Meta's scorched earth approach to AI, Tesla's future, TikTok bill, FTC bans noncompetes, wealth tax

The prediction is framed as happening “within the next several years,” so the time window clearly extends beyond November 2025. Only about 1.5 years have elapsed since the April 26, 2024 podcast, so the stated horizon has not yet expired.

What we can check today:

• Global web search remains overwhelmingly dominated by Google. Statcounter data for October 2025 shows ~90% global market share for Google, with Bing around 4% and the rest split among Yandex, Yahoo, DuckDuckGo, Baidu, etc.; Meta does not appear as a distinct search engine host or engine in these breakdowns. (gs.statcounter.com) This strongly suggests Meta’s traditional web search share is still negligible as of late 2025 and certainly nowhere near “roughly 10 percentage points.”
• Meta has in fact done what the prediction assumed: it integrated its Meta AI assistant directly into the search bars of WhatsApp, Instagram, Facebook, and Messenger starting in early–mid 2024, and has continued rolling it out globally. (androidpolice.com) Meta reports hundreds of millions of monthly users for Meta AI, and it’s now present across many countries and languages inside those apps. (about.fb.com)

However, major measurement services for "global web search market share" still track classic search engines (google.com, bing.com, etc.), not in‑app AI assistant queries inside social apps, and they do not publish a distinct share for Meta AI/Meta-hosted search. (gs.statcounter.com) So we know Meta has not yet achieved a visible ~10‑point share of conventional web search, but we also know the prediction allowed for multiple future years, and there’s no hard, widely accepted metric yet that aggregates all AI-style queries inside Meta’s apps into a single comparable “search market share” number.

Because (1) the stated time horizon (“several years”) has not run out, and (2) the industry does not yet provide a clear, standardized statistic that would let us definitively say Meta will not reach ~10 percentage points in that time window, the fairest assessment is that the prediction is too early to call rather than clearly right or wrong.

The prediction window is 3–4 years after April 26, 2024, i.e., roughly April 2027–April 2028. The current date is November 30, 2025, which is well before that timeframe. Because the prediction is specifically about Meta’s search-market share by that future period, it is too early to determine whether Meta will end up with ~10% of the search market in 2027–2028.

Therefore, the outcome of the prediction cannot yet be evaluated.

Based on public information as of November 30, 2025, this prediction did not come true in the way it was framed.

1. No public, clearly branded “GPT‑5” release in the predicted window

   • The prediction (made April 26, 2024) said GPT‑5 would come out “any day, week or month” after that. That implies a release in the following few weeks to at most a few months (i.e., roughly by late summer 2024) as a distinct next‑generation model.
   • OpenAI publicly launched GPT‑4o in May 2024 as an improved GPT‑4‑class model, but they did not describe it as GPT‑5 or as a new generation beyond the GPT‑4 family. OpenAI’s own materials consistently frame GPT‑4o as part of the GPT‑4 generation (optimized for speed, cost, and multimodal usage), not as “GPT‑5.”
   • As of late 2025, there is no widely documented, clearly named GPT‑5 release matching the "any day, week or month" timeline after April 26, 2024.

2. Quality jump and full‑cycle lead over open source not demonstrably realized as predicted

   • While OpenAI models (e.g., GPT‑4o and subsequent iterative improvements) remain at or near the frontier, open‑source models have also advanced rapidly, with many competitive or specialized systems narrowing the gap rather than obviously falling a full cycle behind.
   • Because there is no clear, acknowledged GPT‑5 release in that timeframe to evaluate, the claim that it would “knock our socks off” and prove OpenAI is “a cycle ahead of the open source community” cannot be validated as stated.

Given that:

• The timing condition (release of GPT‑5 within a few weeks to a few months after April 26, 2024) was not met, and
• The substantive content of the prediction (a clearly labeled GPT‑5 with a large, obvious quality jump that establishes a full‑cycle lead) has not materialized in the way described,

the appropriate classification for this prediction is:

result: "wrong" — the key event (a timely, publicly released GPT‑5 with the described impact) did not occur within the predicted window, and more than enough time has passed to judge it.

