E14: Salesforce acquires Slack, DeepMind's AlphaFold breakthrough, Trust Fund Socialists & more

What was predicted Chamath argued in early December 2020 that DeepMind/Google would likely withhold AlphaFold for years before it “saw the light of day,” implying there would not be a broad public release of AlphaFold (or an equivalent, high‑capability protein-structure tool) for several years due to ethical/control concerns.

What actually happened

1. AlphaFold 2 paper and open‑source code (July 2021)

   • On 15 July 2021, DeepMind’s AlphaFold 2 was published in Nature, accompanied by open‑source code for the model implementation. (en.wikipedia.org)
   • DeepMind’s own blog and GitHub confirm that AlphaFold’s v2.0 inference pipeline was released as open source in mid‑July 2021, freely available to anyone. (github.com)
     This is roughly 7 months after the December 2020 podcast, not “years.”

2. AlphaFold Protein Structure Database (global access starting July 22, 2021)

   • On 22 July 2021, DeepMind and EMBL‑EBI launched the AlphaFold Protein Structure Database (AlphaFold DB), initially providing ~350–365k predicted structures covering nearly the complete human proteome and 20 model organisms, all freely accessible online. (en.wikipedia.org)
   • This database is explicitly described as an openly accessible resource intended to put the “power of AlphaFold into the world’s hands.” (deepmind.google)

3. Massive further expansion (2022 onward)

   • In July 2022, DeepMind and EMBL‑EBI expanded AlphaFold DB to provide predicted structures for “nearly all catalogued proteins known to science,” over 200 million proteins, again freely downloadable and integrated into core bioinformatics resources. (deepmind.google)
   • By 2023–2024, the database contained over 214 million structures and remained openly accessible. (academic.oup.com)

4. Ethical/biological risk handling

   • DeepMind did conduct an ethics and safety review, but their public post on AlphaFold’s release states they chose broad open release (code and DB) and report no known misuse, rather than holding the technology back for years. (deepmind.google)

Why this contradicts the prediction

• Chamath’s claim was that DeepMind/Google would not let AlphaFold (or an equivalent high‑capability tool) “see the light of day” for years due to control/ethical concerns.
• In reality, within about 7–8 months of the December 2020 discussion, DeepMind:
  • Published the core AlphaFold 2 methodology in a top journal,
  • Open‑sourced the inference code, and
  • Launched a freely accessible global database of hundreds of thousands, then hundreds of millions, of AlphaFold predictions.
• This is a rapid and very broad release, not a multi‑year delay. While AlphaFold is a structure‑prediction system rather than a full generative protein-design suite, it is precisely the capability discussed at the time and has become a foundational, openly available tool for protein design and structural biology worldwide.

Given this timeline and breadth of access, Chamath’s forecast that the technology would be held back from broad release “for years” is clearly falsified.

By the 24‑month mark (early December 2022), multiple non‑Google/DeepMind groups had publicly demonstrated protein‑folding models with AlphaFold‑comparable accuracy, effectively replicating the AlphaFold2 result.

• HelixFold (Baidu/PaddlePaddle team, July 2022) – HelixFold is an independent reimplementation of AlphaFold2 using PaddlePaddle. Its authors report that, on CASP14 and CAMEO benchmarks, HelixFold’s accuracy is “on par with AlphaFold2” and they released code and a public web service. This is a non‑Google group matching AlphaFold2’s accuracy on standard benchmarks well before December 2022, i.e., a direct replication of the AlphaFold2 result. (arxiv.org)
• RoseTTAFold (Baker lab, University of Washington, Science 2021) – An academic group developed RoseTTAFold, a three‑track neural network that predicts 3D protein structures and protein–protein complexes with high accuracy. Contemporary commentary notes it is based on the AlphaFold2 framework and achieves high‑accuracy structure prediction, though “not quite as accurate as Alphafold2,” which had been kept private at the time. (espeyearbook.org) This still shows an independent deep‑learning model reaching AlphaFold‑like performance in many cases within 2021.
• ESMFold (Meta AI, 2022) – By November 2022, Nature was describing Meta’s ESMFold as “AlphaFold’s new rival” in protein structure prediction; it was characterized as somewhat less accurate than AlphaFold2 but still an accurate, large‑scale structure predictor, and markedly faster. (en.wikipedia.org)

