E26: State of Venture Capital, plus fan questions on longevity, decentralization & quantum computing

The prediction is that at least one person who was alive as of March 2021 will eventually live past age 200. As of November 2025:

• No human has come close to this age; the oldest verified age in history is 122 years (Jeanne Calment), far below 200.
• The prediction concerns future maximum lifespan of currently living people. To declare it right or wrong, we would need to wait until all individuals who were alive in March 2021 have died, which will not be knowable for well over a century.

Since the claim is about a still‑unfolding future outcome and there is no logical way yet to falsify or confirm it, the status of the prediction cannot be determined at this time.

As of November 29, 2025, there is no quantum computer with ~4000 logical qubits capable of breaking RSA‑2048.

Current leading systems (e.g., IBM, Google, Quantinuum, IonQ) report on the order of hundreds to low thousands of physical qubits, with only small numbers of error‑corrected logical qubits demonstrated in practice:

• IBM’s published roadmaps and recent devices discuss thousands of physical qubits, but logical qubits at scale remain a research goal, not an achieved capability.
• Experimental demonstrations of logical qubits and small code distances exist (e.g., surface codes and other error‑correcting codes), but they are far from the thousands of logical qubits needed for factoring RSA‑2048 using Shor’s algorithm.
• No credible report from major labs, peer‑reviewed papers, or industry announcements claims the ability to factor RSA‑2048 or equivalently strong keys with an actual machine.

Because:

1. The prediction’s time window for the event is 2040–2060, and we are still in 2025.
2. The sub‑claim that there is <5% chance of such a machine before 2040 is equivalent (from today’s vantage point) to saying “it’s very unlikely we’ll have this before 2040,” and nothing observed so far contradicts that; progress is significant but still orders of magnitude away in logical‑qubit scale and error correction.
3. The specific event that would falsify him (a quantum computer with ~4011 logical qubits that can break RSA‑2048) has not occurred yet.

The appropriate evaluation today is that the prediction has not been falsified and is consistent with current reality. Since the key part of the prediction is about not having such a machine before 2040, and we’re still well before 2040 with no such machine, the prediction is best judged as right (so far) rather than wrong or ambiguous.

(No single specific source states “no RSA‑2048‑breaking quantum computer exists,” but this is inferred from the absence of any such claim in the quantum computing literature and industry communications, and from the widely reported current scales of physical and logical qubits.)

As of 29 November 2025, the 10‑year window Friedberg named ("over the next ten years" from 2021, i.e., roughly through 2031) is only about halfway over, and the key parts of his prediction are only partially testable.

What has clearly happened (supports parts of the claim):

1. The area is being “chased really hard” this decade.

   • Major pharma companies have launched explicit quantum‑chemistry and drug‑discovery collaborations (e.g., Boehringer Ingelheim with Google Quantum AI for molecular dynamics and drug design; Roche with Cambridge Quantum; Merck with SEEQC). (en.wikipedia.org)
   • Cloud and platform providers have built chemistry/life‑science offerings around NISQ hardware, such as Microsoft’s Azure Quantum Elements, explicitly targeted at materials, chemistry, and pharmaceutical R&D. (en.wikipedia.org)
   • Investment vehicles like Novo Holdings have committed hundreds of millions of dollars to quantum‑computing startups with a stated focus on life sciences, including protein modeling and accelerating drug discovery. (reuters.com)
   • Numerous reviews and road‑mapping pieces in 2023–2025 outline quantum computing as a key near‑term application for generative chemistry and drug discovery, specifically emphasizing the use of noisy intermediate‑scale quantum (NISQ) devices. (pubmed.ncbi.nlm.nih.gov)
     Conclusion: The claim that this application area would be actively and broadly pursued during the 2020s is already correct.

2. Noisy / pre‑fault‑tolerant devices are being used for real molecular and materials simulations, including some non‑toy results.