The prediction is about a future state of the market: that only around 2029–2031 (5–7 years after April 2024) will autonomous robotaxi-style self-driving services reach even low single-digit percentages of overall ride volume. As of November 30, 2025, we are only about 1.5 years after the prediction, well before the forecast window begins (2029).

Current data shows that autonomous robotaxi deployments from companies like Waymo, Cruise (which has since paused operations in some markets after regulatory issues), and Tesla’s still-evolving FSD/robotaxi plans are limited in geographic scope and experimental/commercial pilot phases in a few cities, not yet a material share of total rides globally or nationally. However, nothing about today’s situation can confirm or falsify what will be true by 2029–2031.

Because the timeframe of the claim has not yet arrived, the correctness of the prediction cannot be judged yet. Therefore the appropriate status is: it’s too early to tell.

The prediction concerns what share of total ride‑hailing/transport rides will be autonomous 5–10 years from April 2024, i.e., roughly in the 2029–2034 timeframe. As of today (November 30, 2025), we are only about 1.5 years past the prediction date and still several years away from even the earliest point in that window (2029). It is therefore too early to evaluate whether autonomous rides will reach 5–10% of the ride‑hailing market in that period. No current market data can yet confirm or falsify a claim about 2029–2034.

Evidence from 2024–2025 is consistent with Jason’s qualitative view that robotaxis are not causing a sudden, mass displacement of human ride‑hailing, but it is too early to say whether his more specific claim of only ~1–2 percentage‑point market‑share gains per year will hold over the longer run.

What we see so far (2024–2025):

• Waymo reports more than 250,000 paid robotaxi rides per week in the U.S. and served about 4 million autonomous rides in 2024, scaling to hundreds of thousands of weekly trips across a handful of U.S. cities. (cnbc.com)
• Baidu’s Apollo Go has also reached roughly 250,000 fully driverless weekly rides and over 17 million total rides, but Baidu’s CEO has emphasized that even in a leading city like Wuhan its share of ride‑hailing is only around 1% and that reaching “meaningful” market share will take many years. (cnbc.com)
• Meanwhile, human‑driver platforms are still enormous. Uber alone handled over 11 billion trips in 2024 and more than 3.2 billion trips in Q2 2025, while Didi in China reports roughly 33 million daily transactions (≈12 billion trips per year). (investor.uber.com) Against tens of billions of human‑driver trips annually, all robotaxi services combined still account for well under 1% of global ride‑hailing volume, implying that their global market share is rising by well under a percentage point per year.

Local exceptions and why they don’t settle the prediction:

• In some constrained geofenced areas, especially San Francisco, robotaxis are making deeper inroads. Yipit data (summarized by The Driverless Digest) suggests that within Waymo’s SF operating zone, its share of ride‑hailing trips grew to roughly 22% by late 2024, comparable to Lyft in that subset of trips—faster than Jason’s “1–2 percentage points per year” heuristic. But this excludes airport and freeway trips and applies only to a limited service area and a short 15‑month window, not to an entire metro market or to the overall industry. (thedriverlessdigest.com)

Why the call is inconclusive rather than right or wrong:

• Jason’s core thesis—no abrupt, global wipe‑out of human ride‑hailing by robotaxis—matches current reality. Market leaders like Waymo and Apollo Go are scaling quickly in absolute numbers but still constitute a tiny fraction of total ride‑hailing worldwide, and even their own executives describe market‑share gains as gradual. (cnbc.com)
• However, his numerical estimate (only ~1–2 percentage‑point share gains per year once deployment begins) is a long‑run structural claim about how adoption will proceed over many years. With commercial robotaxi services only just beginning broader rollout in 2024–2025 and global market share still well below 1%, there is not yet enough multi‑year data to confirm or falsify that specific diffusion rate.

Because the pattern so far aligns with a slow, non‑abrupt transition but the multi‑year percentage‑point trajectory he described cannot yet be robustly measured, the fairest assessment as of November 30, 2025 is “inconclusive (too early)” rather than definitively right or wrong.

