E3: Modern Cold War between US & China, economic recovery, potential mass migration out of San Francisco, pandemic politicization & more with David Sacks & David Friedberg

Chamath’s timeline and mechanism were partly right (vaccines and therapeutics did arrive by late 2020–2021), but his overall prediction that COVID-19 would be “well in hand” within two years—such that it was no longer a major uncontrolled public health crisis—did not hold by May 2022.

By the two‑year mark:

• Effective tools did exist. Multiple vaccines were widely deployed globally and in the U.S., and the first oral antiviral, Paxlovid, received FDA emergency use authorization in December 2021, providing a convenient treatment that reduces risk of hospitalization and death. (fda.gov)
• However, the World Health Organization repeatedly reaffirmed that COVID-19 still constituted a Public Health Emergency of International Concern (PHEIC). In April 2022 and again in July 2022, WHO’s Emergency Committee unanimously agreed the pandemic remained an “extraordinary event” that continued to adversely affect global health and required a coordinated international response. (who.int)
• WHO did not declare an end to the global COVID-19 PHEIC until May 5, 2023, almost a year after Chamath’s two‑year window. (paho.org)
• In the United States, the CDC reported a 7‑day average of about 273 deaths per day as of May 11, 2022, with nearly 1,000,000 cumulative deaths, indicating that COVID-19 was still causing substantial mortality. (archive.cdc.gov) The U.S. federal COVID public health emergency was only allowed to expire on May 11, 2023. (sciencenews.org)

Because by May 2022 COVID-19 was still officially treated as a global and U.S. public health emergency with hundreds of deaths per day in the U.S., it had not become “well in hand” in the sense of no longer being a major uncontrolled public health crisis. Therefore, the prediction is best scored as wrong overall.

Evidence suggests that the U.S. largely “got out of” the COVID‑19 recession within roughly the 2–3 year window David Sacks predicted.

GDP and overall activity

• NBER dates the recession trough to April 2020; the economy was in expansion thereafter.(nber.org)
• Real GDP surpassed its pre‑recession peak by Q1 2021—less than a year after the trough—and later exceeded CBO’s pre‑pandemic projections, indicating a very rapid recovery.(cbpp.org)
• Treasury analysis shows U.S. real GDP more than 5% above its 2019 level by 2023, though still with some shortfall vs. pre‑COVID trend, i.e., essentially recovered but still normalizing.(home.treasury.gov)

Jobs and unemployment

• The U.S. regained all ~22 million nonfarm jobs lost in early 2020 by June 2022—about 26 months after the April 2020 employment trough, within the 2–3 year window he gave.(floordaily.net)
• BLS reports the unemployment rate returned to its pre‑pandemic level (around 3.5–3.6%) in 2022; several labor‑market measures were back at or near pre‑COVID norms by then, even though labor‑force participation remained somewhat lower.(bls.gov)

“Unless some other shoe drops” condition

• While there were negative developments (Delta/Omicron waves, supply‑chain disruptions, and high inflation), these did not produce a second, prolonged COVID‑era recession; output and employment continued to grow and eventually exceeded pre‑pandemic levels.(cbpp.org)

Putting this together: the core economic damage from the initial COVID shock was substantially repaired in about 2 years (by mid‑2022) and clearly by around 3 years out (2023) on most headline metrics. That falls squarely within Sacks’s predicted “2 to 3 year process to get out of this” in the absence of an additional major downturn, so the prediction is best judged as broadly right, if somewhat conservative given how fast GDP itself rebounded.

There are two reasonable ways to interpret what “a 2–3 year process to get back to some sort of [normal]” means, and they give opposite verdicts:

1. If “normal/lower level” just means getting back to single‑digit unemployment (under ~10%)

   • The U.S. unemployment rate (U‑3) peaked at 14.7% in April 2020 and was 13.3% in May 2020, around when the prediction was made.
   • It fell below 10% by August 2020 (8.4%), only about three to four months after the spike, not 2–3 years later.
   • By December 2020, it was already down to 6.7%.

