E6: Big Tech antitrust aftermath, potential effects of an M&A clampdown on Silicon Valley & more

Assessment through 29 Nov 2025
Chamath’s claim was that under the then‑existing U.S. antitrust framework, big tech platforms (Facebook/Meta, Google, Apple, Amazon) would not be broken up, and would instead mostly face regulatory/oversight regimes rather than structural divestitures. As of late 2025, that is what has happened.

1. No structural breakups of the major platforms in the U.S.

• Meta (Facebook): The FTC’s case explicitly sought divestitures of Instagram and WhatsApp. In November 2025, Judge James Boasberg ruled that Meta is not an illegal monopoly and rejected the FTC’s attempt to force a breakup, dismissing the divestiture effort.(washingtonpost.com) No court‑ordered breakup has occurred.
• Google (Alphabet): In the DOJ’s search antitrust case, a federal judge found Google liable and then moved to the remedies phase. In September 2025, the court explicitly declined to break up Google’s search business; instead, it ordered behavioral remedies (ending exclusive default search deals, limits on tying Search/Chrome/Assistant/Gemini, data‑sharing and non‑discriminatory syndication obligations), i.e., ongoing regulatory oversight rather than structural separation.(techcrunch.com)
• Amazon: The FTC’s broad antitrust suit against Amazon, filed in 2023, is still pre‑trial; a joint filing indicates trial is not expected until at least mid‑2026, so no breakup order exists.(retaildive.com) Other enforcement (e.g., a large 2025 Prime enrollment/cancellation settlement) has taken the form of monetary relief and conduct changes, not structural remedies.(the-sun.com)
• Apple: The DOJ and multiple state AGs sued Apple in March 2024 over alleged smartphone monopolization. In 2025, the court denied Apple’s motion to dismiss, allowing the case to proceed, but the litigation is still in early stages and no liability or breakup order has been issued.(michigan.gov)

Across all four firms, there has been no court‑mandated spin‑off or breakup compelled under existing U.S. antitrust statutes as of November 2025.

2. Strong shift toward regulatory and behavioral regimes

While breakups have not materialized, these companies are increasingly subject to regulatory‑style rules and oversight, especially abroad but also via U.S. conduct remedies:

• The EU Digital Markets Act (DMA) has designated Alphabet, Amazon, Apple, and Meta as “gatekeepers,” imposing detailed obligations on app stores, self‑preferencing, data‑combining, interoperability, and pre‑installation, with ongoing compliance reporting and potential fines up to 10% of global turnover.(digital-strategy.ec.europa.eu) Enforcement has already produced non‑compliance investigations and substantial fines for Apple and Meta in 2025, requiring changes to App Store rules and consent models.(digital-strategy.ec.europa.eu)
• In the U.S., the Google search case remedy package focuses on behavioral changes and monitoring (barring exclusivity deals, mandating data access and fair dealing with rivals) rather than structural breakup.(techcrunch.com) Similar patterns—injunctive relief, conduct constraints, and long‑running oversight—define current and prospective remedies against Apple and Amazon as described by DOJ/FTC filings and commentary.(cnbc.com)

3. Alignment with the prediction

Chamath argued in mid‑2020 that, given the antitrust framework at the time, big‑tech platforms would likely avoid being broken up, and that the real pressure would come from more regulation/oversight rather than structural antitrust remedies. More than five years later:

• No major U.S. court or agency has forced Facebook/Meta, Google, Apple, or Amazon to divest core businesses.
• The most significant developments have been new regulatory regimes and behavioral remedies, particularly in the EU (DMA) and in U.S. antitrust cases, which focus on rules, constraints, and supervision rather than dismantling the firms.

Future cases (especially against Apple and Amazon) could still, in theory, produce structural remedies, but judging by outcomes to date—especially Meta’s win against an attempted breakup and Google’s avoidance of structural separation—Chamath’s prediction has been accurate so far.

Conclusion: The prediction has come true to date: big tech platforms have not been broken up under existing U.S. antitrust statutes and have primarily faced regulatory and behavioral constraints instead.

On balance, subsequent antitrust/competition enforcement has roughly matched Chamath’s ranking: Meta/Facebook has faced the heaviest global antitrust pressure, Google substantial but slightly less, Apple somewhat below Google, and Amazon—while clearly targeted—has still seen comparatively lighter competition remedies and outcomes.

Meta / Facebook (predicted: highest risk)

• In December 2020 the FTC, joined by 46 states, sued Facebook (now Meta) for illegal monopolization of personal social networking, seeking structural relief including potential divestiture of Instagram and WhatsApp; the case is ongoing and went to trial in April 2025. (ftc.gov)
• The U.K. Competition and Markets Authority ordered Meta to unwind its completed acquisition of Giphy, forcing a full divestiture—widely noted as the first time a major regulator unwound a Big Tech deal. (cnbc.com)
• In 2024, the European Commission fined Meta about €797.7 million for abusing its dominance by tying Facebook Marketplace to Facebook and imposing unfair conditions on rivals, and ordered it to stop the behavior. (dw.com)
• In 2025, the EU fined Meta a further ~€200 million under the Digital Markets Act (DMA) for its “consent or pay” ad model, and national authorities in countries such as France and Italy opened additional competition probes into Meta’s advertising and WhatsApp/AI practices. (apnews.com)
  Assessment: Meta has clearly been a top target for structural antitrust remedies and very large competition fines—consistent with “highest risk.”

Google (predicted: second-highest risk)

• The DOJ and 11 states sued Google in October 2020 for maintaining monopolies in search and search advertising; in August 2024 a federal judge held that Google violated Section 2 of the Sherman Act. (justice.gov)
• A second DOJ case filed in 2023 targets Google’s ad-tech stack; in April 2025 the court ruled that Google formed an illegal monopoly in digital advertising and signaled that significant structural remedies (such as spinning off parts of the ad-tech business) may be required. (en.wikipedia.org)
• Google has also continued to face multi-state U.S. suits and long‑running EU antitrust enforcement with multi‑billion‑euro fines in prior years, reinforcing its status as a core focus of global competition policy. (ag.ny.gov)
  Assessment: Multiple, large-scale DOJ cases with adverse liability findings, layered on top of EU actions, place Google near the top of the enforcement‑risk spectrum, though Meta’s breakup‑style scrutiny arguably edges it out.

Apple (predicted: third‑highest risk)

• In March 2024 the DOJ and 16 state AGs filed a sweeping Section 2 case accusing Apple of monopolizing smartphone markets by restricting interoperability and foreclosing rival apps/services in the iPhone ecosystem; the complaint explicitly contemplates structural remedies. (justice.gov)
• In March 2024 the European Commission imposed Apple’s first major EU antitrust fine—about €1.84 billion—for anti‑steering rules that prevented music‑streaming apps (e.g. Spotify) from telling users about cheaper subscriptions outside the App Store. (dw.com)
• In April 2025 the EU fined Apple another ~€500 million under the DMA for restricting app distribution and steering, while India’s competition authority has an ongoing abuse‑of‑dominance investigation that Apple says could theoretically expose it to a very large turnover‑based fine. (apnews.com)
  Assessment: Apple has become a major enforcement target, but its big U.S. antitrust case and principal EU fine land later and in somewhat narrower product markets than Google’s long‑standing search/ad-tech dominance cases; its overall exposure appears substantial but plausibly below Google’s and Meta’s.

Amazon (predicted: lowest risk / “most inoculated”)

• In the EU, Amazon settled two antitrust investigations in December 2022—over use of marketplace sellers’ data and preferential treatment in the Buy Box—by offering commitments on ranking, access to Prime/Buy Box, and data use. The Commission accepted behavioral remedies; no fine was imposed, but non‑compliance could trigger penalties up to 10% of global turnover over the next 5–7 years. (techcrunch.com)
• In September 2023 the FTC and 17 states filed a landmark monopolization suit against Amazon, alleging it uses interlocking strategies to maintain monopoly power in the online “superstore” and marketplace services markets, and seeking a permanent injunction and structural relief. A judge allowed most claims to proceed; trial is scheduled for 2026–27. (ftc.gov)
• Separately, the FTC sued Amazon in 2023 over “dark patterns” in Prime sign‑up and cancellation. That case (a consumer‑protection/ROSCA enforcement, not classic antitrust) ended in a 2025 settlement where Amazon agreed to pay $2.5 billion (including a $1 billion civil penalty) and issue about $1.5 billion in customer refunds, plus interface changes. (ftc.gov)
  Assessment: Amazon clearly is under serious regulatory scrutiny, including a major U.S. monopolization case. However, compared with its peers by late 2025:
• It is the only one of the four that has not yet been found liable for an antitrust violation by a U.S. or EU authority (Meta has an EU antitrust judgment and U.K. divestiture order; Google has two adverse DOJ liability findings; Apple has multiple EU competition/DMA fines). (dw.com)
• Its EU matters were resolved via commitments rather than immediate billion‑euro fines, and the main U.S. antitrust case remains pending without a remedial order, whereas Meta, Google, and Apple are already subject to large penalties or structural obligations. (techcrunch.com)
• The largest dollar hit to date (the Prime settlement) arises from consumer‑protection law, not from competition‑law monopolization findings.

