E128: Google enters AI wars, Druck's warning, Trump crushes CNN & more

The prediction has three explicit time horizons:

1. Major U.S. financial crisis in the 2025–2035 period
   As of the current date (November 30, 2025), we are only ~1 year into the 2025–2035 window. While there have been ongoing concerns about U.S. debt levels and fiscal sustainability, there has not been a clearly recognized, discrete "major U.S. financial crisis" in this period comparable to 2008–2009. However, because the prediction is that such a crisis will occur at some point between 2025 and 2035, the window has barely begun, so this cannot yet be judged right or wrong.

2. Spending on seniors reaching ~60% of all U.S. tax revenues by ~2048
   This is a long-run forecast with a target around 2048. We are still more than 20 years away from that date, so it is too early to evaluate whether this will come true.

3. Social Security + Medicare + Medicaid rising from ~12% of GDP (around 2023) to ~24% of GDP by ~2048
   Current CBO and similar projections still only extend partway and are subject to revision, and 2048 has not arrived. This part of the prediction is also inherently long-term and not yet testable.

Because all three components refer to dates that are either in a wide future window (2025–2035) or specifically around 2048, and those dates have not yet passed, the overall prediction cannot be definitively assessed at this time.

Therefore, the correct classification is "inconclusive" (too early).

Available data through late 2025 support both parts of Chamath’s prediction, at least so far.

• Debt-to-GDP has continued to rise rather than fall. CBO figures cited by the Senate Budget Committee put federal debt held by the public at about 97.3% of GDP in 2023 and project it rising to roughly 99–100% of GDP by 2024–2025, not declining. (budget.senate.gov) Other summaries of 2025 data similarly report public debt near 100% of GDP and total (gross) federal debt around 123% of GDP, confirming an upward trajectory from 2023 levels. (crfb.org)
• No systemic break or cessation of the U.S. economy has occurred. CBO’s January 2025 outlook, while warning that debt is on an “unsustainable upward trajectory,” still projects continued real GDP growth around 1.8% annually over the next decade, with unemployment and inflation normalizing—i.e., a functioning, growing economy despite higher debt. (crfb.org) Late‑2025 reporting likewise describes policy debates, a recent government shutdown, and recession worries, but explicitly notes that overall U.S. economic growth remains intact and a broad recession is not expected. (reuters.com)
• Forward-looking projections align with his view. Multiple recent CBO-based analyses show debt held by the public climbing from about 100% of GDP in 2025 to well above 110%–150% of GDP in the coming decades, reinforcing the idea that debt-to-GDP is on a rising, not falling, path while the economy is still assumed to operate. (crfb.org)

Because (1) debt-to-GDP has indeed risen since 2023 rather than being reduced, and (2) this increase has not caused the U.S. economy to "stop" or suffer a systemic breakdown as of November 2025, the prediction is best classified as right (so far), with the obvious caveat that longer-term outcomes beyond 2025 remain unresolved.

As of 30 November 2025, neither condition in Chamath’s bet has occurred:

• U.S. federal debt-to-GDP in 2023–2025: Publicly available data from sources such as the U.S. Treasury and IMF show that U.S. gross federal debt-to-GDP has been roughly in the ~120–130% range (well above 50% but far below 200%) through 2023–2025. Various statistical series (e.g., IMF World Economic Outlook, Federal Reserve / FRED data) all place recent U.S. debt-to-GDP around this band, not near either 50% or 200%.
• Long-term projections: CBO long-term budget outlooks project federal debt held by the public (a narrower measure than gross debt) to rise in coming decades, potentially exceeding its WWII highs and continuing upward, but these are projections, not realized outcomes, and they do not show any near-term path to 50% of GDP. They also do not show debt reaching 200% of GDP as of 2025; those levels, if they occur, are many years in the future. (Multiple CBO reports up to 2025 reinforce this directional trend.)

Because the bet is explicitly about which event happens first in the future starting from the 2023 level — debt falling to 50% of GDP vs. rising to 200% of GDP — and neither threshold has been reached yet, there is no way to declare a winner.