   • IBM’s work on "quantum utility" shows noisy quantum processors (e.g., Heron) solving Ising‑model problems relevant to materials science beyond brute‑force classical simulation, marking a first practical regime where NISQ devices provide scientifically useful results. (newswire.ca)
   • Hybrid quantum–classical algorithms like Quantum‑Selected Configuration Interaction (QSCI) have been used by IBM and partners to push chemical computations "beyond the scale of exact diagonalization" on classical machines, demonstrating the largest‑scale chemistry simulations on quantum hardware to date. (en.wikipedia.org)
   • Google’s 2025 Quantum Echoes algorithm, running on the noisy 105‑qubit Willow processor, achieved a quantum advantage benchmark about 13,000× faster than the best classical methods and was applied to NMR‑related molecular dynamics, uncovering previously unknown atomic structures in molecules with 15 and 28 atoms. (reuters.com)
   • A 2024 Nature paper (summarized in Drug Discovery News) and related work show hybrid quantum computing applied to realistic drug‑discovery tasks (e.g., prodrug activation energetics and covalent KRAS‑G12C inhibition by sotorasib), demonstrating that small noisy quantum devices can contribute to modeling non‑toy chemical problems. (drugdiscoverynews.com)
     Conclusion: We do see genuine research breakthroughs in quantum simulations of molecules and materials on noisy hardware, consistent with his claim that “super noisy, super low fidelity” machines would start yielding breakthroughs this decade.

What has not clearly happened yet (limits of the claim as of 2025):

1. No widely accepted, end‑to‑end “practical breakthrough” in enzyme/drug design attributable to NISQ quantum computers.
   Recent surveys and industry analyses still describe quantum computing for drug discovery as early‑stage / emerging, with tools “limited in availability” as of mid‑2025 and use cases framed as proofs of concept or pilots rather than standard industrial workhorses. (mdpi.com)
   Reviews on quantum computing in drug discovery and medicine emphasize potential and early demonstrations, but do not report approved drugs, industrial enzymes, or nitrogen‑fixation catalysts that owe their discovery primarily to quantum computers. (mdpi.com)
   In other words, most impact so far is in research demos and methodology, not yet in widely acknowledged breakthroughs like new commercial enzymes or drugs designed thanks to quantum advantage.

2. His strong sub‑claim that “we already have enough compute power” remains questionable.
   Even in 2020–2022, experts in quantum‑accelerated drug discovery repeatedly noted that hardware was “still in its infancy” and "not quite at a scale" to solve the toughest real‑world drug problems, and more recent work still frames many life‑science applications as contingent on further scaling and error‑correction over the next decade. (pharmaceutical-technology.com)
   The most advanced 2024–2025 results (e.g., Willow + Quantum Echoes, hybrid pipelines) are still presented as stepping stones toward practical drug and materials design, not evidence that today’s NISQ machines are already sufficient for the full class of industrially relevant problems Friedberg invoked. (livescience.com)

Why the overall verdict is “inconclusive” rather than right or wrong:

• The strongest, most specific part of Friedberg’s prediction is that within about 2021–2031, noisy quantum computers will deliver practical breakthroughs in simulating quantum states of materials/atoms/molecules that directly translate into things like better enzymes for nitrogen fixation, drug discovery, or targeted proteins, using the compute power already available at the time he spoke.
• By late 2025, we do have:
  • Widespread pursuit of exactly these applications in industry and academia. (en.wikipedia.org)
  • Credible, noisy‑hardware demonstrations of non‑toy molecular simulations and even quantum advantage on chemistry‑related benchmarks. (newswire.ca)
• But we do not yet have clear, broadly agreed‑upon industrial outcomes like new approved drugs or widely deployed enzymes that owe their discovery to these NISQ simulations; authoritative sources still characterize this as a developing, not yet fully realized, capability. (mdpi.com)
• Since his horizon runs until roughly 2031, there is still substantial time for such practical breakthroughs to emerge as hardware and algorithms continue to improve.

Because an essential part of his prediction concerns outcomes by the end of the decade and those outcomes have neither clearly materialized nor clearly failed by 2025, the fairest assessment today is “inconclusive (too early)” rather than definitively right or wrong.

Chamath argued that as companies like Prenuvo gathered more data and improved ML, two things would happen: (1) full‑body MRI scan times would fall from ~60–90 minutes to roughly 40 minutes, and (2) prices would drop from ~$1,500–$2,500 into the low hundreds of dollars while preserving diagnostic quality.