   These data imply that the move from ~15% to clearly lower, single‑digit unemployment happened in well under a year, so on that interpretation his 2–3 year timeline is wrong. (bls.gov)

2. If “normal/lower level” means returning roughly to pre‑pandemic labor‑market conditions (~3.5–4% unemployment)

   • Before COVID, the unemployment rate was about 3.5% (February 2020). (bls.gov)
   • The rate kept improving through 2021 and reached 3.9% by December 2021, and around 3.5% again in mid‑2022 (e.g., July 2022 was 3.5%), essentially back to its pre‑COVID level.
   • From May 2020 (~13.3%) to December 2021 (3.9%) is ~1.5 years; to mid‑2022 (3.5%) is about 2.1–2.2 years—squarely inside his "2 to 3 year" window for getting back to something like the old normal.

   On this stricter endpoint definition, his 2–3 year horizon is roughly correct. (fortune.com)

Because the podcast quote stops at “some sort of, you know …” and the user’s normalization adds its own gloss (“e.g., around single‑digit unemployment”), we can’t be sure which level Sacks meant: any single‑digit rate (which happened fast) or a near‑full normalization to ~3.5–4% (which did take about 2 years). Both readings are plausible and lead to opposite grades, so the fairest overall judgment is ambiguous.

Summary
Friedberg predicted that outside very high‑density cities like New York, places such as Dallas, Houston and San Diego would not see a New York‑style second wave with extreme daily death rates, but instead a prolonged low‑level “slow burn” of cases. In reality, all three metros experienced clearly defined, severe waves (summer 2020 and especially winter 2020–21) with sharp spikes in deaths and hospital strain—i.e., surges rather than a mere slow burn—although their per‑capita peaks did remain below New York City’s first‑wave extremes. Overall, the core dynamic he forecast (no big waves, just a slow burn) did not hold.

What actually happened

1. New York City’s benchmark

   • NYC’s first wave was uniquely intense: from Feb 29–Jun 1, 2020 there were 18,679 deaths, with deaths peaking around the week of April 5 at a mean ~566 deaths per day; single‑day peaks reached roughly 600–800 deaths.(cdc.gov)
   • With a population of about 8.3 million, that corresponds to ~6–8 deaths per 100,000 residents per day at peak—an outlier even within the U.S.

2. Dallas (Dallas County)

   • After the podcast (May 2020), Dallas had a summer 2020 wave: by mid‑July 2020 the county recorded a deadliest week with 74 deaths in one week, sharply above prior levels.(dallasnews.com)
   • A much larger winter 2020–21 wave followed: late January 2021 saw the county’s deadliest week with 183 deaths, and on Jan 27 and Feb 3, 2021 the county reported record single‑day tolls of 40 and then 50 deaths, respectively.(keranews.org)
   • With ~2.6M people in Dallas County, 50 deaths in a day is ~1.9 deaths/100k/day—well below NYC’s first‑wave per‑capita peak but still a pronounced surge, not a low‑level burn.

3. Houston (Harris County / Texas Medical Center)

   • Harris County saw a major summer 2020 surge: by late July 2020, 109 deaths were confirmed in a single week and roughly one‑third of all county deaths to date were reported after July 10, indicating a rapid, wave‑like escalation.(communityimpact.com)
   • Earlier in that surge, the seven‑day rolling average of daily deaths more than doubled within days, again reflecting a sharp spike rather than a steady plateau.(communityimpact.com)
   • At the regional level, Houston’s Texas Medical Center reported ICU occupancy at 97% of capacity by June 24, 2020 due to COVID‑19, a hallmark of an acute wave stressing hospital systems.(en.wikipedia.org)
   • Texas as a whole went on to experience multiple high‑intensity waves, including the winter 2020–21 and Delta waves, with ICU demand exceeding 100% of capacity in many hospitals in August 2021.(en.wikipedia.org)

4. San Diego County

   • San Diego had relatively modest early numbers, but then a large winter 2020–21 wave. On Jan 22, 2021, county officials reported a record 79 COVID‑19 deaths in a single day, breaking prior records that same month. The report notes 53 consecutive days with >1,000 cases and 25 days in 30 with >2,000 cases—classic wave dynamics.(kpbs.org)
   • With ~3.3M residents, 79 deaths in a day is ~2.4 deaths/100k/day—less than NYC’s first‑wave peak, but still a substantial, concentrated spike.