Overall judgment
While all four companies have faced aggressive enforcement since mid‑2020, the realized competition‑law outcomes support the relative ordering Chamath suggested:

• Meta/Facebook: breakup‑style FTC case + EU/UK structural remedies and massive fines → clearly top tier of antitrust risk.
• Google: two major DOJ cases with liability findings and potential structural break‑up of core ad/search businesses → very high risk, likely second only to Meta.
• Apple: one major DOJ case plus significant but more targeted EU competition/DMA fines → substantial but somewhat lower relative risk.
• Amazon: one U.S. monopolization case still pre‑judgment, EU cases resolved via commitments, and major but non‑antitrust consumer‑protection enforcement → material scrutiny but, so far, the least severe antitrust outcomes of the four.

Because Amazon has indeed faced comparatively lighter antitrust/competition sanctions and findings than Meta, Google, and Apple over the ensuing years, Chamath’s directional prediction—that Amazon would be the most “inoculated” and Facebook the most exposed, with Google then Apple in between—is broadly right.

Why this prediction is wrong

Chamath’s claim was that, starting around 2020 and "for the foreseeable future," large M&A by the Big Four (Facebook/Meta, Google/Alphabet, Apple, Amazon) would be impossible due to antitrust/regulatory pressure, implying a de facto halt to major acquisitions.

In the 5+ years since that July 2020 statement, multiple large, multi‑billion‑dollar acquisitions by these firms have not only been announced but successfully closed despite heavy scrutiny:

• Amazon

  • Closed its $8.45 billion acquisition of MGM in March 2022 after U.S. and EU review; the FTC let the waiting period expire without blocking the deal. (forbes.com)
  • Closed its $3.9 billion acquisition of primary‑care provider One Medical in February 2023; the FTC investigated and issued a second request but ultimately did not stop the merger. (cnbc.com)
  • Its attempted $1.4 billion acquisition of iRobot was blocked in practice by U.S. and EU opposition and then abandoned in January 2024, showing scrutiny is real but not an absolute bar. (en.wikipedia.org)

• Google / Alphabet

  • Completed the $5.4 billion acquisition of cybersecurity firm Mandiant in September 2022 after DOJ and other antitrust reviews concluded without blocking the deal. (en.wikipedia.org)
  • Agreed in 2025 to acquire cybersecurity startup Wiz for $32 billion, its largest deal ever; the transaction is moving through regulatory review, with reports that DOJ has cleared it to close in 2026. (ft.com)

• Meta (Facebook)

  • Announced the purchase of Within Unlimited (VR fitness app Supernatural) for roughly $400 million. The FTC sued to block the deal, but a federal judge denied the injunction; the FTC eventually dropped its challenge and the deal closed in February 2023. (cnbc.com)
  • Meta also closed its acquisition of customer‑service software firm Kustomer, first announced in 2020, after conditional EU approval in early 2022. (euronews.com)
  • Conversely, Meta was forced to divest Giphy by the UK CMA, showing heightened enforcement but again not a total ban on other deals. (techcrunch.com)

• Apple

  • Apple has faced escalating antitrust actions (e.g., U.S. v. Apple in 2024, German and EU gatekeeper designations) but continues to make smaller, strategic acquisitions; it has chosen not to pursue mega‑deals recently, rather than being uniquely barred while peers are not. (en.wikipedia.org)

Regulatory climate vs. prediction content

Antitrust scrutiny against Big Tech has unquestionably intensified since 2020 (multiple DOJ/FTC suits against Google and Meta, new cases against Apple, and abandoned deals like Amazon–iRobot). (en.wikipedia.org) But Chamath’s prediction was stronger: he said large M&A by the Big Four was "impossible" and implied a sustained halt to major acquisitions.

The actual record shows:

• Several very large deals (>$3–5B) by Amazon and Google completed post‑2020.
• Medium‑sized tech deals (hundreds of millions to low billions) by Meta proceeded despite direct agency lawsuits, with courts sometimes siding against regulators (e.g., Within Unlimited).
• Regulators have blocked or deterred some deals, but not to the point that “large acquisition and M&A … is impossible” for these firms as a group.

Because multiple Big Four companies have completed exactly the kind of large acquisitions that the prediction said would be impossible, and they did so within a few years of the claim under intense modern antitrust scrutiny, the prediction is wrong.

Evidence since 2020 shows a brief IPO boom rather than the sustained structural shift toward public exits for emerging‑growth U.S. tech companies that Chamath predicted.

1. Overall U.S. IPO activity:

   • U.S. IPOs surged to 480 in 2020 and a record 1,035 in 2021 (driven heavily by SPACs), but then collapsed to 181 in 2022, 154 in 2023, and roughly the low‑200s in 2024, far below the 2021 peak. (stockanalysis.com)
   • Analysis from a major asset manager notes that after 2021, IPO activity fell below the 20‑year average, contradicting the idea of a persistently more active public market vs the 2000–2020 period. (ft.com)

2. Tech‑specific IPOs vs 2000–2020:

   • For U.S. tech IPOs, data compiled by Visual Capitalist show 261 in 2000, then mostly double‑digit counts each year from 2001–2019; total tech IPOs over 2000–2019 are 947, an average of about 47 per year. (visualcapitalist.com)
   • Post‑prediction: tech IPOs were 48 in 2020 and 126 in 2021, but only 6 in 2022 and 9 in 2023, among the lowest on record. (visualcapitalist.com)
   • Averaging 2020–2023 (48, 126, 6, 9) yields roughly the same annual tech‑IPO volume as 2000–2019, not a clearly higher and more supportive regime. Excluding the one‑off 2021 spike, the post‑2020 years are markedly worse than the 2000–2020 period.

3. Many major growth tech firms remain private:

   • Reporting highlights that leading growth companies such as SpaceX, Stripe, and OpenAI increasingly choose to remain private, with private markets exceeding $5.7 trillion and supplying ample late‑stage capital, while U.S. IPO counts in 2024 are far below 2021 levels. (investopedia.com)
   • Other coverage notes that the number of listed companies in the U.S. is shrinking and IPO activity has been weak, due to abundant private capital and listing frictions, again implying no broad shift toward public markets as the primary growth‑company exit. (thetimes.co.uk)

4. Big‑tech M&A not fully shut, undermining the premise:

   • While antitrust scrutiny of dominant platforms clearly increased (new DOJ/FTC merger guidelines and high‑profile suits against Google, Meta, Apple, and attempts to block deals such as Nvidia–Arm and Microsoft–Activision), several large tech acquisitions still closed, and law‑firm and policy analyses describe a slowdown and higher scrutiny, not the effective disappearance, of big‑tech M&A. (aoshearman.com)

Putting this together: there was a short‑lived, pandemic‑era IPO/SPAC boom that partially fits his thesis, but from 2022 onward IPO volumes (including tech) have been below long‑run averages and many top emerging‑growth tech firms continue to rely on private capital instead of going public. Over the multi‑year horizon since 2020, U.S. public markets have not become sustainably more active or more supportive of emerging‑growth tech listings than in 2000–2020, so the prediction is best judged wrong overall.

Summary The prediction overstated both the scale and the persistence of any antitrust‑driven pullback. There is evidence of some regulatory “chill” on certain startup deals and a temporary dip in Big Tech startup acquisitions around 2023, but acquisition activity by the major platforms did not stay substantially reduced, nor did Silicon Valley suffer the kind of sustained, regulation‑driven “disaster” implied.

1. Did the major platforms “substantially reduce” acquisition activity?

Immediate and medium‑term behavior (2020–2022) contradicts a sharp pullback:

• Post‑hearing, Big Tech continued to pursue very large deals. Examples include Microsoft’s acquisition of Activision Blizzard for ~$69B (announced 2022, closed 2023) and other mega‑deals that required intensive regulatory review but still went through.(en.wikipedia.org)
• Sector‑wide, 2021–2022 were boom years for tech M&A overall; PwC/Dealogic data show that $500M+ tech, media & telecom deals were actually more numerous under Biden than under Trump, and median time to close rose by only one day (76→77 days), indicating no broad freeze in large deals.(theregister.com)

There was a notable dip around 2023, but it was not permanent and had broader causes:

• Global TMT M&A value fell 46% and deal volume 26% in 2023 vs 2022, driven largely by higher interest rates and weaker growth, with regulatory scrutiny cited as one of several contributing factors.(verdict.co.uk)
• A Crunchbase analysis in mid‑2023 found that the five most valuable U.S. tech companies (Apple, Microsoft, Google, Amazon, Nvidia) had made just five known startup acquisitions year‑to‑date, putting 2023 on pace for the smallest number of startup acquisitions in years.(news.crunchbase.com) This supports a temporary reduction, especially for small startup deals.

But by 2024–2025, Big Tech was again very active in M&A:

• One 2024 summary estimated that the five major tech firms collectively spent over $127B on acquisitions and strategic investments in 2024, a ~340% increase over pre‑pandemic levels, with 73% of that focused on AI‑related technologies.(konceptual.ai)
• Alphabet/Google went on to announce its largest acquisition ever—the $32B purchase of Wiz in 2025—after an earlier attempt at a ~$23B deal was shelved due to antitrust concerns, showing that even intense scrutiny did not stop it from pursuing huge M&A.(reuters.com)

Taken together, this is inconsistent with the claim that the major platforms, in general, would sustainably reduce acquisitions. There was a cyclical slump and some hesitation around certain contested deals, but not a lasting withdrawal from M&A.