Therefore, the prediction’s status as of 30 November 2025 is “inconclusive (too early)”: we do not yet know whether U.S. federal debt-to-GDP will hit 200% before it ever falls to 50%.

As of November 30, 2025, the conditions in Friedberg’s conditional prediction have not occurred, and the prediction is explicitly about what will eventually happen if those conditions are met.

Key facts:

• Debt-to-GDP has not reached ~200%. Gross federal debt is around 119–125% of GDP in 2024–2025, well below 200%.(pewresearch.org)(tradingeconomics.com)(fiscal.treasury.gov) Treasury’s long‑term projections suggest debt might exceed 200% of GDP around mid‑century (e.g., by 2049) under unchanged policy, but that is a projection, not a current reality.(fiscal.treasury.gov)(fiscal.treasury.gov)
• Interest costs are far below the implied ~$2T and <50% of revenue. Net interest outlays are just under $1 trillion per year as of FY 2025, not $2T, and amount to roughly 14–19% of federal revenues, not “more than half.”(gurufocus.com)(debt-clock-usa.com)(siai.org)
• Tax burdens are nowhere near the “70%+” level he describes. Federal receipts are about 17% of GDP in 2024, in line with historical norms.(bidenwhitehouse.archives.gov)(ycharts.com)(gurufocus.com) The top marginal federal income tax rate is 37%, and overall federal, state, and local tax collections are on the order of a few tens of percent of GDP, not 70%.(eitc.irs.gov)(pgpf.org)

Because:

1. The triggering scenario (debt ≈ 200% of GDP at ~4% rates, with interest consuming over half of government income) has not yet occurred; and
2. The prediction is about what will happen eventually under that future scenario (no specific deadline),

there is not yet a factual basis to say whether his claim about future 70%+ tax burdens is right or wrong. It remains a theoretical, forward‑looking argument about fiscal arithmetic and political choices. Therefore, the status of the prediction is inconclusive (too early to tell).

As of November 30, 2025, the U.S. Department of the Treasury has not issued 50‑year or 100‑year nominal Treasury bonds. The longest standard maturity remains the 30‑year Treasury bond.

Evidence:

• The Treasury’s current marketable securities lineup (bills, notes, bonds, and TIPS) continues to list bonds only out to 30 years as the longest maturity; there is no 50‑year or 100‑year instrument offered as a regular product.
• Discussions about ultra‑long U.S. Treasuries (50‑ or 100‑year) have occurred repeatedly in policy and market commentary (e.g., around 2016, again in 2019–2020, and after large deficits in the 2020s), but each time the Treasury Borrowing Advisory Committee and Treasury officials have ultimately decided against introducing such maturities, citing concerns about demand depth, market functioning, and overall financing strategy.

Because the prediction was that “we’ll have 50 year US bonds. We’ll have 100 year US bonds” and enough time has passed to check whether such instruments exist (they do not), the forecast is incorrect as of now.

As of now (through November 30, 2025), the U.S. Treasury has not issued any 100-year Treasury bonds.

Public reporting and Treasury documentation show that while ultralong maturities (including 50- and 100-year bonds) have been repeatedly studied or discussed, they’ve never been brought to market in the modern era. The Treasury Borrowing Advisory Committee and Treasury officials have confirmed on multiple occasions that ultralong bonds were evaluated but not issued, with supply focused instead on the existing curve out to 30 years.

Because the prediction is explicitly about the yield level “if and when” 100‑year bonds are issued, and that condition has not occurred yet, there is no market yield to observe and thus no way to test whether the yield would, in fact, come in below 1%.

So the prediction is neither confirmed nor falsified at this time; it depends on a future policy decision and subsequent market pricing that have not happened yet.