• What the law required: The Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA), signed April 24, 2024, required TikTok’s Chinese parent ByteDance to divest U.S. operations within 270 days (deadline January 19, 2025), with up to a 90‑day presidential extension, or face a ban via app‑store and hosting prohibitions. (en.wikipedia.org)
• Court outcome: TikTok challenged the law; the D.C. Circuit in December 2024 and the U.S. Supreme Court in January 2025 upheld PAFACA’s constitutionality, confirming that TikTok would either need a “qualified divestiture” or be effectively banned in the U.S. after the deadline. (law.justia.com)
• What actually happened to the deadline: Instead of allowing TikTok to be shut down, the President repeatedly used (and arguably stretched) statutory authority to delay enforcement of the ban: enforcement that was to apply January 19, 2025 was pushed to April 5, then June 19, then September 17, and then December 16, 2025, while the White House endorsed a divestiture framework intended to let Americans keep using TikTok. (whitehouse.gov)
• Further extensions and ongoing sale: In September 2025, President Trump approved a divestiture plan and further extended the effective ban date to January 20, 2026, drawing criticism from Senator Ed Markey for repeatedly extending deadlines rather than letting the ban take effect. (reuters.com)
• Status as of late November 2025: TikTok remains available in the United States. ByteDance is in the process of selling roughly 80% of TikTok’s U.S. assets to a consortium led by Oracle and Silver Lake, with ByteDance expected to retain under a 20% stake, but neither a completed divestiture nor a U.S. shutdown has occurred yet. (reuters.com)

Because TikTok has not been shut down in the U.S. under the April 2024 divest‑or‑ban law—and instead continues to operate while working toward a sale—the prediction that TikTok would fail to divest and “just get shut down” is wrong as of November 30, 2025.

As of November 30, 2025, the timeframe implied by 'within the next few years' after the April 24, 2024 TikTok law has not fully elapsed, so it is too early to say definitively whether Sacks’s forecast will play out.

Evidence so far:

• The Protecting Americans from Foreign Adversary Controlled Applications Act (the April 2024 TikTok law) explicitly targets TikTok/ByteDance; it has not been applied to Telegram, and no new federal law or executive action has been adopted that would ban or broadly restrict Telegram in the U.S. (en.wikipedia.org)
• Some U.S. political messaging now explicitly groups Telegram with TikTok as a platform exploited by terrorists. For example, an August 2024 Homeland Security subcommittee press release on a bipartisan bill warns about foreign‑controlled apps like TikTok and encrypted messaging platforms like Telegram being used by terrorist organizations, but it only mandates DHS threat assessments rather than proposing a ban. (homeland.house.gov)
• U.S. and international media have for years reported that Hamas and other extremist groups use Telegram; this continued after April 2024, with investigations describing Hamas‑aligned propaganda channels and possible radicalization of U.S. users. (nypost.com) However, this is not yet a broad, coordinated U.S. media campaign uniquely focused on Telegram.
• Claims that Telegram is effectively compromised or backdoored by Russian intelligence are being pushed mainly by Ukrainian officials (who banned Telegram on official devices over spying concerns) and by investigative outlets such as IStories (amplified in English by smaller sites), not as a central talking point of major U.S. media or top U.S. officials. (tbsnews.net)
• U.S. government action has so far targeted specific extremist networks that use Telegram, such as designating the far‑right Terrorgram collective as a terrorist group, rather than Telegram itself as a platform. (en.wikipedia.org)

Taken together, there are some elements of the dynamic Sacks described (messaging about terrorists’ use of Telegram and some discussion of Russian‑linked risks), but there is not yet clear evidence of the kind of sustained, coordinated U.S. media‑plus‑industry campaign to justify a nationwide Telegram ban that he portrayed as a 'foregone' conclusion. Because both the horizon ('next few years') and key causal steps he described are still unfolding, the prediction is best classified as inconclusive at this time.