In addition, reviews from 2022 summarize that several deep‑learning methods (including RoseTTAFold and other two‑step approaches) can reach accuracy similar to state‑of‑the‑art AlphaFold2 on many targets, reinforcing that comparable performance had been achieved outside DeepMind within this time window. (academic.oup.com)

Because at least one non‑Google/DeepMind group (notably HelixFold, and arguably RoseTTAFold and ESMFold as well) had publicly demonstrated AlphaFold‑level or near‑AlphaFold‑level protein‑folding accuracy before December 2022, Friedberg’s prediction that “in the next 24 months, someone else will replicate this” is fulfilled.

• What was predicted?
  Friedberg said that within ~1 year of the Dec 2020 episode, “a lot of startups” would replicate AlphaFold-like techniques for protein folding and use them for drug discovery, and that around 50 such startups would be funded 12–18 months out (roughly Dec 2021–Jun 2022).(podscripts.co)

• Evidence that some of this happened:

  • DeepMind open‑sourced AlphaFold 2 and launched the AlphaFold Protein Structure Database in mid‑2021, which is widely acknowledged as catalyzing new research and commercial efforts around AI‑driven protein structure and drug discovery.(en.wikipedia.org)(forbes.com)
  • Multiple AI protein‑design / protein‑engineering startups were indeed founded in 2021–2022, often explicitly building on AlphaFold-era advances. Examples include:
    • Isomorphic Labs (Alphabet AI drug‑discovery company, explicitly drawing on AlphaFold technology, incorporated Feb 2021 and announced Nov 2021).(en.wikipedia.org)
    • Cradle (founded 2021, using machine learning and later generative AI for protein engineering; seed round announced in 2022).(en.wikipedia.org)
    • A BiopharmaTrend survey lists 12 AI‑powered biotech companies founded in 2021, including Perspix Biotech (AI‑driven protein engineering) and an early Profluent Bio effort in protein design, which together illustrate a broader wave of AI‑first protein and drug‑discovery startups in that period.(biopharmatrend.com)
  • These examples support the qualitative part of his claim that a noticeable cohort of new AI‑protein / protein‑design startups emerged within roughly a year of AlphaFold 2’s breakthrough.

• Where the prediction becomes unverifiable:

  • Friedberg’s more specific forecast was that on the order of 50 startups based on “some novel protein idea” would be funded in the 12–18 months after Dec 2020.(podscripts.co)
  • Publicly available surveys and retrospectives on AI‑powered biotech and protein‑design startups (e.g., the BiopharmaTrend 2021 list, general discussions of an expected “ecosystem” of AlphaFold‑driven startups in Forbes) describe a growing wave of companies but do not provide a comprehensive count anywhere near granular enough to verify whether ~50 distinct AlphaFold‑style protein startups actually closed funding rounds specifically between Dec 2021 and Jun 2022.(biopharmatrend.com)(forbes.com)
  • Determining the true count would require systematic access to private deal databases (PitchBook, Crunchbase, etc.) with detailed tagging for both "novel protein" focus and AlphaFold‑like techniques—information not available in open sources. No credible industry report or academic survey was found that retrospectively tallies “dozens” or “about 50” such companies funded in that exact 12–18‑month window.

• Conclusion:

  • The directional part of the prediction (a wave of new AI protein‑design / AlphaFold‑inspired startups within a year) is broadly supported by examples and contemporary commentary.
  • The quantitative part (≈50 funded startups in the Dec 2021–Jun 2022 window) cannot be reliably confirmed or refuted from public, citable sources.

Because the core of the prediction hinges on that specific magnitude ("you’ll see 50 startups getting funded 12 to 18 months from now") and there is no comprehensive data to check it, the overall assessment has to be ambiguous rather than clearly right or wrong.