5. Comparison with the prediction

   • Correct component: None of these metros reached per‑capita daily death rates as extreme as New York City’s early‑April 2020 peak; NYC remained an outlier in intensity.(cdc.gov)
   • Incorrect core claim: Friedberg specifically said that in cities like Dallas, Houston and San Diego, “you’re not going to see that… you’re going to have more of the slow, steady burn… I wouldn’t expect us to have a New York style second wave… just this slow burn going forward of new cases.” The actual data show pronounced, discrete waves in all three metros (summer 2020, winter 2020–21 and later), featuring rapid rises in cases and deaths, weekly and daily death records, and serious hospital strain—not merely a flat, low‑level “burn.”(dallasnews.com)

Because the central behavioral forecast—that these cities would avoid large second (and subsequent) waves and instead see only a gradual, steady burn—was contradicted by substantial, wave‑like surges in deaths and hospitalizations, this prediction is best classified as wrong overall, despite being directionally right that few places would match New York City’s uniquely extreme first‑wave death rates.

Donald Trump lost the November 3, 2020 U.S. presidential election to Joe Biden, who won both the popular vote and the Electoral College. Congress certified Biden’s victory on January 6–7, 2021, and Biden was inaugurated on January 20, 2021.

Chamath framed Trump’s reelection as “overwhelmingly likely” and specifically tied that likelihood to voter frustration over COVID-19 lockdowns, implying this frustration would drive Trump to victory unless Democrats ended lockdowns early. In reality:

• Joe Biden won 306 Electoral College votes to Trump’s 232.
• Biden won the popular vote by over 7 million votes.

Given that the prediction was explicitly about the 2020 election outcome and that outcome is now known, the forecast that Trump was “overwhelmingly likely to win” is incorrect.

Therefore, the prediction must be rated as wrong.

Evidence from polling, exit polls, and post‑election analyses indicates that COVID‑19 and related restrictions were indeed a salient issue by November 2020, but they hurt Trump rather than becoming a key advantage for him.

1. The antecedent mostly held: COVID and pandemic policy were top election issues.

• Surveys in mid‑ to late‑2020 found the coronavirus outbreak was among the top issues for voters, alongside the economy and racial inequality.
  • Pew (Aug & Oct 2020) shows majorities saying the coronavirus was a very important factor in their vote.(pewresearch.org)
  • KFF’s September 2020 tracking poll found the coronavirus outbreak named as the single most important issue by about one‑fifth of voters (second only to the economy).(kff.org)
• NBC/ CNBC exit polling on Election Day found the economy, racial inequality, and the Covid‑19 crisis as the top three issues; 52% of voters said controlling the pandemic even at the expense of the economy was more important than reopening.(cnbc.com)

So pandemic conditions and restrictions remained highly salient into November.

2. Voters largely disapproved of Trump’s COVID handling and trusted Biden more.

• By July–October 2020, national and swing‑state polls showed solid majorities disapproving of Trump’s handling of the pandemic (e.g., 61% disapprove vs. 35% approve in KFF July poll; ~40% approve vs. ~58–61% disapprove in late‑Oct Marquette and other polls).(kff.org)
• Pew in April 2020 already found about two‑thirds of Americans saying Trump had been too slow in his initial response.(pewresearch.org)
• Analyses noted that as COVID worsened, more voters preferred Biden over Trump as crisis manager; Brookings summarized that the more the election became a referendum on Trump’s pandemic management, the worse his prospects looked.(brookings.edu)

3. COVID as a top issue benefited Biden, not Trump.

• An internal post‑election autopsy by Trump’s own pollster, Tony Fabrizio, concluded Trump “lost the 2020 election largely because of his handling of the coronavirus pandemic”; voters in 10 key states rated the pandemic as their top issue and gave Biden significantly higher marks.(washingtonpost.com)
• Reporting on that autopsy notes that voters who saw coronavirus as a top priority broke for Biden by nearly 3‑to‑1.(independent.co.uk)
• NBC exit polls found that among voters prioritizing the pandemic and favoring stronger virus control over faster reopening, support skewed heavily toward Biden.(cnbc.com)

4. Scholarship finds the pandemic reduced Trump’s vote share.

• A peer‑reviewed study on COVID‑19 and the 2020 election found that higher local COVID case incidence decreased support for Trump, especially in urban counties, swing states, and states he had won in 2016; they estimate that with even modestly fewer cases, Trump likely would have been reelected.(pmc.ncbi.nlm.nih.gov)
  This directly contradicts the idea that ongoing pandemic conditions and restrictions became an electoral asset for him.