Evidence from official results shows that Donald Trump, the Republican nominee, won the 2024 U.S. presidential election held on November 5, 2024, defeating Democratic nominee Kamala Harris with 312 electoral votes to 226 and a popular‑vote plurality of 49.8% to 48.3%. (en.wikipedia.org) He was then inaugurated as the 47th president of the United States on January 20, 2025, marking the start of his second, nonconsecutive term. (en.wikipedia.org) Since the prediction was that Trump would be elected president in the next (2024) U.S. presidential election following the May 2023 town hall, and he in fact won that election and assumed office, the prediction is therefore right.

Friedberg’s May 2023 claim was that Donald Trump had a real chance of being reelected in the 2024 U.S. presidential election. The election was held on November 5, 2024, and official results show that Republican nominee Donald Trump defeated Democratic nominee Kamala Harris, winning 312 electoral votes to 226 and securing the popular vote as well, thereby becoming president again in a non-consecutive term.​ (en.wikipedia.org) Since Trump did, in fact, win the 2024 election, the subsequent outcome is fully consistent with the assertion that he had a serious shot at reelection. Therefore, this prediction is right.

Sacks’ statement was: “I do think it makes him more likely to be the nominee.” This is a claim about the change in probability of Donald Trump winning the 2024 Republican presidential nomination as a result of the CNN town hall, not a direct prediction that Trump will or won’t become the nominee.

Factually, Donald Trump did go on to win the 2024 Republican presidential nomination; he was formally nominated at the Republican National Convention in July 2024, after dominating the primary process and securing the required delegates earlier in the year.

However, whether the CNN town hall increased the likelihood of that outcome is a counterfactual causal claim: we cannot observe the alternative timeline in which the town hall did not occur and compare probabilities. Because of that, the truth of the specific normalized prediction — that the town hall made Trump more likely to win the nomination — cannot be definitively verified or falsified from observable data, even though we know the eventual outcome.

Therefore, the appropriate classification is "ambiguous": enough time has passed and the nomination outcome is known, but the causal effect of the town hall on Trump’s likelihood of winning cannot be determined from available evidence.

The key conditions of Sacks’s prediction did not materialize as described:

• Trump did become the 2024 Republican nominee and general-election candidate. He clinched the GOP nomination on March 12, 2024, was officially nominated in July, and went on to win the November 5, 2024 election. (en.wikipedia.org)

• Biden did not actually run the general-election campaign against Trump. Although he announced a reelection bid on April 25, 2023 and then dominated the Democratic primaries, Biden withdrew from the 2024 race on July 21, 2024 after a widely panned June 27 debate with Trump. Kamala Harris then launched her own campaign the same day; she became the Democratic nominee on August 5, 2024 and ultimately lost to Trump. (en.wikipedia.org) So the specific scenario Sacks described—Biden vs. Trump in the fall campaign—never fully occurred.

• While Biden was still a candidate in early 2024, his campaign activity was substantially more than “once a week” and not purely a low-key Rose Garden posture. An Axios analysis of his schedule found that through June 30, 2024, Biden held about 90 domestic events outside the D.C. area, roughly 3–4 per week, and had visited more swing states than Trump by that point. His travel was lighter than Obama’s 2012 reelection schedule but heavier than Trump’s pandemic-affected 2020 schedule, indicating a moderately active campaign, not a once‑a‑week response operation. (axios.com) Major campaign events included high-profile speeches near Valley Forge (Jan. 5), at Mother Emanuel AME in Charleston (Jan. 8), and a large March 28 fundraiser at Radio City Music Hall with Obama and Clinton. (en.wikipedia.org)

• Biden did adopt some elements criticized as a Rose Garden–style or low-visibility approach—mainly in press access, not campaign travel. He held relatively few formal press conferences and sit-down interviews compared to recent presidents, and commentators and allies at various points described or contemplated a “Rose Garden strategy” or “split-screen” approach that leveraged incumbency while Republicans fought among themselves. (washingtonpost.com) But this referred primarily to media and messaging strategy, not to the kind of once‑weekly campaign schedule Sacks predicted.