Reasoning

• The prediction (made in December 2020) is explicitly about ~30 years in the future: 2020 + 30 = 2050.
• The current date is November 29, 2025, which is less than 5 years after the prediction and about 25 years before the target timeframe.
• The claim is about the relative importance and size of the biodefense industry by around 2050, compared to today’s major defense or healthcare sectors. This is a long‑term structural forecast about global industry composition.
• There is no way, in 2025, to know whether the biodefense sector in 2050 will in fact be “one of the most important and sizable global industries” on par with defense or healthcare.

Because the prediction is explicitly about conditions around 2050 and we have not yet reached that timeframe, it is too early to evaluate whether it has come true.

Therefore the correct classification is: inconclusive (too early to tell).

It’s too early to decisively judge this prediction.

1. Timeframe hasn’t elapsed
Chamath’s claim was about “the next 20 or 30 years” from ~2020. As of today (late 2025), only about 5 years have passed, so we cannot yet know whether Google will, over that whole period, be under “layers and layers of oversight” analogous to nuclear regulation.

2. Evidence of growing oversight (in direction of his claim)

• The EU AI Act (in force since August 1, 2024) creates a comprehensive AI regulatory framework, including governance bodies (European AI Office, national supervisory authorities) and obligations for general‑purpose and “systemic risk” AI models, which will come fully into effect over 2025–2027. These rules clearly apply to major providers like Google/DeepMind. (digital-strategy.ec.europa.eu)
• Analysis of the AI Act notes that general‑purpose AI models with systemic risk (a category that would include Google’s largest models) face extra requirements: model evaluation, risk mitigation, incident reporting, and cybersecurity—on top of baseline transparency and documentation. (crfm.stanford.edu)
• Google has stated it will sign the EU’s code of practice for general‑purpose AI, designed to operationalize compliance with the AI Act—another layer of formal oversight. (techcrunch.com)
• In the U.S., Google is one of the companies that entered into voluntary AI safety commitments with the Biden administration, covering pre‑deployment testing, security, transparency, and risk research—an early, softer form of regulatory oversight. (bidenwhitehouse.archives.gov)
• UK and international scrutiny of Google DeepMind’s frontier models (e.g., Gemini) and alleged breaches of safety pledges has intensified political pressure for mandatory regulation and stronger oversight regimes. (time.com)

3. But not yet “nuclear‑style” multi‑layer oversight focused on Google/DeepMind
Despite these developments, current oversight is:

• Still emerging and fragmented (EU, U.S., U.K., etc. each moving at different speeds),
• Largely sector‑wide (applies to many AI providers and platforms) rather than uniquely centered on Google, and
• Not yet as entrenched, stringent, and specialized as the nuclear regulatory regimes Chamath invoked by analogy.

So far, events are partially consistent with his direction of travel (growing, multi‑layer AI oversight that affects Google), but we lack the full 20–30‑year perspective, and today’s institutions are not yet clearly comparable to nuclear oversight in scope and rigidity. Therefore the prediction cannot be fairly scored as right or wrong yet, making the appropriate label “inconclusive (too early)”.

Reasoning about time horizon

• The prediction is explicitly about what will happen "over the coming decades." As of Nov 2025, only ~5 years have elapsed since the 2020 statement, so the forecast period is far from over.

What we do see by 2025

• Advanced biological and AI-enabled design tools have clearly become more capable and somewhat more accessible:
  • AI-driven biodesign systems (e.g., the Moremi Bio agent) have been shown experimentally to generate large numbers of novel toxic proteins and small molecules, raising concerns about dual-use misuse and accessibility to people with limited expertise. (arxiv.org)
  • A Microsoft-led study generated tens of thousands of toxic protein sequences that evaded existing DNA-synthesis screening, illustrating how modern AI tools can create dangerous biological designs at scale. (washingtonpost.com)
  • A 2023 review on AI and biological misuse finds that large language models and specialized biological design tools together could both lower barriers for non-experts and expand capabilities for sophisticated actors, potentially broadening the set of people who could attempt serious biological misuse. (arxiv.org)
  • The U.S. National Academies (2025) conclude existing AI-enabled biological tools can already design or redesign toxins, while stressing that physical production and technical limits still constrain their use for large-scale or pandemic-level attacks. (nationalacademies.org)