2. Was there a “sustained chilling effect” on M&A that harmed the startup ecosystem via fewer acquisition exits?

Evidence that antitrust policy increased friction and chilled some startup deals:

• Legal and academic analyses document that, under Lina Khan (FTC) and Jonathan Kanter (DOJ), challenges to startup acquisitions rose sharply: only three startup acquisitions were challenged between 2012–2019 vs fourteen between 2020–2023. High‑profile cases or threats of suit led parties to abandon deals like Adobe/Figma, Qualcomm/Autotalks, Sanofi/Maze—and an earlier Google/Wiz attempt. These analyses explicitly describe a “chilling effect” spreading across Silicon Valley, making acquisitions by large incumbents riskier and less attractive as an exit path.(corpgov.law.harvard.edu)
• A related NYU‑Law–based policy piece similarly argues that expanded merger review and higher uncertainty have made acquisitions more costly, lengthier, and less appealing, and warns that this can reduce capital flowing into the startup sector by weakening M&A as an exit option.(pulseforinnovation.org)

But the broader startup‑exit slump has larger macro causes and isn’t uniquely driven by Big Tech pullback:

• PitchBook data cited by Reuters show that global VC exit value in 2023 fell to a six‑year low ($234.3B), with U.S. VC exits at their weakest since 2016, reflecting high interest rates, a frozen IPO market, and risk‑off investor sentiment.(reuters.com) These macro factors affected all exit routes, not just Big Tech acquisitions.
• At the same time, FTC statistics and subsequent commentary indicate that the share of deals receiving intensive merger scrutiny (second requests) has stayed around 1–2% over the last decade, and that the perception of a broadly hyper‑aggressive FTC is driven by a few very high‑profile cases rather than a systemic clampdown on most deals.(forbes.com)

So while there is credible evidence of a non‑trivial antitrust‑driven chill on some startup M&A—especially high‑profile or strategically sensitive deals—the data do not support the prediction of a broad, sustained collapse of acquisition exits uniquely attributable to the 2020 antitrust push.

3. Overall assessment relative to Sacks’s prediction

Sacks predicted that the July 2020 House antitrust hearings would cause the big platforms to systematically second‑guess “every acquisition,” triggering a lasting chill in M&A and a “disaster” for the Silicon Valley startup ecosystem via fewer acquisition exits.

What actually happened:

• Big Tech continued to do major acquisitions in 2020–2022 and then resumed very large and numerous deals again by 2024–2025, including record‑setting transactions. There was a dip in small/startup acquisitions around 2023, but it was not permanent, nor did Big Tech exit the M&A market.(verdict.co.uk)
• Startup exits (IPOs + M&A) did fall sharply in 2022–2023, but largely in tandem with rising rates and a depressed IPO market. Antitrust policy contributed to the difficulty of some specific deals and created real uncertainty, yet the overall pattern of exits is better explained by a broad funding and macro downturn than by a Big Tech acquisition strike alone.(reuters.com)

Because Big Tech did not maintain a long‑term, across‑the‑board reduction in acquisitions, and because the startup ecosystem’s challenges cannot be cleanly or primarily attributed to that kind of M&A freeze, the core prediction—of a sustained, regulation‑driven collapse in platform acquisitions that would be a “disaster for Silicon Valley”—has not clearly materialized.

Hence the prediction is best classified as wrong, albeit with a kernel of truth that antitrust developments did create some localized chilling effects on specific startup deals and contributed at the margin to a more challenging exit environment.

How the prediction was framed
Chamath argued in mid‑2020 that if Big Tech M&A was effectively “off the table” due to antitrust, then:

1. Large‑cap tech companies would no longer reliably buy late‑stage startups at a ~2× premium to the last round.
2. As a result, late‑stage private tech valuations would go “way down” relative to the 2010s, since investors could no longer underwrite that strategic M&A markup.

To evaluate this, we need to check (a) whether Big Tech M&A really became effectively impossible, and (b) how late‑stage valuations in 2022‑2024 compare to the 2010s.

1. Big Tech M&A did not go “off the table”

Regulatory pressure and antitrust enforcement did increase meaningfully after 2020:

• US and EU authorities brought major antitrust cases against Google, Meta, Apple and others, and blocked or forced abandonment of some large deals such as Nvidia–ARM and Adobe–Figma.(en.wikipedia.org)
• Microsoft’s acquisition of Activision Blizzard faced intense global scrutiny and delay, but ultimately closed in October 2023.(en.wikipedia.org)
• Regulatory scrutiny contributed to a sharp slowdown in mega‑cap tech acquisitions: the six biggest tech companies (Apple, Microsoft, Alphabet, Amazon, NVIDIA, Meta) made only 9 acquisitions in 2023 vs. 33 in 2022, and overall tech mega‑deals (> $10B) fell sharply.(cooleyma.com)

However, “off the table” is much stronger than “slowed”:

• An NGO study finds Big Tech firms (Apple, Microsoft, Alphabet, Amazon, Meta) acquired at least 191 companies between 2019 and 2025, averaging a new acquisition roughly every 11 days, even though deal counts fell after 2022.(somo.nl)
• Big tech still executed or announced large strategic deals: Amazon–MGM (content), Amazon–One Medical (healthcare), Microsoft–Activision (gaming), and Alphabet’s planned $32B acquisition of Wiz (cloud security), among others.(helprange.com)

So while enforcement clearly chilled some big transactions and reduced volumes, M&A was not functionally eliminated as an exit path for startups. The prediction’s key precondition (“M&A off the table”) did not fully materialize.

2. Late‑stage private valuations vs. the 2010s

Chamath’s quantitative claim was that late‑stage private valuations would go “way down” once this clampdown took effect, relative to the 2010s era when investors underwrote 2× strategic exits.

What actually happened:

1. Huge boom first (2020–2021):

   • Late‑stage VC valuations surged to record highs in 2020–2021, far above 2010s levels, despite antitrust talk already intensifying. PitchBook’s US VC data show average late‑stage pre‑money valuations jumping from about $134–154M in 2018–2019 to $170M in 2020 and then ~$339M in 2021.(scribd.com)

2. Correction from peak (2022–2023):

   • Rising interest rates, public‑market tech sell‑offs and a shut IPO window led to a sharp markdown in late‑stage deals in 2022–2023. Median Series D/E valuations on Carta fell 55–80% year‑over‑year at the trough in Q1 2023.(carta.com)
   • PitchBook’s 2023 US VC Valuations report notes that late‑stage median valuations fell and step‑up multiples compressed, but the 2023 annual median late‑stage pre‑money valuation was still the third‑highest of the last decade, i.e., above most 2010s years.(scribd.com)
   • A Q4‑2023 summary based on PitchBook data says explicitly: late‑stage valuations in 2023 fell to a six‑year low, meaning roughly back to ~2017 levels—not structurally below the 2010s baseline.(pilot.com)

3. Still not ‘way down’ vs 2010s baseline:

   • The same PitchBook time series shows average US late‑stage VC pre‑money valuations by year: roughly $100–154M in 2014–2019, versus $214.7M in 2023 and $150.9M in early 2024.(scribd.com)
   • So even after the correction, late‑stage valuations have been similar to or higher than late‑2010s levels; they are certainly not “way down” relative to the 2010s. The main change is that they are down from the extraordinary 2021 bubble, not from the pre‑2020 regime he was referencing.

There is evidence that venture‑backed tech M&A exit prices in 2023 were often below the last private round (median exits ~30% below prior funding rounds), reflecting stress at late stage.(cooleyma.com) But that’s a cyclical compression from bubble highs rather than a structural reset to levels meaningfully below the 2010s.

3. Causality vs. Chamath’s thesis

Chamath’s mechanism was: antitrust → Big Tech can’t buy → no 2× strategic exits → investors slash late‑stage valuations.

The data and deal history suggest a different story:

• M&A was constrained but not eliminated; Big Tech still does acquisitions and has even announced new multi‑billion‑dollar startup deals in 2024–2025.(somo.nl)
• The sharp valuation reset tracks macro factors (interest‑rate hikes, public tech multiple compression, IPO freeze) at least as closely as it tracks M&A antitrust. Reports from CB Insights, PitchBook, and Carta all emphasize rates, public comps, and liquidity as primary drivers of the 2022–2023 pullback.(cbinsights.com)
• And even after that pullback, late‑stage valuations remain around or above their late‑2010s range, not materially below it.

Conclusion

Because:

• Big Tech M&A was slowed but far from “off the table,” and
• Late‑stage private valuations, while down sharply from 2021 bubble highs, have not gone “way down” relative to 2010s levels and in 2023 were still among the highest years of the decade,

Chamath’s prediction about a structural, antitrust‑driven collapse in late‑stage private tech valuations (relative to the 2010s) has not come true.

Verdict: wrong.

Regulators did in fact become substantially more aggressive about blocking or conditioning acquisitions by large internet and tech platforms in the early‑to‑mid 2020s, matching the dynamic Chamath was describing.

In Europe and the U.K., competition authorities forced Meta to unwind its completed acquisition of Giphy on the grounds that it would further increase Meta’s already significant power in social media and display advertising markets — the first time a Big Tech firm was ordered to divest a closed deal on competition grounds. (cnbc.com) Regulators also pushed Amazon to abandon its $1.4–1.7 billion acquisition of iRobot over concerns that Amazon could use its dominant marketplace to disadvantage rival device makers, and their stance was widely framed as part of growing global antitrust scrutiny of big tech M&A. (apnews.com) Likewise, Adobe and Figma terminated Adobe’s $20 billion takeover after the EU Commission and U.K. CMA signaled they were likely to block it because it would remove an emerging rival and entrench Adobe’s position in design and creative software. (apnews.com) All of these are exactly cases where regulators acted to prevent already‑large tech firms from compounding their advantages via acquisitions.

