onpodcastai.com) Because Jason’s exact weight change on that vacation is not publicly documented and the show appears not to have reported a clear outcome of the bet, there is no verifiable way to determine whether he gained “about 15 pounds” as predicted.

Given that the event is long past but the outcome is not knowable from available sources, the prediction’s accuracy must be rated as ambiguous rather than right, wrong, or merely too early to tell.