But key parts of the prediction remain unverified

• Major policy and security assessments emphasize that, so far, AI’s effect on overall biological risk has been limited and mostly theoretical:
  • A 2024 CNAS report states that, although AI could greatly increase biocatastrophic risk in the future, its actual impact on biological risk to date has been marginal, and many worst‑case capabilities remain speculative. (cnas.org)
  • OpenAI’s 2024 evaluation of GPT‑4 for biological threat creation found at most a mild uplift in performance, not a dramatic enabling of non‑expert attackers. (openai.com)
  • A 2025 CSIS report frames AI‑enabled bioterrorism as a domain where barriers are expected to fall and risks may accelerate, but again treats this mainly as a forward‑looking concern requiring preparation, not as an already‑realized surge in catastrophic capability among many actors. (csis.org)
• There is no public evidence by late 2025 of a measurable, large increase in the number of malicious actors actually attempting catastrophic bio‑attacks, nor of catastrophic attacks carried out using these new tools.

Conclusion

• The premise of the prediction (tools getting more powerful and more accessible, prompting expert concern about more potential malicious users) is increasingly reflected in the literature and in policy debates.
• However, the core claim—that this will materially raise global catastrophic biorisk via a substantially larger pool of malicious actors over the coming decades—is about a longer timeframe and depends on developments (capabilities, governance, actual attacker behavior) that have not yet fully unfolded.

Because the relevant decades-long period is far from complete and the most consequential part of the claim (a large realized increase in catastrophic bio‑risk from many new malicious actors) cannot yet be empirically assessed, the prediction is too early to call and is therefore inconclusive.

The prediction is about the latter half of the 21st century (roughly 2050–2100), stating that biodefense will become one of the biggest industries on Earth by then. As of the current date (2025-11-29), that time period has not begun, so it is too early to determine whether biodefense will in fact become comparable to or larger than sectors like conventional defense, energy, or IT. Therefore, the correctness of this prediction cannot yet be evaluated.

Summary
The prediction links a counterfactual (Giuliani exiting quickly) to election outcomes in Georgia. We can verify the factual premise (Giuliani’s continued prominence), but not the counterfactual effect on GOP chances with enough rigor to call it clearly right or wrong, so the outcome is ambiguous.

1. Did Giuliani “stay on the stage” through the Georgia runoff period?

Yes. Rudy Giuliani remained a leading public face of Donald Trump’s election‑fraud allegations well into December 2020 and up to the Georgia runoffs on January 5, 2021:

• He led or fronted multiple post‑election lawsuits and public events alleging widespread fraud in several states in late November and December 2020.
• He appeared at legislative-style hearings and press events pushing fraud claims, and continued to do media hits and public statements contesting the 2020 presidential result during December 2020, overlapping with the Georgia runoff campaign period.

So the if part of the prediction—Giuliani remaining publicly prominent with fraud claims during the runoff period—did happen.

2. What actually happened in the Georgia runoffs?

On January 5, 2021, Democrats Raphael Warnock and Jon Ossoff both won their Georgia Senate runoffs, flipping control of the U.S. Senate. Republican turnout in some strongly pro‑Trump areas underperformed relative to November 2020, and analysts have argued that Trump’s and his allies’ fraud rhetoric may have depressed GOP turnout or created confusion. But this is an inference, not a directly observable fact.

3. Why the prediction is not clearly verifiable

The prediction’s core claim is causal and counterfactual:

With Giuliani staying prominent, Republican chances in the Georgia runoffs will be materially worse than if he had exited quickly.

To judge this, we would need credible evidence about a hypothetical alternate world in which Giuliani quickly stopped advancing fraud claims while everything else stayed constant. In reality:

• Many actors besides Giuliani (Donald Trump himself, Sidney Powell, Lin Wood, right‑wing media, etc.) were loudly promoting fraud narratives in Georgia, making it impossible to isolate Giuliani’s unique marginal effect.
• Post‑election analyses and academic studies can suggest that fraud rhetoric in general may have hurt Republicans in Georgia, but they do not meaningfully disentangle Giuliani’s role from that broader ecosystem or quantify a “material” effect relative to a scenario where he disappeared early.
• Different plausible models of voter behavior could support both of these stories: (a) Giuliani and fraud rhetoric hurt GOP chances, or (b) other factors (Democratic mobilization, demographic trends, campaign strategy) were decisive and Giuliani’s marginal impact was negligible.