Because (1) Biden did not end up running the full 2024 campaign as the Democratic nominee against Trump, and (2) during the time he was a candidate his schedule involved many more than one brief appearance per week and a real—if somewhat lighter—campaign travel cadence, the prediction that he would essentially retreat to a low‑energy, once‑weekly Rose Garden campaign turned out to be wrong.

Sacks’ forecast was explicitly conditional on a Biden vs. Trump rematch, where sustained media saturation around Trump would create renewed “Trump fatigue” and lead enough voters to re‑elect Biden despite concerns about his cognitive decline.

In reality:

• Joe Biden withdrew from the 2024 race on July 21, 2024, citing the best interests of his party and the country, and endorsed Kamala Harris to be the Democratic nominee.(en.wikipedia.org)
• Kamala Harris became the Democratic nominee and faced Donald Trump in the general election.(en.wikipedia.org)
• On November 5, 2024, Donald Trump defeated Kamala Harris, winning 312 electoral votes to 226 and carrying the popular vote as well; Biden left office on January 20, 2025.(en.wikipedia.org)

Because the key condition of the prediction—a Biden vs. Trump general election—never occurred, we cannot directly observe whether voters would have re‑elected Biden out of Trump fatigue despite worries about his cognition. The actual outcome (Trump defeating Harris) doesn’t cleanly confirm or falsify the conditional claim itself.

Given that the contingency failed, the prediction’s truth value cannot be determined, so it is best classified as ambiguous rather than right or wrong.

Biden did in fact run for reelection in 2024 and did not exit the race before the election, so the event Friedberg was implicitly concerned about (Biden not running) did not occur. However, his explicit prediction was only that there was a non-zero chance this could happen. Since almost any contingent future political event has some non-zero probability, this is trivially true and not practically falsifiable. Because the statement doesn’t specify a concrete probability threshold or a calibration standard we can test against, we can’t rigorously classify it as right or wrong—only that the hypothesized outcome did not materialize. Therefore, the most appropriate label is "ambiguous".

Evaluating this prediction depends heavily on subjective judgments about what qualifies as an “even worse debacle” for Ukraine than in May 2023.

Key facts about the war in May 2023 (baseline):

• Russia’s full‑scale invasion had stalled after major Russian failures around Kyiv (2022) and significant Ukrainian counteroffensives in Kharkiv and Kherson (late 2022).
• By spring 2023, the front lines were relatively static, with Russia mounting costly offensives (e.g., around Bakhmut) for limited gains, while Ukraine was preparing a counteroffensive with growing Western aid and advanced weapons (HIMARS, tanks, etc.).

At that time, many analysts saw Ukraine as having serious challenges but still with realistic prospects for further gains, given Western support and Russia’s earlier battlefield failures.

Situation by the end of 2023:

• Ukraine’s 2023 counteroffensive (launched in June) achieved limited territorial gains and fell far short of optimistic public expectations, with very high Ukrainian casualties and significant equipment losses. Western and Ukrainian officials and independent analysts widely described the offensive as disappointing or a stalemate.
• Russia largely held its defensive lines in the south and east, and by late 2023 was transitioning back to offensive operations in some sectors, leveraging mobilization and a war‑time economy.
• Ukraine continued to suffer heavy damage to its energy and civilian infrastructure from Russian missile and drone attacks, while Western political debates about sustaining aid (especially in the U.S. and E.U.) intensified in late 2023.
• However, Ukraine did not experience a strategic collapse: no major new loss of a capital‑scale city beyond places already contested (like Bakhmut); Kyiv and much of the territory liberated in 2022 remained under Ukrainian control; and Ukraine scored some notable successes, especially in the Black Sea (e.g., pushing back the Russian Black Sea Fleet presence and enabling alternative grain export routes).