1. Telegram has not been targeted under the new “foreign‑adversary app” authority.
   The Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA) was enacted in April 2024 and explicitly targets ByteDance/TikTok and its subsidiaries as “foreign adversary controlled applications,” while creating a process to designate other apps in the future. As of late 2025, public descriptions of the law and its enforcement still identify only TikTok/ByteDance as covered by the statute; there is no record of Telegram being designated or restricted under PAFACA or a similar federal foreign‑app authority. (en.wikipedia.org)

   There have been other actions involving Telegram—e.g., France’s criminal investigation of CEO Pavel Durov, Ukraine and some U.S. states (like Montana) restricting Telegram on government devices—but these are unrelated to PAFACA and do not amount to a U.S. federal ban or material nationwide restriction of Telegram under the new foreign‑controlled‑app framework sacks was discussing. (en.wikipedia.org)

2. The conditional part of the forecast (“if Biden wins a second term…they will set their gunsights on X”) cannot occur, because Biden did not win a second term.
   The 2024 U.S. presidential election was won by Donald Trump, who was inaugurated as the 47th president on January 20, 2025. (en.wikipedia.org) Since Biden is no longer president and does not hold a 2025–2029 term, the specific scenario sacks described—Biden using this authority in a second term to pressure Elon Musk to divest or curtail X—has been precluded by events.

3. No evidence of U.S. authorities using this authority to pressure Musk to divest X.
   Public reporting through November 30, 2025 shows no instance of the U.S. government invoking PAFACA or a similar “foreign adversary controlled applications” power against X (Twitter) or Elon Musk. Legal and regulatory actions involving Musk in this period—such as the SEC’s securities‑disclosure lawsuit about his 2022 Twitter acquisition—concern securities law and data‑access issues, not foreign‑controlled‑app national‑security powers or forced divestiture of X. (en.wikipedia.org)

Because: (a) Telegram has not been the first major target of the new foreign‑controlled‑app authority, and (b) the Biden‑second‑term condition for the later X‑related enforcement never materialized—and there is no sign of such enforcement against X under that authority—the prediction about the sequence of enforcement is best evaluated as wrong in light of actual developments to date.

• Joe Biden did not win a second term. The 2024 U.S. presidential election was won by Donald Trump, with Trump/Vance defeating the Harris/Walz ticket, so there is no “second Biden term” in which the described conduct by Biden’s attorney general could occur. (en.wikipedia.org)

• The Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA) was signed by Biden in April 2024 and gives the president and attorney general powers over “foreign‑adversary‑controlled applications,” explicitly targeting TikTok/ByteDance and allowing similar designations for other apps under specified conditions. (en.wikipedia.org)

• Under President Trump, use of these powers has focused on TikTok (including multiple executive orders delaying and shaping enforcement) rather than on X, and there is no public record of X being designated or investigated under PAFACA by federal authorities. (whitehouse.gov)

Because Biden never obtained a second term, the conditional scenario (“if Biden wins a second term, then his administration will open a PAFACA‑style investigation into X to pressure Musk to divest”) never arose. We therefore cannot know from real‑world events whether that hypothetical would have occurred. The prediction is about a counterfactual world rather than the world that actually unfolded, so its correctness cannot be determined from observed outcomes.

As of November 30, 2025, the forced sale of TikTok’s U.S. business has not actually been completed, only negotiated in principle.

After Congress passed the Protecting Americans from Foreign Adversary Controlled Applications Act in April 2024, ByteDance faced a divest-or-ban deadline, upheld by the Supreme Court in TikTok, Inc. v. Garland. TikTok temporarily went offline in January 2025, after which President Trump delayed enforcement and began negotiating a divestiture framework. 【0search13】【3search19】

By mid‑September 2025, reporting described only a “framework of a deal” for the U.S. operations of TikTok to be sold to a U.S.‑led investor consortium (Oracle, Silver Lake, Andreessen Horowitz, others), with Trump extending the ban/divestment deadline to December 16, 2025. These accounts consistently characterize the arrangement as pending or to be finalized, not closed. 【1news11】【1news13】【3search18】【0news16】【0news17】

Because the divestiture/ownership transfer has not yet closed, the condition in Friedberg’s prediction ("if it does go through" – i.e., a completed forced sale/auction of TikTok’s U.S. business) has not been met. Without a completed deal, there is naturally no post‑closing media cycle yet that would focus on (1) disclosed advisory fees to investment banks for running a completed auction, and (2) retrospective analyses on how the tech and media landscape has been reshaped by a finalized TikTok ownership change.