Major platform deals that did proceed often faced unusually heavy conditions. The U.K. CMA initially moved to block Microsoft’s $68.7 billion acquisition of Activision Blizzard to prevent Microsoft from using the deal to dominate cloud gaming, then only cleared a restructured transaction in which Activision’s cloud‑streaming rights were divested to Ubisoft and subject to long‑term licensing commitments designed to keep the market open. (cnbc.com) This is a textbook example of regulators insisting on structural changes to stop a dominant platform from further entrenching itself in a fast‑growing adjacent market.

U.S. agencies also shifted toward challenging more ecosystem‑strengthening acquisitions by dominant tech firms. The FTC sued to block Meta’s purchase of VR fitness app maker Within, arguing Meta was trying to “buy its way to the top” instead of competing on the merits and that acquiring Within would lessen innovation and choice in VR fitness — a case aimed squarely at Meta extending its existing VR platform. (cnbc.com) The same FTC and the DOJ have brought major monopolization and merger cases against Google, Meta, Amazon and Microsoft that explicitly seek to prevent further entrenchment of already‑large platforms, even where courts have ultimately been skeptical.

Beyond individual cases, regulators introduced structural tools specifically aimed at preventing gatekeeper platforms from reinforcing their power through smaller deals. The EU’s Digital Markets Act designates the largest online platforms as “gatekeepers” and is explicitly intended to keep digital markets contestable and fair, limiting conduct by entrenched firms that would let them leverage dominance into adjacent services. (en.wikipedia.org) Separately, the European Commission’s 2021 guidance on Article 22 encouraged member states to refer below‑threshold mergers — particularly in tech and pharma — to catch so‑called killer acquisitions of small but strategically important rivals by incumbents. (skadden.com) Even after the EU’s top court curtailed this specific tool, the Commission and national authorities signaled they would look for other ways to scrutinize such deals, and by 2025 senior officials were openly discussing bringing Big Tech “acquihire” transactions under merger review for the same reason. (reuters.com) These policy moves are motivated by exactly the fear Chamath described: that unchecked acquisitions let dominant platforms build compounding advantages that regulators now see as their responsibility to limit.

While enforcers have not won every case — U.S. courts, for example, ultimately rejected the FTC’s bid to unwind Meta’s Instagram and WhatsApp deals — the overall pattern since 2020 is clearly toward more frequent, more aggressive attempts to stop or heavily condition acquisitions that would let already‑dominant tech platforms further entrench their ecosystems. That observed trend aligns closely with Chamath’s prediction that regulators would increasingly act to prevent such compounding advantages through M&A.

Evidence since 2020 shows a clear shift toward treating core internet services as essential / critical infrastructure and subjecting large platforms to sector‑specific regulation, not just traditional antitrust, in several major jurisdictions.

1. Explicit "critical / essential infrastructure" treatment of digital services (EU)

• The EU’s revised cybersecurity framework (NIS2 Directive) classifies “digital infrastructure”—including cloud computing providers, data centers, internet exchange points, DNS providers, and public electronic communications networks—as a sector of high criticality, alongside energy, transport, banking, and health, with mandatory risk‑management and incident‑reporting obligations. (twobirds.com)
• NIS2 and its 2024 implementing rules extend similar cybersecurity and reporting requirements to online marketplaces, online search engines, and social networking platforms, treating them as critical entities or networks whose disruption has systemic impact and must be overseen by national authorities. (digital-strategy.ec.europa.eu)
  These moves are structurally analogous to how aviation, energy or transport infrastructure is regulated.

2. Sector‑specific ex ante regimes for large platforms (EU)

• The Digital Markets Act (DMA) creates an EU‑wide, ex ante regulatory regime for designated “gatekeepers” providing core platform services (search engines, app stores, social networks, video‑sharing platforms, communication apps, operating systems, ad services, etc.), imposing ongoing conduct rules and interoperability/data‑access obligations that go well beyond case‑by‑case antitrust. (en.wikipedia.org)
• The Digital Services Act (DSA) adds a parallel, risk‑based regulatory framework for “Very Large Online Platforms” and “Very Large Online Search Engines,” requiring systemic‑risk assessments (elections, public security, health, protection of minors), transparency reports, algorithmic auditing and data access for regulators and researchers. (en.wikipedia.org)
• These acts are actively enforced with investigations and large fines against Apple, Meta, Google, Amazon and others, confirming that large internet services are now under an ongoing sector‑specific regulatory regime, not just classic antitrust enforcement. (apnews.com)

3. UK: broadcast‑style regulator extended to online services

• The UK’s Online Safety Act 2023 mandates that many internet services (social networks, search, user‑to‑user platforms, some porn sites) meet statutory duties of care to manage harmful and illegal content; Ofcom is empowered to fine up to 10% of global turnover or block access, much like a broadcasting and communications regulator. (en.wikipedia.org)
  This is a sector‑specific regime for core online services, analogous in structure to historic radio/TV regulation.

4. United States: partial convergence, but still mainly antitrust‑driven

• The U.S. continues to rely primarily on antitrust enforcement against dominant platforms (e.g., FTC v. Meta over social‑media monopoly; DOJ cases against Google in search and adtech), which are traditional competition‑law tools rather than a new infrastructure‑style regime. (en.wikipedia.org)
• However, legislative proposals like the Digital Platform Commission Act explicitly call for a new, independent sector‑specific regulator (an “FCC for digital platforms”) with authority to designate “systemically important digital platforms” and set ongoing rules—illustrating that policymakers increasingly conceptualize large platforms as infrastructure‑like and in need of bespoke oversight, even though these bills have not passed. (bennet.senate.gov)
• Separately, U.S. homeland‑security policy long designates Communications and Information Technology as critical infrastructure sectors, covering internet backbone and major providers, though this focuses on security/resilience more than economic conduct rules. (cisa.gov)

Assessment
By late 2025, major governments—especially in the EU and UK—do treat core internet services and infrastructure (cloud, DNS, data centers, platforms, search, social networks, marketplaces) as critical or highly critical sectors and have imposed ongoing, sector‑specific regulatory regimes on the largest platforms that resemble utility / transport / broadcasting regulation more than case‑by‑case antitrust. The U.S. has moved more slowly and remains mainly antitrust‑centric, but the conceptual shift toward utility‑style oversight is explicit in current legislative proposals.

Given that the prediction was broad (“over the coming years, governments will come to treat…”) rather than confined to one country, and that in some of the world’s most important regulatory jurisdictions this shift has clearly occurred, the prediction is best judged as broadly right, even though implementation is uneven across countries.

Summary
By late 2025, a large and durable slice of news and commentary has moved to bottom‑up, subscription‑driven platforms (notably Substack) with individual writers monetizing directly. On top of that, a layer of newsletter‑based media companies and bundles—effectively “next‑generation magazines”—has clearly emerged. Some details (like true cross‑creator bundles inside Substack) haven’t materialized, and legacy outlets still dominate overall news consumption, but the core direction Chamath described is playing out strongly enough to count this as right.

1. Bottom‑up subscription journalism on Substack‑like platforms

Since 2020, Substack and similar platforms have become a major channel for journalists and commentators to leave traditional newsrooms and monetize directly via paid subscribers:

• Substack reports millions of paid subscriptions and tens of millions of active subscribers, with over 50,000 publishers making money and the top 10 publishers earning tens of millions of dollars annually. (en.wikipedia.org)
• High‑profile journalists have exited legacy outlets to launch paid Substack newsletters or Substack‑based outlets, including Bari Weiss (founding Common Sense, rebranded as The Free Press), Mehdi Hasan (Zeteo), Taylor Lorenz (User Mag), and Paul Krugman (leaving the New York Times opinion section to run an independent Substack newsletter). (en.wikipedia.org)
• Entire newsrooms or collectives have formed subscription sites that mirror the Substack model even when not using Substack’s infrastructure—for example Defector Media (ex‑Deadspin staff) and Hell Gate NYC, both funded primarily by reader subscriptions rather than advertising. (en.wikipedia.org)

This is exactly the “bottom‑up” rebuild Chamath describes: individuals and small teams going direct‑to‑reader, with audiences “voting with their subscription dollars” on which writers and outlets survive.

2. Emergence of “next‑generation magazines” and bundles on top of newsletters

Chamath also predicted that aggregation tools and bundles would form on top of these individual subscriptions, creating magazine‑like products.

We now see several concrete realizations of that idea:

• Puck is explicitly built as a bundle of star‑journalist newsletters: one subscription gives access to multiple personalities and verticals (media, politics, tech, Wall Street). Readers confirm that a single Puck membership unlocks all contributing writers’ emails, functioning as a multi‑author bundle under one paywall. (axios.com)
• The Ankler, launched as a single Hollywood newsletter, has grown into a profitable mini‑network with nearly a dozen specialized newsletters authored by a team of journalists, all under one subscription/business umbrella—very close to a “next‑gen trade magazine” layered on top of the newsletter model. (axios.com)
• The Free Press, which began as Bari Weiss’s Substack newsletter, scaled into a larger outlet with over 100,000+ paid subscribers and multimillion‑dollar annual revenue, then was acquired by Paramount/CBS, with Weiss installed as editor‑in‑chief of CBS News. This is a textbook case of a bottom‑up Substack newsletter evolving into a full‑fledged media brand with magazine‑like breadth and institutional influence. (en.wikipedia.org)

Outside Substack’s own ecosystem, these companies look almost exactly like the “next‑generation magazines and content subscriptions” Chamath described: bundles of multiple creator/newsletters, sold as a unified subscription product, built on direct reader revenue rather than mass‑market advertising.