5. Lockdown anger did not override perceptions of incompetence in 2020.

• While there was visible backlash to lockdowns, polling in key swing states showed many voters instead blamed Trump for reopening too quickly and failing to model protective behaviors; a CNBC/Change poll found 55% agreed that Trump was pushing states to reopen too fast to help his reelection, and many cited him as the top person responsible for surging cases.(cnbc.com)
• Axios’s retrospective review of COVID’s political impact notes that COVID was a central issue in 2020 and that voters prioritizing virus control over the economy helped defeat Trump—though lockdown and school‑closure backlash later aided Republicans in 2024, not 2020.(axios.com)

Conclusion
By November 2020, COVID and pandemic restrictions were still major issues, but instead of “superseding” concerns about Trump’s earlier mishandling and turning into a key pro‑Trump issue, they remained tightly linked to perceptions of his incompetence and were a central reason he lost the election. The core prediction—that if COVID/lockdowns were still salient, the issue would primarily benefit Trump—is therefore wrong.

Evidence since 2020 shows that California did, in fact, experience large‑scale net out‑migration to other U.S. states, commonly described as an exodus and widely linked to high costs and taxes.

Scale of out‑migration (“in droves”)

• Census‑based summaries indicate that from roughly spring 2020 to July 2024, California lost nearly 1.5 million residents on net to other states. (californiainsider.com)
• State and Census data show very large annual domestic net losses:
  • About 692,000 people left California for other states in 2021 while fewer than 337,000 moved in, a net domestic loss of ~355,000. (governing.com)
  • For 2022, estimates show a net domestic out‑migration of around 343,000, the largest numeric loss of any state. (dailywire.com)
  • Losses have moderated but remain substantial, with net domestic outflows around 200,000–250,000 per year as of 2023–24. (dof.ca.gov)

Cumulatively, this constitutes a very large, historically unusual out‑migration over the “coming years” after May 2020, consistent with the colloquial claim that people would “leave in droves.”

Role of high costs and taxes

• Analyses of California’s recent population decline by the Public Policy Institute of California identify high housing costs as a major driver of people leaving for other states, with about one‑third of Californians saying they are considering leaving the state because of housing costs. (ppic.org)
• Reporting on the 2021–2023 population losses notes that critics of state policy explicitly tie the out‑migration to higher taxes, high cost of living, homelessness, and housing shortages under Democratic governance. (khsu.org)
• A Census‑based summary of 2021–22 migration notes that hundreds of thousands left California and cites high cost of living, housing shortage, rise of remote work, and high taxes as key motivations. (en.wikipedia.org)
• A 2025 analysis of IRS migration data by the National Taxpayers Union Foundation finds California losing taxpayers on net to lower‑tax states such as Florida and Texas and directly points to California’s high tax burden and cost of living as central factors in this wealth and taxpayer outflow. (nypost.com)

Taken together, these data show that in the years after May 2020 California experienced very large net domestic out‑migration, and widely cited explanations emphasize high housing costs, overall cost of living, and high taxes—matching the substance of Chamath Palihapitiya’s prediction.

Assessment
Chamath’s prediction has held up as of November 29, 2025.

1. Existing main manufacturing operations did not relocate out of California

• Tesla’s Fremont Factory in California remains a major production hub, described by Tesla itself as its hub for Model S, Model 3, Model X and Model Y production. It is still one of the largest manufacturing sites in California and is actively hiring. (tesla.com)
• Production there has not been wound down; instead, Tesla has expanded activity. In 2024 the Fremont plant celebrated producing its 1,000,000th Model Y and reported installed annual capacity of ~550,000 Model 3/Y plus 100,000 S/X. (teslamagz.com)
• In 2025, Tesla was expanding 4680 cell manufacturing at Fremont, indicating ongoing investment rather than relocation away from California. (teslaoracle.com)
• While Tesla moved its corporate headquarters from Palo Alto, California, to Austin, Texas in 2021, Elon Musk explicitly stated this did not mean Tesla was leaving California and that the company intended to keep increasing output from Fremont. (cnbc.com)

Taken together, this shows Tesla did not relocate its existing core manufacturing operations out of California; Fremont remains a key factory.