Some coverage, including a WSJ‑based exclusive cited by Voz.us, discusses a proposed multi‑billion‑dollar fee the U.S. government might charge buyers for brokering the TikTok deal, and only mentions typical investment‑bank fee percentages as contextual background; it is not a WSJ feature about banks’ auction fees on a consummated transaction. 【2view0】

Given that the triggering event (a completed forced sale of TikTok U.S.) has not yet occurred, the prediction cannot be definitively judged right or wrong at this time; it remains contingent on future developments.

Biden’s FY2025 budget proposed a 25% “minimum income tax on the wealthiest taxpayers,” calculated on total income including unrealized capital gains for households with over $100 million in wealth, along with higher rates on capital income that implied a top long‑term capital gains rate of roughly 44.6% for high earners (39.6% ordinary rate plus a 5% NIIT). (pwc.com)

However, these measures remained proposals only. Contemporary tax commentary and planning guides for tax year 2025 state that long‑term capital gains are still taxed at preferential rates of 0%, 15%, or 20%, with the effective top rate of 23.8% once the existing 3.8% NIIT is included—no 44.6% top federal long‑term capital gains rate was enacted. (ubs.com)

Likewise, there is no federal tax on unrealized gains or a 25% billionaire minimum tax in force as of late 2025; analyses by the Tax Foundation and the Committee for a Responsible Federal Budget still describe the 25% minimum tax on unrealized gains as a proposal and model its hypothetical revenue impact, indicating it has not become law. (taxfoundation.org)

Instead, the major tax legislation actually enacted in 2025 was the Republican "One Big Beautiful Bill Act," which extends and modifies the 2017 Trump tax cuts and makes various tax changes, but does not implement a wealth tax on unrealized gains or raise the top long‑term capital gains rate to 44.6%. (en.wikipedia.org)

Given that the FY2025 budget cycle (covering the fiscal year beginning October 1, 2024) has concluded without Congress enacting either the 25% unrealized‑gains wealth tax or the ~44.6% top long‑term capital gains rate from Biden’s FY2025 budget, Chamath’s prediction that “these things will never pass” within that 2025 budget cycle is borne out by the subsequent legislative record.

The prediction was explicitly conditional: Sacks said that if Democrats held the presidency and both chambers of Congress in 2024, and did so without Manchin and Sinema, one should “price in the strong possibility” that a wealth‑tax‑style measure on unrealized gains would pass.

In reality, the condition did not occur:

• In the 2024 elections, Republicans won the presidency and gained control of both the Senate and the House, producing a Republican rather than Democratic trifecta for the 119th Congress.(en.wikipedia.org)

In the actual 119th Congress under this GOP trifecta:

• Congress passed the One Big Beautiful Bill Act, which largely extends and expands Republican-style tax cuts; it does not create any federal tax on unrealized gains or a wealth tax.(en.wikipedia.org)
• As of October 2025, there is still no federal wealth tax in the United States; wealth-tax or unrealized-gains proposals (such as Biden’s “Billionaire Minimum Income Tax” and H.R. 6498) have been introduced but have not been enacted into law.(taxrep.us)

However, Sacks’s statement was not that such a tax would pass in the real world unconditionally; it was that if Democrats held a Biden-led trifecta without Manchin/Sinema, there would be a high likelihood ("strong possibility") of passage. Because that underlying political scenario never materialized, we have no way to observe whether his conditional probability assessment was accurate.

Therefore, the prediction cannot be scored as right or wrong based on actual outcomes and is best classified as ambiguous rather than correct, incorrect, or merely “too early.”