On the tooling side, aggregation for consumption (versus billing) has also grown:

• The Substack app aggregates all of a reader’s subscriptions into a single feed with discovery, recommendations, and social‑style “Notes,” making it a de facto front‑end for multiple newsletters at once. (reddit.com)
• Third‑party tools like Readwise Reader integrate paid Substack emails into a single reading queue, so users can read all their paid newsletters in one interface. (reddit.com)

These don’t always handle payments as a unified bundle, but they do represent the “overlay aggregation tools” layer he foresaw, where multiple subscriptions are experienced together like a personalized magazine.

3. Where the prediction falls short

There are two meaningful caveats, but they don’t overturn the main thesis:

1. No platform‑level à‑la‑carte bundles on Substack itself (yet).
   Substack still generally sells subscriptions creator‑by‑creator. Users have explicitly complained that they cannot just choose, say, 5–10 authors for a single blended fee, and Substack has not shipped a formal bundle marketplace. (reddit.com)

2. “Where the world is going” is overstated in scope.
   While subscription newsletters and newsletter‑based bundles have grown dramatically, they coexist with (and are often overshadowed by) other dominant information channels like TikTok, YouTube, and legacy outlets’ own paywalled sites and apps. The entire world of “information and content” has not consolidated around Substack‑style products.

Those points mean the vision isn’t universally true, but they don’t negate the core directional prediction that journalism would significantly re‑form around direct, subscription‑driven, writer‑centric platforms and that bundled newsletter enterprises would arise on top.

4. Overall assessment

By 2025:

• Direct‑to‑subscriber platforms (Substack and similar) have become a major, durable part of the news and commentary ecosystem, with many prominent journalists and entire outlets funded primarily through reader subscriptions rather than ads. (en.wikipedia.org)
• Multi‑newsletter bundles and newsletter‑native media companies (Puck, The Ankler, The Free Press, Defector, etc.) function in practice as “next‑generation magazines,” validating the expected aggregation layer on top of individual creators. (axios.com)

Given how closely these developments match the structure of Chamath’s forecast—even if some implementation details differ and the trend doesn’t dominate all information consumption—the prediction is best judged as right in substance.

Chamath’s prediction has effectively come true. Since 2020, multiple governments have created new, platform‑specific regulatory frameworks that go beyond earlier, more general laws and directly govern how large internet and social media platforms operate and manage societal risks.

In the EU, the Digital Services Act (DSA) became law in 2022 and is now fully applicable. It establishes a comprehensive accountability regime for online intermediaries and very large online platforms (VLOPs) such as major social networks and marketplaces, including systemic risk assessments, obligations to mitigate harms (e.g., to fundamental rights and electoral processes), transparency of recommender systems, and independent audits—significantly updating and tightening the older 2000 E‑Commerce Directive framework. (fra.europa.eu) In parallel, the Digital Markets Act (DMA) introduced a new competition regime for designated “gatekeepers” like Alphabet, Meta, Apple, Microsoft, Amazon, and ByteDance, imposing binding “do’s and don’ts” on how their core platform services may leverage market power and data, again going beyond pre‑existing competition rules. (digital-markets-act.ec.europa.eu)

In the UK, the Online Safety Act 2023 created a new statutory framework that places explicit legal duties on social media companies and search services to protect users from illegal content and harms to children, enforced by Ofcom, marking a shift from largely self‑regulatory approaches to detailed operational obligations on platforms. (gov.uk) In the United States, while there is still no single federal framework of comparable breadth, several states—such as New York’s SAFE For Kids Act (regulating “addictive” social media feeds for minors), California’s SB 976 (restricting addictive feeds and notifications for minors), and Virginia’s SB 854 (limiting minors’ time on social media and requiring age verification)—have enacted targeted platform regulations addressing algorithms, feeds, and youth usage. (en.wikipedia.org)

Collectively, these developments show that post‑2020 governments did in fact implement new, more direct regulatory frameworks for major internet and social media platforms aimed at governing their operations and societal impact, consistent with Chamath’s prediction.

Chamath’s prediction has largely played out as stated.

1. Context at the time of the prediction (mid‑2020)
   When he spoke on July 31, 2020, Amazon had not yet faced a large, headline federal monopolization case in the U.S., unlike Google and (soon after) Facebook/Meta. The U.S. DOJ filed a major Sherman Act monopolization suit against Google in October 2020 over search and search advertising, and the FTC plus 46 states filed an antitrust case against Facebook (now Meta) in December 2020 over social‑networking monopoly power and the Instagram/WhatsApp acquisitions.(en.wikipedia.org) That matches his premise that the main antitrust lens at the time was focused on Google and Facebook, not yet on Amazon.

2. Escalating U.S. antitrust actions against Amazon (2021–2025)

   • In May 2021, the Washington, D.C. attorney general filed an antitrust lawsuit against Amazon alleging that Amazon illegally used and maintained monopoly power in online retail by imposing “most favored nation” style pricing restrictions on third‑party sellers, which allegedly inflated prices across the online market.(oag.dc.gov) This was an early formal move to treat Amazon as a dominant platform harming competition.
   • On September 26, 2023, the U.S. Federal Trade Commission, joined by 17 state attorneys general, filed a landmark antitrust lawsuit against Amazon. The complaint accuses Amazon of illegally maintaining monopoly power in both the “online superstore” market for consumers and the market for marketplace services to sellers, using punitive and coercive tactics (including steering rules and tying logistics services) to exclude rivals and entrench its dominance. FTC Chair Lina Khan explicitly described Amazon as “a monopolist” exploiting its monopolies and framed the case as a major effort to restore competition.(cnbc.com) This is widely characterized as a landmark Big Tech antitrust case.
   • In parallel, private and state actions have also advanced, including a large nationwide consumer antitrust class action over Amazon’s pricing and marketplace policies, certified in 2025, which alleges Amazon used its platform power to inflate prices through restrictions on third‑party sellers.(reuters.com)
   • Separately, the FTC has brought multiple additional cases and enforcement actions against Amazon (for example, over Prime “dark patterns,” resulting in a record civil penalty settlement in 2025), underscoring that Amazon is now a central focus of the agency’s tech‑platform enforcement strategy.(ft.com)

3. Comparison to how regulators treat Google and Meta
   Google and Meta remain under aggressive U.S. antitrust scrutiny: DOJ monopolization suits against Google over search and digital advertising, with a 2024 ruling that Google violated Section 2 in search; and the FTC’s ongoing monopolization case against Meta over its acquisitions of Instagram and WhatsApp, which went to trial in 2025.(en.wikipedia.org) The FTC/AGs Amazon monopolization case is structurally and rhetorically similar—a large, multi‑sovereign lawsuit accusing a dominant Big Tech platform of unlawful tactics to acquire and maintain monopoly power and seeking structural or behavioral remedies. Media and legal commentary commonly group Amazon with Google and Meta as core Big Tech antitrust targets.

4. Timing versus the prediction window
   Chamath predicted that “within 5 or 10 years” Amazon would be viewed through the same antitrust lens as Facebook and Google. The key FTC + 17‑state monopolization suit arrived in 2023, roughly 3 years after his July 2020 statement—well inside his 5–10 year window. By that point, Amazon clearly became a primary target of U.S. antitrust enforcement, comparable to Google and Meta in regulatory focus.

Given that (a) Amazon was indeed not under a comparable federal monopolization case at the time he spoke, and (b) within a few years U.S. regulators brought a major, headline monopolization suit and a cluster of related cases that place Amazon squarely in the same Big Tech antitrust crosshairs as Google and Meta, the prediction is best judged as right.

Assessment
Chamath’s claim was that, in the 2020 Big Tech antitrust cycle and its aftermath, “In terms of Amazon, I think they’re going to get away unscathed”—i.e., Amazon would avoid being broken up and avoid any meaningful sanctions arising from that wave of scrutiny.

What actually happened (2020–2025)

1. Major antitrust fines and remedies in Europe

   • In December 2021, Italy’s competition authority (AGCM) fined Amazon €1.128 billion for abusing its dominant position in e‑commerce logistics (tying marketplace advantages to Fulfilment by Amazon) and imposed behavioral remedies overseen by a monitoring trustee. (en.agcm.it)
   • In September 2025, an Italian court cut the fine amount but upheld the core abuse‑of‑dominance findings, still leaving Amazon facing a very large penalty (around €750 million after removing a 50% surcharge). (reuters.com)

2. Record EU data‑protection fine, upheld in 2025

   • In July 2021, Luxembourg’s data protection authority (CNPD) hit Amazon with a €746 million (≈$800+ million) GDPR fine over unlawful personal‑data processing for targeted advertising. (cnbc.com)
   • On 18 March 2025, Luxembourg’s administrative court rejected Amazon’s appeal and confirmed the fine and corrective measures, meaning Amazon must comply with the sanctions (subject only to any further appeal). (cnpd.public.lu)

3. Large U.S. enforcement actions, including antitrust

   • In September 2023, the FTC and 17 U.S. states filed a sweeping antitrust lawsuit accusing Amazon of illegally maintaining monopoly power in online retail and seeking injunctive and potentially structural relief to “pry loose Amazon’s monopolistic control.” The case is ongoing, but it is a central, high‑stakes antitrust action emerging from the same policy cycle that followed the 2020 House investigation. (ftc.gov)
   • Separately, in 2025 Amazon agreed to a $2.5 billion settlement with the FTC over “deceptive” Prime enrollment and cancellation practices, including a $1 billion civil penalty (the largest in FTC history) and $1.5 billion in refunds, plus mandated changes to its subscription and cancellation flows. (washingtonpost.com)

4. Context from the 2020 U.S. antitrust cycle

   • The 2020 House Judiciary antitrust investigation concluded that Amazon and other tech giants wield “monopoly power” and recommended structural separations and tougher antitrust enforcement, setting the stage for later actions even though Congress did not immediately break up Amazon. (cnbc.com)

Conclusion
Amazon has not been broken up, but across 2021–2025 it has faced:

• Multiple record‑scale fines (Italy antitrust, EU GDPR), each in the hundreds of millions to over a billion euros, plus ongoing behavioral obligations; and
• A landmark U.S. antitrust case seeking structural remedies, as well as a $2.5 billion FTC settlement with significant mandated practice changes.