Because the claim rests on an unobservable counterfactual and Giuliani was just one of multiple prominent messengers, we cannot definitively determine whether his continued prominence materially worsened Republican odds relative to an early exit.

Conclusion: The premise (Giuliani stayed very visible) is true; the outcome (GOP lost the runoffs) is known; but the specific causal comparison the prediction makes cannot be reliably confirmed or falsified. Therefore the prediction’s accuracy is ambiguous rather than clearly right or wrong.

Outcome of the January 5, 2021 Georgia runoffs

• Both contested U.S. Senate seats in Georgia went to Democratic candidates in the January 5, 2021 runoff elections.
• Jon Ossoff (Democrat) defeated incumbent David Perdue (Republican) in the regular Class II Senate election. (en.wikipedia.org)
• Raphael Warnock (Democrat) defeated incumbent Kelly Loeffler (Republican) in the special election for the Class III seat. (en.wikipedia.org)
• As a result, Democrats captured both seats and secured effective control of the U.S. Senate. (en.wikipedia.org)

Assessment of the prediction
Chamath’s prediction was that “there is a very good chance that these guys are going to win these two Senate seats”—i.e., he assigned a high likelihood that Democrats would win both Georgia runoff races. Since Democrats did, in fact, win both Senate seats in the January 5, 2021 runoffs, the prediction is right.

Available evidence shows that some prominent Republicans did publicly criticize Trump’s post‑election legal efforts, but there was no broad move by leading congressional Republicans to curtail or repudiate Rudy Giuliani specifically, nor was he pushed off center stage before the January 5, 2021 Georgia runoffs.

Key points:

1. Giuliani stayed the face of Trump’s effort well past January 5. He continued leading strategy sessions and working with allies at the Willard Hotel “command center” in December, and remained central to the broader attempt to overturn the election into early January 2021, including planning around the January 6 certification fight. (en.wikipedia.org)

2. Criticism from within the GOP was scattered and often generalized, not a coordinated congressional repudiation of Giuliani. Some Republicans criticized the legal team or the fraud claims:

   • Former New Jersey governor Chris Christie (not a member of Congress) called Trump’s legal team, including Giuliani, a “national embarrassment” and urged Trump to stop the challenges. (uproxx.com)
   • Sen. Joni Ernst slammed Sidney Powell’s conspiracy claims as “absolutely outrageous” and “offensive,” but framed it as a rebuke of Powell and the broader claims, not a specific move to sideline Giuliani. (iowastartingline.com)
   • Sen. Ben Sasse, in a December 31 statement, denounced the project to overturn the election as a “dangerous ploy” and “swampy politics,” criticizing the fundraising and legal challenges generally, not Giuliani personally. (ottawa.citynews.ca)

3. Congressional leadership did not publicly move to shut Giuliani down. Senate Majority Leader Mitch McConnell acknowledged Biden’s win on December 15 and privately urged senators not to join House objections to Electoral College results, but he did not publicly single out Giuliani or call for his removal from the effort. Meanwhile, figures like Josh Hawley and others still backed objections to the electoral count, aligning with the broader challenge Giuliani was fronting. (en.wikipedia.org)

4. Internal dissatisfaction with Giuliani did not translate into congressional action. Reporting indicates Trump himself and some allies worried Giuliani and other lawyers were “fools” making him look bad, yet Giuliani remained Trump’s point man on the election challenge. These were White House/political complaints, not moves led by congressional Republicans to remove him from center stage. (cnbc.com)

Given that Giuliani remained the public and strategic leader of the post‑election effort through and beyond the Georgia runoffs, and that congressional Republican criticism was limited, fragmented, and not clearly aimed at curtailing Giuliani’s role, the prediction that “leading congressional Republicans” would publicly move to sideline him before January 5, 2021 did not come true.