Why the prediction is ambiguous rather than clearly right or wrong:

• In favor of “right”: Relative to expectations in May 2023, by December 2023 Ukraine’s position looked worse than hoped: the counteroffensive underperformed, casualties were very high, Russia was adapting militarily and industrially, and Western support looked less assured. Many commentators described the situation as a stalemate trending unfavorably for Ukraine.
• In favor of “wrong”: A “debacle” usually implies an outright disaster or collapse. Ukraine did not lose the war, Kyiv did not fall, and Ukraine maintained political cohesion, continued to receive major Western support, and achieved some strategic successes. The line of contact in late 2023 was not dramatically worse for Ukraine than in May 2023; in some domains (naval/Black Sea, international support frameworks), Ukraine’s situation was arguably stronger.

Because “even worse debacle” is a highly normative and qualitative claim, and the end‑2023 situation can reasonably be interpreted either as:

• a significantly deteriorated strategic position, or
• a costly stalemate but not an outright disaster relative to May 2023,

there is no objective, widely accepted benchmark that clearly confirms or falsifies the prediction. Reasonable analysts can and do disagree about whether the change from May to December 2023 meets the threshold of an “even worse debacle.”

Therefore, the fairest classification is: ambiguous — the outcome cannot be determined as clearly right or clearly wrong based on available, mainstream factual assessments.

By the time of the May 12, 2023 episode, the U.S. was indeed in acute banking turmoil following the collapses of Silicon Valley Bank, Signature Bank, and First Republic—three of the largest bank failures in U.S. history.(en.wikipedia.org) But the normalized prediction you gave treats Sacks as forecasting that this turmoil would continue and develop into a serious, extended banking crisis over the ensuing months.

What actually happened is that the crisis was sharp but short-lived and was largely contained:

• In all of 2023 there were 5 U.S. bank failures; only three were large, with two small additional failures later in the year.(forbes.com) In 2024–2025 there were just a handful of small additional failures, far fewer than during 2009–2011.(archive.fdic.gov) This pattern does not resemble a prolonged, system‑wide fiasco.
• The Federal Reserve’s Bank Term Funding Program and other interventions are widely credited with averting a broader systemic banking crisis; the program expired in March 2024 after doing its job.(federalreserve.gov) The BIS similarly noted that large 2023 failures “did not lead to a systemic crisis.”(bis.org)
• By 2024–2025, the FDIC reported its insurance fund recovering faster than expected, and U.S. banks returned to solid profitability with rising deposits and strong capital and liquidity.(reuters.com)

So while there was a real but brief banking shock in spring 2023, it did not evolve into the extended, escalating fiasco implied by the prediction. Hence the prediction is best judged wrong.

The prediction was explicitly conditional: Sacks said that if the 2024 election were Biden vs. Trump, it would be a nail-biter/toss‑up with no clear favorite.

That condition never occurred. President Joe Biden withdrew from the 2024 race on July 21, 2024 and endorsed Vice President Kamala Harris, who became the Democratic nominee. The actual general‑election matchup was Donald Trump vs. Kamala Harris, not Biden vs. Trump. (nypost.com) Because the hypothesized Biden–Trump contest never happened, we can’t directly test how close it would have been “heading into the election.”

We do have polling data from when Biden was still running. In late June–July 2024, national polling averages and aggregators generally showed Trump leading Biden by about 2–4 percentage points nationally, e.g. RealClearPolitics, The Hill, and other averages had Trump up roughly 3 points at the time Biden exited. (en.wikipedia.org) Some modelers, such as Nate Silver, explicitly described Trump as a clear favorite rather than a toss‑up, putting Trump’s win probability around 65–66% vs. Biden. (nypost.com) That evidence suggests the race at that time was somewhat close in the numbers but not universally regarded as an even “toss‑up.”

After Biden withdrew, the race became Trump vs. Harris, and many analysts did characterize that contest as extremely close—Harry Enten, for example, called the 2024 election the closest presidential race in at least 60 years based on Harris–Trump polling. (vanityfair.com) But that pertains to a different matchup than the one Sacks specified.

Because the specific Biden–Trump general election Sacks was talking about never actually occurred, and we can’t observe how that hypothetical race would have looked right before Election Day, the truth of his conditional prediction cannot be definitively determined even though the 2024 cycle is over.

Hence the assessment: ambiguous.