These are substantial, high‑profile sanctions and regulatory constraints by any normal standard. Describing Amazon as having emerged from this antitrust and regulatory cycle “unscathed” or “without meaningful sanctions” is therefore inconsistent with what actually happened.

Result: "wrong"

The prediction is a counterfactual conditional: Friedberg claims that if today’s large tech platforms were forcibly broken up into many smaller, less profitable firms, then those smaller firms would be unable to fund large-scale, capital‑intensive R&D (examples he gives: Amazon’s multibillion‑dollar satellite constellation; Alphabet’s self‑driving cars), so such projects would not proceed at the same scale or pace.

As of 29 November 2025, the key antecedent of this prediction has not occurred in the United States:

• No major U.S. tech platform (Amazon, Alphabet/Google, Meta, Apple, Microsoft, etc.) has been forcibly broken up by antitrust authorities. There are ongoing lawsuits and regulatory actions (e.g., U.S. vs. Google search/ads, FTC vs. Meta, FTC/DOJ scrutiny of app stores and ad markets), but these have not resulted in structural breakups of the core firms.
• Because the structural breakup scenario hasn’t happened, we do not observe a world in which the big platforms have been split into dozens of much smaller, low‑profit firms that must self‑fund R&D. That means we cannot directly test the claim that, under that structure, large-scale projects like Kuiper or Waymo “would not get done” or would be much smaller/slower.

We can see that large, integrated platforms have continued to fund such projects since 2020 (e.g., Amazon continuing Project Kuiper; Alphabet continuing Waymo and other bets), which is consistent with Friedberg’s premise that high‑profit scale can support big R&D. But the core predictive content is about what would happen if those same firms were forcibly broken up, and that hypothetical has not been realized.

Because the necessary condition for evaluating the prediction (forcible breakup into many smaller firms) has not occurred, the prediction cannot be empirically verified or falsified at this time.

Therefore, the proper status is:

• Result: inconclusive – the world Friedberg is describing has not occurred, so the prediction cannot yet be tested in the real world.

Overall judgment: The prediction is wrong. While the stock market did rally and vaccines ultimately proved highly effective (later in 2020), the core political and epidemiological claims did not occur: Trump did not "go all in" on masks, U.S. daily COVID deaths did not fall below 250 by mid‑September, he was widely judged to have lost the first debate, and he lost re‑election.

Below is a breakdown of each element:

1. “Trump goes all in on masks” in the weeks after July 31, 2020

   • Trump briefly called mask‑wearing "patriotic" (July 20) and at a July 21 briefing urged people to wear masks when they couldn’t distance. (cnbc.com)
   • But he refused to impose a national mask mandate and explicitly opposed one, saying he wanted people to "have a certain freedom" and did not believe in a national requirement. (washingtonpost.com)
   • Through August–September he waffled publicly, saying "the mask, perhaps, helps," but also stressing that "masks have problems" and mocking Biden for frequent mask use; he continued to hold packed rallies where many supporters were unmasked. FactCheck.org concluded in late September 2020 that Trump "has not been clear" in support of masks and that his messaging was inconsistent. (factcheck.org)
     Verdict: This is false; he never adopted a strong, consistent, national pro‑mask stance that matches "goes all in."

2. “Vaccine starts to show promise right as the debates start” (late September 2020)

   • The first major Phase 3 efficacy announcements—the kind of news showing clear promise—came in November, not September: Pfizer/BioNTech reported >90% efficacy on November 9, 2020, and Moderna reported ~94.5% efficacy on November 16, 2020. (pfizer.com)
   • Earlier in 2020 there were Phase 1/2 results described as “promising,” but no major new breakthrough efficacy news landed around the first debate on September 29. The big step‑change headlines came weeks later in November.
     Verdict: The directional idea that vaccines would look promising in 2020 was right, but the timing (“as the debates start”) is wrong.

3. “Deaths plummet… under 250 deaths per day in America by September 15, 2020”

   • On September 15, 2020, U.S. cumulative COVID deaths were about 194,500, up hundreds from the day before. One daily summary (from JHU data) shows +455 U.S. deaths over the prior 24 hours, well above 250. (haitilibre.com)
   • The U.S. passed 200,000 total deaths by September 21, 2020, implying an average daily increase of roughly 700–800 deaths around that period—again far above 250/day. (en.wikipedia.org)
     Verdict: False; daily deaths did not fall below 250 by mid‑September.

4. “The market rips” in this period (Aug–Sept 2020)

   • The S&P 500 closed 3,271.12 on July 31, 2020, rising to 3,401.20 by September 15, 2020, and had already rallied more than 50% off the March lows. (statmuse.com)
   • August 2020 was the S&P 500’s best August since 1986 and saw new all‑time highs; financial press described this as a record‑setting rally that fully erased the pandemic crash losses. (cnbc.com)
     Verdict: This part is basically correct—U.S. equities did rally strongly.

5. “Trump does well in the late‑September debate and ‘sails in’ to re‑election”

   • The first presidential debate took place on September 29, 2020 in Cleveland. (en.wikipedia.org)
   • Post‑debate polls found viewers thought Biden won by a wide margin: a CNN/SSRS poll reported 60% of viewers thought Biden won vs. 28% for Trump; a CBS poll showed Biden 48% vs. Trump 41%. Commentators across outlets called the debate a "disgrace" and "the worst presidential debate" they had seen. (en.wikipedia.org)
   • In the November 3, 2020 election, Joe Biden won 306–232 in the Electoral College and about 51.3% to 46.8% in the popular vote, defeating Trump. (dashpolitics.com)
     Verdict: False; Trump was widely judged to have lost the first debate and he did not win re‑election.

Synthesis
Out of the chain of events Jason predicted:

• Masks: wrong.
• Vaccine timing around the first debate: largely wrong on timing.
• Deaths below 250/day by mid‑Sept: clearly wrong.
• Market rally: right.
• Debate performance + easy re‑election: wrong.

Because the major political and public‑health claims failed and only the market piece really materialized, the overall prediction is best classified as wrong.

Time window. The prediction covers roughly August 1–October 31, 2020.

What actually happened.

Upside vaccine news in that period:

• Pfizer/BioNTech (Aug 20, 2020): Released Phase 1 data showing strong neutralizing antibody responses in younger and older adults and signaled they were on track to seek regulatory review as early as October, with plans to supply up to 100 million doses by end of 2020 and 1.3 billion in 2021. (pfizer.com)
• Moderna (Sep 29, 2020): Published Phase 1 data in older adults showing mRNA‑1273 induced high neutralizing antibody titers in age 56–70 and 71+ cohorts, with the vaccine generally well tolerated—messaged as providing “optimism” about protecting high‑risk groups. (investors.modernatx.com)
• Timeline/rollout optimism (early Sept 2020): Pfizer’s CEO publicly said the company could have initial efficacy results by late October and be ready to quickly seek FDA authorization if results were positive. Around the same time, CDC asked states to prepare for possible vaccine distribution as early as early November, reported as giving “optimism” for a vaccine while also raising concerns. (fiercepharma.com)

Downside / negative vaccine news and framing:

• AstraZeneca trial pause (Sept 8, 2020): Its Oxford AZD1222 Phase 3 trial was put on hold globally after a participant developed an unexplained serious illness, widely covered as a safety-related setback. (pharmtech.com)
• Johnson & Johnson trial pause (Oct 12, 2020): J&J’s large Phase 3 trial was likewise paused for an unexplained illness in a participant, another headline "pause" that fueled public concern about vaccine safety. (cnbc.com)
• Overall tone of U.S. COVID coverage (including vaccines): A National Bureau of Economic Research analysis of English‑language COVID‑19 articles in 2020 found that 91% of stories from major U.S. outlets were negative in tone, and that this unusually high negativity persisted even in areas with positive scientific developments like vaccine trials. (nber.org) This indicates that even when data from vaccine trials were scientifically promising, mainstream coverage tended to emphasize risks, uncertainties, and controversies rather than good news.

Evaluation against the prediction. The predictor asserted that over the next 2–3 months people would “for sure” hear more upside than downside news on vaccines. In reality:

• There were several high-profile positive trial updates and optimistic timelines from Pfizer, Moderna, and U.S. agencies. (pfizer.com)
• But there were also very salient negative events (AstraZeneca and J&J trial pauses) that dominated headlines, and robust empirical analysis shows that U.S. COVID‑19 news overall—including vaccine‑trial coverage—remained overwhelmingly negative in tone even when science was progressing. (pharmtech.com)

Given that mainstream audiences in this period were exposed to vaccine stories framed largely with caution and concern, and that the best available quantitative evidence shows persistent negativity even around vaccine trials, the claim that upside vaccine news would clearly outweigh downside news did not come true.

Summary of the prediction
On 31 July 2020, Friedberg asserted that effective COVID-19 vaccines would indeed exist as real, deployable products, and that the remaining uncertainty was about how many doses could be produced and when they would be available.