Assessment
The prediction was that in the 2022 U.S. midterm elections, Republicans would:

1. Win back the majority in the U.S. House of Representatives, and
2. As a result, Nancy Pelosi would cease to be Speaker of the House.

What actually happened

• In the November 8, 2022 midterm elections, the Republican Party won a majority of seats in the U.S. House of Representatives for the 118th Congress.
• Nancy Pelosi, a Democrat, had been Speaker of the House in the 116th and 117th Congresses. After Republicans gained the majority, she did not stand for the speakership in the 118th Congress and was succeeded by Republican Kevin McCarthy, who was elected Speaker on January 7, 2023.
• Therefore, as a consequence of Republicans winning the House, Pelosi ceased to be Speaker.

Conclusion Both specific elements of the prediction came true: Republicans won back the House majority in the 2022 midterms, and Nancy Pelosi stopped being Speaker as a result. Thus the prediction is right.

It’s too early to know whether this 20‑year forecast is right or wrong.

What Chamath said

• Quote (Dec 4, 2020): “And we are about to go through over the next 20 years, $30 trillion of wealth transfer.”
• Normalized prediction: roughly 2020–2040, about $30T of intergenerational wealth transfer in the U.S.

What current research says

• A 2018 Cerulli Associates report projected about $68T in U.S. wealth transfer between 2018 and 2042.
• Later work (e.g., a much‑cited 2021 analysis by Cerulli/others) estimates on the order of $70–80T+ of U.S. wealth transfer between ~2020 and mid‑2040s, depending on exact date range and definitions (all inheritances vs. just Boomer wealth, etc.).

These estimates:

• Are projections spanning roughly similar but not identical periods (e.g., 2018–2042 or 2021–2045 rather than exactly 2020–2040).
• Use varying definitions (total U.S. household wealth vs. just Baby Boomer wealth; inter vivos gifts vs. bequests).
• Extend well beyond 2025, so they have not yet played out.

Why the prediction cannot be scored yet

• The forecast period (≈2020–2040) is only about 5 years in as of Nov 29, 2025.
• The ultimate amount depends heavily on future asset prices, mortality, tax policy, and behavior over the next 15 years.
• While many current projections are higher than $30T for similar periods, they are not realized facts and differ in scope, so we cannot definitively say his ~$30T figure will be too low or wrong for his exact 20‑year window.

Because:

• The forecasted period is far from complete, and
• Existing numbers are themselves forward‑looking estimates, not outcomes,

the correct classification today is “inconclusive (too early)”, not yet clearly right or wrong.

Summary
Chamath predicted in December 2020 that:

1. "We are all going to need a vaccine" — i.e., broad population in developed countries would need/receive COVID‑19 vaccination.
2. These vaccines would come from for‑profit companies that thrived on top of R&D funded by prior drug profits (i.e., the standard big‑pharma model, not purely state labs or non‑profits).

By late 2025, both parts are substantially correct.

1. Did the broad population end up needing and receiving vaccines?

• In the United States, COVID‑19 vaccines became widely recommended for essentially the entire adult population and later for children. CDC data show that by mid‑2022, about 79% of the total U.S. population had received at least one dose, and over 90% of adults had received at least one dose.
• In the EU and other high‑income countries, vaccination rates were similarly high; for example, the EU reported over 70% of adults fully vaccinated by late 2021, with even higher one‑dose coverage.
• Major public‑health bodies (CDC, WHO, EMA, etc.) recommended primary series and boosters broadly, not just for a tiny at‑risk subset, confirming that vaccination was considered needed at a population level in developed countries.

While not literally every single person was vaccinated, in context the prediction was about broad population‑level need and uptake. That did happen.