As of November 30, 2025 (about 19 months after the April 26, 2024 episode), there is no U.S. federal law that creates a new, special tax or levy directly on assets held inside 401(k)s, IRAs, or similar private retirement accounts. These accounts remain tax-advantaged under federal law: contributions are generally tax-deferred or deductible, withdrawals are taxed as income, and required minimum distributions continue under long‑standing rules, but there is no new asset-based charge on retirement balances themselves.(en.wikipedia.org)

Recent federal changes since the prediction have mostly expanded or adjusted incentives for retirement saving, not targeted the accounts as a new revenue source. The IRS has increased 401(k) contribution limits for 2025–2026, and retirement-related legislation (including SECURE 2.0’s implementation) emphasizes higher contribution and catch‑up limits and later RMD ages, rather than imposing new wealth-style taxes on retirement accounts.(apnews.com) Major 2025 tax legislation (the so‑called One Big Beautiful Bill Act) focuses on extending or modifying income-tax rates, SALT caps, remittance taxes, and various credits and deductions; public summaries do not show any provision that specifically imposes a new tax on 401(k) or IRA account balances.(en.wikipedia.org)

Although there has been active debate about wealth taxes at both federal and state levels, notable proposals either have not been enacted or explicitly exclude pensions and retirement accounts from their tax base.(nasdaq.com) Because Sacks’s prediction was explicitly about what would happen over the coming decades as federal liabilities grow, and only a short period has passed with no such policy adopted, there is not yet enough elapsed time to conclude that the prediction is either clearly right or clearly wrong. Therefore, the status is inconclusive (too early).

Evidence since April 2024 shows no U.S. federal move comparable to the TikTok divest‑or‑ban law that specifically targets Telegram for a nationwide restriction or ban.

Key points:

• The Protecting Americans from Foreign Adversary Controlled Applications Act, signed April 24, 2024, is written to explicitly and uniquely target TikTok and other ByteDance‑controlled apps, not Telegram. Its designation language and enforcement provisions are tailored to ByteDance and its subsidiaries, with no similar statutory designation of Telegram as a foreign‑adversary app. (en.wikipedia.org)

• As of late 2025, Telegram remains freely available in U.S. app stores and widely used. Overviews of Telegram’s legal status and global censorship list multiple countries that have blocked or restricted Telegram (Iran, Russia historically, China, Vietnam, Nepal, etc.), but do not list the United States as having imposed an operational ban or broad nationwide restriction on the service. (en.wikipedia.org)

• Post‑2024 U.S. legislative and policy activity that mentions Telegram has focused on threat assessments and research, not bans. For example, Homeland Security Subcommittee legislation requires DHS to conduct annual terrorism‑threat assessments of terrorist organizations’ use of “foreign cloud‑based mobile and desktop messaging applications like Telegram,” and related press releases explicitly group Telegram with other platforms (TikTok, etc.) in the context of studying online radicalization. These measures do not restrict user access to Telegram or order app‑store removal. (homeland.house.gov)

• Some state‑level and government‑device policies classify Telegram as a “high‑risk app” and ban it on state networks or devices (e.g., Montana’s 2023 ban on Telegram, WeChat, and Temu on government devices; North Carolina’s SB 83 proposal likewise listing TikTok, WeChat, and Telegram as high‑risk for public networks and devices). Those are limited, pre‑ and post‑2024 device/network rules, not a new, nationwide effort to ban or severely limit Telegram’s operation for the general public after the TikTok law. (itechpost.com)

• After the April 2024 TikTok law, the most aggressive actual restrictions in the U.S. have continued to fall on ByteDance products (TikTok and related apps) and, separately, on some newer Chinese apps like DeepSeek, RedNote/Xiaohongshu, and Lemon8, which have been targeted for bans on government devices in states like Texas. There is no comparable post‑law action singling out Telegram as “next on the hit list.” (en.wikipedia.org)

Together, this shows that while Telegram has drawn political and security scrutiny in U.S. discourse, the prediction that it would be the next major foreign‑owned communication app to face U.S. government efforts to significantly restrict or ban its operation—in a way analogous to TikTok’s divest‑or‑ban regime—has not materialized as of November 30, 2025.