For Prenuvo itself, neither target has been reached as of late 2025. Prenuvo markets its whole‑body MRI as “fast (under 1 hour)” with typical scan times around 45–60 minutes, not ~40 minutes.(prenuvo.com) Its current U.S. pricing for a comprehensive whole‑body scan is about $2,499, with enhanced packages at $3,999–$4,499, i.e., still in or above the original $1,500–$2,500 range rather than in the low hundreds.(prenuvo.com) Industry reporting likewise places Prenuvo’s full‑body scans around $2,400–$2,500.(bodyspec.com)

Across the broader “full‑body MRI” market, typical consumer preventive scans remain roughly $1,000–$2,500, with some lower‑cost offerings like SimonMed’s simonONE Body scan at ~$650 and about an hour of scanner time—cheaper, but still not “low hundreds.”(goodhousekeeping.com)

At the same time, newer AI‑driven services such as Ezra/Function Health have introduced substantially faster and cheaper MRI‑based screening. Function Health now offers an MRI scan (covering head, neck, abdomen, and pelvis) that takes about 22 minutes and costs $499, explicitly described as a reduction from earlier 60‑minute, ~$1,500 scans, and marketed as screening for hundreds of conditions.(functionhealth.com) However, this product does not cover the entire body in the same way as Prenuvo’s flagship scan (it omits chest, lungs, and limbs), and there is not yet clear, independent evidence that its diagnostic performance is fully equivalent to longer, more expensive protocols.

Because Chamath’s prediction was framed around Prenuvo‑style full‑body MRI but the outcomes differ depending on which provider and definition of “full‑body” you use—and because equivalence of diagnostic resolution is not yet well established—the evidence is mixed rather than clearly validating or falsifying his claim.

As of 29 November 2025, the target year (2030) has not yet been reached, so it is too early to determine whether the prediction will ultimately be correct.

Available data show that global venture capital (VC) investment has not yet reached $1 trillion annually, but that does not invalidate a “by 2030” prediction:

• KPMG reports global VC investment hit a record about $671 billion in 2021, up from $347 billion in 2020. (kpmg.com)
• S&P Global Market Intelligence data show global VC funding totaled about $435.7 billion in 2022 and then fell to about $260 billion in 2023. (spglobal.com)
• Other analyses put 2023 and 2024 global VC investment in the mid–300 billions, with Chambers & Partners citing USD 345.7 billion in 2023 and USD 368.3 billion in 2024, reflecting a modest recovery but still well below 2021’s peak. (practiceguides.chambers.com)

In all major data series, annual global VC investment is far below $1 trillion as of 2024, and 2025 data are incomplete. However, because the prediction’s horizon is “by 2030” and there are roughly five years remaining, its truth value cannot yet be determined. Therefore the correct classification today is “inconclusive (too early).”

The prediction is clearly stated in the All-In E26 transcript: Jason says that in 2021 they are “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)

To judge this, we would need reliable, deal-by-deal or year-end figures for TheSyndicate.com’s 2021 activity: number of SPVs/deals and total dollars deployed through the syndicate SPVs specifically.

Public data, however, are fragmentary and inconsistent:

• VC and crypto databases attribute only a few dozen total investments to TheSyndicate.com across all years. For example, Cointime lists 29 investments with the most recent in November 2021, not broken out by amount syndicated or confirming 60 deals in that calendar year.(cointime.ai) Unicorn Nest similarly shows 44 total investments with 2021 as the “peak activity year,” but again without a clear 2021 deal count or capital total, and its list is clearly incomplete relative to other sources.(unicorn-nest.com)
• Kingscrowd shows many separate 2021 SPVs titled “X, a Series of Jason’s Syndicate 2021,” each raising hundreds of thousands to roughly $1.4M (e.g., Giggster, Kippo), confirming active 2021 syndicate deal flow but not giving a complete universe or roll‑up total.(kingscrowd.com)

Later interviews show scale but not a backward-looking 2021 tally:

• In an October 4, 2022 Acquired episode, Jason says his angel syndicate is the largest in the world and that he has deployed $185M in his career and is “doing $50 million a year now.” This is a current run‑rate claim as of late 2022, not an audited statement about calendar‑year 2021 performance at TheSyndicate.com.(acquired.fm)
• In a 2023 Meb Faber interview, he explains that only about half of the companies his fund invests in actually choose to do a syndicate; the syndicate thus represents only a subset of his overall capital deployment, which further complicates any attempt to infer 2021 syndicate volumes from his total investing activity.(mebfaber.com)

Because TheSyndicate.com is a private angel syndicate, there is no comprehensive, public, year‑end 2021 report of (a) exact number of syndicate deals closed or (b) total dollars deployed via those SPVs. Third‑party databases clearly undercount relative to known deals, so they can’t be used to safely declare the prediction wrong, and Jason’s later public comments don’t explicitly confirm or deny whether the 2021 target of 60 deals and ~$50M through TheSyndicate.com was actually met.

Given the lack of authoritative, complete data, the accuracy of this specific 2021 forecast cannot be determined from public information, so the result has to be rated as ambiguous rather than clearly right or wrong.