What actually happened

• In December 2020, the Pfizer–BioNTech COVID‑19 vaccine became the first to receive emergency authorization and be deployed to the public (UK emergency authorization on 2 Dec 2020; US FDA Emergency Use Authorization on 11 Dec 2020).(en.wikipedia.org)
• The Moderna mRNA‑1273 vaccine received US FDA Emergency Use Authorization on 18 Dec 2020, and began deployment shortly thereafter.(en.wikipedia.org)
• By late 2020 and through 2021, multiple vaccines (mRNA and others) were being manufactured and administered globally at scale, not just in clinical trials.
• As of August 2024, about 13.5–13.7 billion COVID‑19 vaccine doses had been administered worldwide, with over 70% of the global population receiving at least one dose, demonstrating massive real‑world deployment and large‑scale manufacturing.(en.wikipedia.org)

Assessment

• Friedberg’s key claim was existence and deployability of effective vaccines ("we're going to have a vaccine"), not just ongoing trials. Multiple such vaccines were indeed developed, authorized, and widely deployed within months of his statement.(en.wikipedia.org)
• The issues that followed were in fact manufacturing volume, distribution, and timing of rollouts across countries, exactly the dimensions he highlighted as the remaining uncertainties.(en.wikipedia.org)

Because effective, widely deployed COVID‑19 vaccines did materialize and the main challenges proved to be production and timing, Friedberg’s prediction is substantively correct.

Evidence from August–early November 2020 shows that information about COVID-19 vaccines did not "only continue to improve" nor did it on net strengthen public confidence during that window.

1. What Freiberg predicted (July 31, 2020)
He claimed that from then until November, vaccine news would only get better and provide “more and more indications of success,” implying a steadily improving information stream and rising confidence in vaccines’ effectiveness and inevitability.

2. Actual vaccine trial news, Aug–early Nov 2020

Positive / optimistic signals

• By late July 2020, large phase 3 efficacy trials for mRNA vaccines from Moderna and Pfizer/BioNTech were starting, a clear sign of scientific progress.(pmc.ncbi.nlm.nih.gov)
• By late September, U.S. health officials testified that four vaccine candidates (Pfizer, Moderna, AstraZeneca, Janssen) were in or about to enter phase 3, and they expressed “growing optimism” that one or more would prove safe and effective by late 2020 or early 2021.(fda.gov)

Negative / confidence‑reducing signals

• AstraZeneca/Oxford trial pause (September 2020): On Sept. 8–9, 2020, AstraZeneca paused global trials of its AZD1222 vaccine for a safety review after a participant developed an unexplained illness suspected to be a serious neurological event.(pharmtech.com) This was heavily covered as a setback and highlighted safety uncertainties.
• Johnson & Johnson/Janssen trial pause (October 2020): On Oct. 12, 2020, J&J temporarily paused dosing in all of its COVID-19 vaccine trials, including the large phase 3 ENSEMBLE trial, due to an unexplained illness in a participant.(jnj.com) Again, this was widely reported as a potential safety problem.
• These high‑profile pauses directly contradicted the idea that there would be only more indications of success; they were explicit signals of uncertainty and potential risk.

Key positive efficacy data arrived after the election

• The first major public announcements showing ~95% efficacy for the Pfizer–BioNTech and Moderna vaccines came on Nov. 9 and Nov. 16, 2020, after the Nov. 3 U.S. election and thus outside Freiberg’s prediction window.(pmc.ncbi.nlm.nih.gov) Up to Election Day, there was no definitive public efficacy readout yet.

3. Public confidence trends in that period

Survey data show that public confidence in a COVID-19 vaccine fell, rather than rose, by mid‑September 2020:

• Pew Research Center found that the share of U.S. adults who said they would definitely or probably get a COVID-19 vaccine dropped from 72% in late April–early May 2020 to 51% by Sept. 8–13, 2020—a 21‑point decline.(pewresearch.org)
• The same survey reported that 78% of Americans were more worried that the approval process would move too fast and not fully establish safety and effectiveness, indicating growing skepticism rather than increasing confidence.(pewresearch.org)

These attitudinal shifts occurred squarely within the July 31–early November 2020 window and are consistent with the impact of safety pauses and politicization of the vaccine timeline.

4. Assessment vs. the normalized prediction

The normalized claim is that, between July 31 and the early November election:

• The aggregate flow of vaccine information would increasingly indicate success, and
• New trial data and announcements would on net strengthen confidence that vaccines were effective and coming soon.

What actually happened:

• Scientifically, there were both positive signals (multiple candidates in phase 3, official optimism) and highly publicized negative signals (AstraZeneca and J&J safety pauses) before the election.(fda.gov)
• In terms of public reaction, measured willingness to be vaccinated declined sharply by mid‑September, indicating that, on balance, the information landscape did not strengthen overall confidence.(pewresearch.org)
• The strongest, unambiguous efficacy announcements that truly cemented confidence arrived after Election Day.

Because significant negative information emerged during the window and public confidence measurably fell rather than rose, Freiberg’s prediction that vaccine news would “only continue to improve” and provide ever‑stronger indications of success **did not come true.

Therefore, the prediction is best classified as: wrong.

Outcome check

• The prediction was that, as of July 31, 2020, Donald Trump would win the November 2020 U.S. presidential election.
• In reality, Joe Biden won the 2020 U.S. presidential election and became the 46th President of the United States, with Donald Trump losing the election.

Since the predicted winner (Trump) did not win, the prediction is wrong.

Outcome check

• The prediction (July 31, 2020) was that Joe Biden was on track to win the November 2020 U.S. presidential election and that Donald Trump was a “huge underdog.”
• In the actual 2020 U.S. presidential election held on November 3, 2020, Joe Biden won both the popular vote and the Electoral College, defeating incumbent President Donald Trump. Biden received 306 electoral votes to Trump’s 232 and a popular vote margin of over 7 million votes.

Assessment

• The core claim is probabilistic/forward-looking: that as of late July 2020, Biden was likely to win and Trump’s chances were low.
• The eventual outcome (Biden winning) matched the predicted direction of the race; the “huge underdog” characterization is consistent with polling averages at that time, which showed Biden with a significant national lead. (This is an inference from the widely reported polling context in mid-2020.)

Conclusion

• Because Biden did in fact win the 2020 election, the directional prediction that Biden was on track to win and Trump was a heavy underdog is substantively correct.

Therefore:

• result: right

Context of the prediction

Chamath said on July 31, 2020 that once policymakers decide what to do on enhanced unemployment, it would take 4–8 weeks for state systems to re‑implement and restart those payments.

The relevant 2020 "decision" matching the user’s normalization is the Continued Assistance for Unemployed Workers Act, signed into law on December 27, 2020, which extended PUA/PEUC and reinstated Federal Pandemic Unemployment Compensation (FPUC) at $300/week.(nga.org) The U.S. Department of Labor guidance specified that states had to pay the $300 FPUC supplement for weeks of unemployment beginning after December 26, 2020, with the first payable week ending January 2 or 3, 2021, depending on the state’s benefit week.(dol.gov)

What actually happened in the states

Evidence shows that many states restarted the $300 top‑up in about 1–2 weeks, not 4–8 weeks:

• A union summary and CNBC report from January 5, 2021 list more than 20 states (e.g., Alabama, Arizona, California, Connecticut, Delaware, D.C., Georgia, Idaho, Illinois, Louisiana, Maine, Maryland, Massachusetts, Minnesota, Mississippi, Nevada, New Hampshire, New York, North Carolina, Oregon, Rhode Island, Tennessee, Texas, West Virginia) as already issuing or processing the new $300/week enhancement that week—i.e., within 9 days of the law being signed.(afacwa.org)
• Delaware explicitly states that it began paying the extra $300 on January 4, 2021, the first week the benefit could be paid, for weeks ending January 2 onward.(labor.delaware.gov)
• Mississippi’s UI agency announced on January 8, 2021 that it had already begun paying the $300 FPUC supplement for the week ending January 2, 2021.(mdes.ms.gov)
• South Dakota reports it began paying the additional $300 weekly FPUC effective week ending Jan. 2, again implying implementation in roughly one week.(drgnews.com)
• Illinois’ employment department likewise notes it began paying the $300 FPUC benefit beginning January 4, 2021, covering weeks ending January 2.(wifr.com)

More generally, CNBC notes that after the CARES Act’s original $600 FPUC in March 2020, it took about a month for all 50 states and D.C. to administer the enhancement, and that for the December 2020 extension, experts expected the majority of states to have the new $300 payments out by mid‑January 2021—again, about 2–3 weeks from enactment.(cnbc.com)

States that were slower

Some states did experience longer lags for certain programs and claimant groups:

• Nebraska’s labor department said on January 13, 2021 that it anticipated all amended CARES Act programs (PUA, PEUC, FPUC, MEUC) would be fully implemented in approximately four‑to‑six weeks.(dol.nebraska.gov)
• Trackers in January 2021 showed states like Kansas still not paying the new $300 FPUC as of mid‑January while they updated systems, implying delays closer to several weeks.(savingtoinvest.com)

However, these slower implementations were the exception, not the norm, and even they tend to cluster around the lower end of Chamath’s 4–8 week window.

Assessment vs. the prediction

Chamath’s prediction was that, once the federal decision was made, it would take 4–8 weeks for state agencies to technically implement and restart the enhanced unemployment payments.