2. Were the vaccines from for‑profit pharma companies funded by prior profits‑driven R&D?

The first and dominant vaccines in the U.S. and most of Europe were:

• Pfizer‑BioNTech (Comirnaty) – Pfizer is a large for‑profit pharmaceutical company with extensive R&D funded historically by its drug revenues (e.g., Lipitor and many others). BioNTech is also a for‑profit biotech firm.
• Moderna (Spikevax) – a for‑profit biotech company, already heavily funded by equity, partnerships, and prior revenue/pipeline value, investing heavily in mRNA R&D before COVID‑19.
• Johnson & Johnson (Janssen) – J&J is a large for‑profit pharma/medical products company with long‑standing R&D funded largely by profits from its product portfolio.
• AstraZeneca‑Oxford – AstraZeneca is another major for‑profit pharma company relying on profitable drugs and R&D investment.

These firms did receive substantial public funding and advance‑purchase guarantees (e.g., Operation Warp Speed in the U.S., EU and U.K. procurement deals), but that does not contradict Chamath’s claim. His point was about who the producers are and their business model. The leading vaccines were indeed:

• Developed and manufactured by large for‑profit pharma/biotech companies.
• Building on pre‑existing R&D infrastructure and platforms, which in turn were funded over years by prior drug revenues and private capital.

While there were also vaccines from state‑linked or non‑profit actors (e.g., Russia’s Sputnik V, China’s Sinovac/Sinopharm, some public‑lab efforts), these did not dominate vaccination campaigns in the U.S. and Western Europe and do not negate the general statement about developed‑world vaccination.

Conclusion

• The broad population in developed countries did end up needing and receiving COVID‑19 vaccines.
• These vaccines were overwhelmingly produced by large, for‑profit pharmaceutical and biotech firms with R&D capabilities built on years of investment funded largely by prior commercial success.

Therefore, the prediction is right in substance, even if not every single individual was vaccinated and even though governments contributed significant emergency funding alongside those existing profit‑driven R&D engines.

Why this prediction is marked “right”

• In the SPACsgiving episode E13 (Nov 25, 2020), the hosts explicitly discuss starting “an AngelList syndicate for every All‑In podcast listener” and Jason says “We’ll do it with the syndicate.com slash all‑in podcast… we’ll aggregate all these subscribers.”(podscripts.co) This is the initial commitment to a listener co‑investment mechanism.
• In E14 (Dec 4, 2020), they report that this has actually been set up: “Everybody go to the syndicate.com slash all in. If we do this, I’m collecting the emails now… If you want to invest with the besties…” and then Jason closes with: “I asked my crack team… to set up the syndicate.com slash all in… We’re going to do something to let the people who listen invest alongside us. We’re going to figure it out in 2021 at some point.”(podscripts.co) That same episode’s show notes label this as “Invest with the Best…ies!” and link directly to https://www.thesyndicate.com/allin, calling it the All‑In Syndicate.(podscripts.co)
• Thus, by December 4, 2020—well before and then throughout calendar year 2021—the hosts had launched a dedicated sign‑up mechanism for podcast listeners that funneled them into Jason Calacanis’s existing angel syndicate for co‑investing.
• In a later 2021 episode (E26, “State of Venture Capital”), Jason describes The Syndicate’s 2021 activity: “we’re on track this year to do five deals a month, 60 deals, and we’ll put 50 million to work this year at thesyndicate.com.”(podscripts.co) The whole point of that platform is to let outside LPs co‑invest alongside him in those deals, so any All‑In listener who signed up via thesyndicate.com/allin in 2021 could participate in at least some of these investments.
• Even though the thesyndicate.com/allin URL 404s now, multiple independent archives (Libsyn, Apple Podcasts, Podscripts/Metacast) preserve the All‑In Syndicate link and branding tied to E14, confirming that this was a real listener-facing mechanism, not just offhand banter.(allinchamathjason.libsyn.com)

Interpretation

The core of the prediction was that by 2021 the hosts would create “some mechanism (e.g., via The Syndicate) that allows podcast listeners to co‑invest alongside them in at least one deal.” By:

1. Creating and promoting the All‑In‑branded syndicate signup (thesyndicate.com/allin) to listeners, and
2. Running many syndicated startup deals through that platform in 2021,

they did in fact provide such a mechanism within the specified timeframe (in fact, slightly early in late 2020, and then active during 2021). On that basis, the prediction is best scored as right.