What actually occurred after the December 27, 2020 law:

• A large number of states resumed or began issuing the $300 supplemental payments within about 1–2 weeks (by January 4–8, 2021).
• Experts and contemporaneous reporting expected and observed that the majority of states would be paying by mid‑January, i.e., within 2–3 weeks of enactment, not 4–8 weeks.(cnbc.com)
• Only some states and some sub‑programs (like full PUA/PEUC extensions) took on the order of 4–6 weeks to fully implement, and 8‑week delays were not typical.

Because the prediction was framed as a general statement about how long it would take to re‑implement unemployment insurance once the decision was made, and the dominant empirical pattern across states was implementation in sub‑4‑week timeframes, his 4–8 week forecast substantially overstated the actual delay.

Conclusion

Overall, the real‑world rollout of the 2020 federal UI extension shows that most states restarted the enhanced payments in roughly 1–3 weeks, not 4–8 weeks, with only a minority hitting the lower half of his forecasted range and virtually none clustered around eight weeks. On that basis, the prediction is best classified as wrong.

Joe Biden won the November 3, 2020 U.S. presidential election, receiving 306 electoral votes to Donald Trump’s 232 and being inaugurated as the 46th President of the United States on January 20, 2021. This confirms Chamath’s prediction (made on July 31, 2020) that Biden would win the November 2020 election when forced to stake his “entire net worth.”

Key facts:

• Major news outlets and official tallies report that Joe Biden secured more than 270 electoral votes and the popular vote in the 2020 election, defeating incumbent President Donald Trump.
• Congress certified Biden’s Electoral College victory on January 7, 2021, and he was inaugurated on January 20, 2021.

Therefore, the prediction that Biden would win the November 2020 U.S. presidential election is correct.

Evidence from the aftermath of the November 2020 U.S. election indicates that the low‑probability scenario Sacks described did in fact occur in substance:

1. Serious attempt to overturn a certified election result
   After Election Day, Donald Trump and allies launched an extensive campaign to overturn the outcome, including more than 60 lawsuits (all unsuccessful), pressure on state officials to change results, and a coordinated fake‑elector scheme, all aimed at preventing Joe Biden’s lawful victory from taking effect. (en.wikipedia.org)

2. January 6 and pressure on Congress/Vice President
   These efforts culminated in the January 6, 2021 attack on the U.S. Capitol, explicitly intended to disrupt and delay the Electoral College count and pressure Congress and Vice President Pence to overturn the result. (en.wikipedia.org) This went beyond Bush v. Gore–style litigation into violence and an attempted subversion of constitutional procedures.

3. Ongoing legitimacy crisis around elections
   The false "stolen election" narrative (“Big Lie”) became the basis of a broad election‑denial movement that continues to claim U.S. elections are rigged despite lack of evidence, eroding confidence in electoral integrity. (en.wikipedia.org) Congressional testimony and polling summarized in the January 6 investigation record show that majorities of Americans by early 2022 believed U.S. democracy was “in crisis and at risk of failing” and had low confidence in honest elections—clear indicators of a legitimacy crisis. (govinfo.gov)

4. Serious political damage at home and abroad
   The Capitol attack and subsequent denialism were widely described as a grave attack on democracy, including by foreign leaders who said that armed supporters of an outgoing president undermined the principle of legitimate elections. (en.wikipedia.org)

While some officials argue the U.S. ultimately avoided a full‐blown constitutional breakdown, the combination of an attempted overturning of the election, the Jan 6 insurrection, and a persistent crisis of electoral legitimacy matches—and arguably exceeds—the "constitutional or legitimacy crisis" scenario Sacks outlined. Therefore the prediction is best scored as right.

Summary of prediction
Chamath predicted that the 2020 U.S. presidential election would:

1. Be clearly decided in the Electoral College rather than hinge on a Bush v. Gore–style recount / “hanging chad” controversy.
2. Not be razor-thin in electoral votes, but instead be “incredibly decisive on an electoral college level.”

What actually happened

• In the 2020 election, Joe Biden won 306 electoral votes to Donald Trump’s 232, a margin of 74 electoral votes.
• While there were numerous legal challenges and disputes, they did not produce a Bush v. Gore–style single-state recount crisis, nor was the outcome contingent on a tiny number of disputed ballots in one state. The result was certified across all states, and the Electoral College vote on December 14, 2020, confirmed Biden’s 306–232 victory.

Assessment vs. prediction

• “No hanging chad issue” / no Bush v. Gore–style crisis: Correct. There was litigation and controversy, but no equivalent of the 2000 Florida recount hinging on punch-card ballots and a tiny margin.
• “Incredibly decisive on an electoral college level”: Also broadly correct. A 306–232 result is a comfortable Electoral College margin and not razor-thin.

Conclusion
Given that the election was clearly resolved in the Electoral College without a razor-thin margin or a decisive single-state recount crisis, Chamath’s prediction is right.

Why this prediction is wrong

Sacks made two linked claims:

1. The 2020 election would not be close, but a Biden “blowout” where he would “run away with it.”

   • Final results: Joe Biden won about 51.3% of the popular vote to Donald Trump’s 46.9%, a margin of roughly 4.5 percentage points and about 7.1 million votes. (pewresearch.org)
   • In the Electoral College, Biden won 306–232. (presidency.ucsb.edu)
   • Key tipping‑point states (e.g., Wisconsin, Arizona, Georgia) were decided by 0.6 percentage points or less; if Trump had flipped all three, the Electoral College would have been 269–269. (pollbludger.net)
   • Analysts widely characterized this as a decisive but relatively modest victory, not a landslide akin to, for example, Reagan in 1984.

   Given the narrow margins in pivotal states and the single‑digit national margin, the outcome was not the kind of overwhelming “blowout” or runaway race that would normally render disputes moot.

2. Because it wouldn’t be close, the U.S. would “avoid the crisis” (a post‑election constitutional/legitimacy crisis).
   In reality, the aftermath produced exactly the sort of crisis he thought would be avoided:

   • Trump and allies launched a sustained campaign to overturn the results, filing around 60 lawsuits and pressuring state and federal officials to discard or reverse certified outcomes. (en.wikipedia.org)
   • On January 6, 2021, a pro‑Trump mob attacked the U.S. Capitol in an attempt to disrupt and delay the Electoral College vote count and pressure Congress and Vice President Pence to overturn Biden’s victory. (en.wikipedia.org)
   • Congress’ certification of the electoral votes was temporarily halted by violent disruption and only completed later that night, after the building was secured. (theguardian.com)
   • The House established the January 6th Committee, which concluded that January 6 was the culmination of a multi‑part effort by Trump to overturn the election and described it as an attempted coup, leading to legislative reforms (e.g., the Electoral Count Reform and Presidential Transition Improvement Act of 2022) specifically intended to prevent a repeat of this crisis. (en.wikipedia.org)

The combination of a relatively narrow, contestable outcome in pivotal states and an unprecedented attempt by the incumbent to overturn the result did produce a major legitimacy and constitutional crisis, including violence aimed at stopping the peaceful transfer of power.

Because (a) the election was not a clear “blowout” and (b) the post‑election period did feature exactly the kind of crisis he said would be avoided, Sacks’ prediction is best classified as wrong.

What was predicted

Jason speculated that there was a 5% chance that Donald Trump would drop out of the 2020 presidential race before the election in exchange for some kind of implicit or explicit deal not to prosecute him after leaving office.

What actually happened

• Donald Trump remained the Republican nominee through Election Day (November 3, 2020) and did not drop out of the race.
• He served out his term until January 20, 2021, when Joe Biden was inaugurated.
• There is no evidence of any public or reported private deal in which Trump agreed to drop out of the race in exchange for non-prosecution; instead, after leaving office he has faced multiple civil and criminal investigations and prosecutions at federal and state levels, which directly contradicts the idea of a non‑prosecution deal.

Because the event described in the prediction (dropping out in a non‑prosecution deal) did not occur, the prediction about that scenario coming to pass is wrong, even though Jason only assigned it a low probability (5%).

Outcome: The prediction was right.

Reasoning: In the November 3, 2020 U.S. presidential election, Joe Biden defeated incumbent President Donald Trump, securing both the popular vote and the Electoral College majority, and was inaugurated as the 46th President of the United States on January 20, 2021. This matches Friedberg’s revised prediction (as of July 31, 2020) that Joe Biden would win the November 2020 U.S. presidential election.

Sources: Official U.S. election results and widely reported coverage from major news outlets (e.g., AP, major U.S. networks and newspapers) all confirm Joe Biden’s victory in the 2020 presidential election.

Why this prediction is judged right

• The prediction (July 31, 2020) was that, despite the then‑current stalemate in Congress over further COVID‑19 relief, “they will eventually pass a bill” that includes additional unemployment or stimulus support.
• On December 27, 2020, the U.S. Congress and President Trump enacted the Consolidated Appropriations Act, 2021, a roughly $900 billion COVID-19 relief package. It included:
  • A $300 per week federal unemployment supplement for several months.
  • New direct stimulus checks of up to $600 per adult and $600 per child. These are exactly the kinds of unemployment extensions and stimulus the prediction was referring to.
• Subsequently, on March 11, 2021, President Biden signed the American Rescue Plan Act, which again provided:
  • Another round of direct stimulus checks up to $1,400 per person.
  • Extended and enhanced unemployment benefits (including the $300/week supplement and longer eligibility).

Because Congress and the administration did in fact pass further major COVID‑19 relief bills with both unemployment extensions and stimulus payments after the July 2020 impasse, the prediction that “they will eventually pass a bill” of this nature is fulfilled.