2. Incremental facilities were built in other locations, driven largely by tax incentives

• After 2020, Tesla built and ramped Gigafactory Texas near Austin, a major new vehicle plant producing the Model Y and Cybertruck. (mysanantonio.com)
• To land this factory, Texas entities offered substantial tax incentive packages, matching Chamath’s rationale:
  • The Del Valle Independent School District approved a Chapter 313 property-tax limitation that effectively grants Tesla tens of millions of dollars in school-tax savings over about a decade. (dallasnews.com)
  • Travis County approved an economic development agreement rebating 70–80% of certain property taxes for up to 20 years, estimated at roughly $14–15 million in savings over the first decade. (kut.org)
• In 2025 Tesla also moved to build a new Megapack “megafactory” near Houston, Texas, under a tax abatement deal with Waller County that reduces its property taxes in exchange for large capital investment—another incentive-driven siting choice. (reuters.com)
• More broadly, Tesla has continued adding major facilities outside California—Gigafactory Berlin, further Gigafactory Texas build‑out, planned Gigafactory Mexico—while simultaneously expanding some operations inside California (e.g., the Lathrop Megafactory and a large engineering HQ in Palo Alto). (en.wikipedia.org)

Conclusion
Chamath claimed (1) it was not practical to move Tesla’s existing California manufacturing out of the state and (2) incremental facilities would instead be built elsewhere, guided heavily by tax incentives. Fremont remains a major and even expanding manufacturing site, while Tesla’s major new plants and megafactories have indeed been located in other states (and countries) using large incentive packages. That aligns closely with his prediction, so the forecast is right.

Why this prediction is classified as inconclusive

• The prediction window is “in 20 years” from 2020—i.e., by around 2040. As of today (late 2025), only about 5 years have elapsed since the prediction, leaving roughly 15 years in the forecast horizon. The claim is about what could happen by 2040, so it cannot yet be definitively judged right or wrong.

• Empirically, humanity is far from eradicating all infectious diseases as of 2025. Smallpox remains the only human infectious disease that has been eradicated globally. WHO and related summaries describe smallpox as the only human disease ever eradicated, certified in 1980 after the last natural case in 1977. (who.int) Many other major infectious diseases (e.g., malaria, TB, HIV, a wide spectrum of respiratory and enteric infections) persist at large scale.

• Even for polio, a single human infectious disease with a long-running, well-funded eradication campaign, eradication has not yet been completed. As of 2022–2023, wild poliovirus type 1 remains endemic in Afghanistan and Pakistan, and outbreaks of vaccine-derived polioviruses continue in multiple countries. (asm.org) This illustrates how challenging eradication is, even for one pathogen with effective vaccines and decades of effort.

• The trajectory of antimicrobial resistance (AMR) is moving in the opposite direction of global eradication: a 2025 WHO report shows that in 2023, about one in six laboratory-confirmed bacterial infections worldwide was resistant to antibiotics, with resistance rising in over 40% of pathogen–antibiotic combinations between 2018 and 2023. (who.int) AMR is projected to cause millions of deaths and major economic damage by 2050 without strong intervention, which makes the idea of eradicating all infectious disease even more technically and logistically daunting. (ft.com)

• Scientifically, most global health and infectious-disease experts consider the complete eradication of all infectious diseases (including those with animal or environmental reservoirs, rapidly mutating viruses, and increasingly drug-resistant bacteria) extremely unlikely in any 20‑year window. However, the prediction is phrased in a counterfactual/possibilistic way (“we could… the only thing holding that up is regulation”), which makes it harder to falsify outright: it depends on assumptions about future political, regulatory, and funding choices as much as on biology.

Given that:

1. The deadline (circa 2040) has not yet arrived, so the prediction’s outcome is not yet observable, and
2. The statement is partly about what could be possible under different regulatory conditions rather than a hard, time‑stamped forecast of what will happen,

the fairest classification at this point is “inconclusive (too early)”.

Substantively, based on current evidence on AMR trends, remaining eradication challenges even for a single disease like polio, and the enormous ongoing burden of infectious diseases worldwide, the prediction looks highly implausible—but it cannot yet be declared definitively wrong under the 2040 timeline.

Assessment Chamath’s 21 May 2020 claim that “this is the beginning of the modern Cold War… America versus China” predicted a prolonged, Cold War–style strategic rivalry marked by systemic geopolitical and economic competition. Developments since 2020 align strongly with that description.

Key evidence that a Cold War–style rivalry emerged and persisted

1. U.S. policy framing China as a long‑term strategic competitor

   • In February 2021, President Biden announced a Pentagon China Task Force specifically to craft strategy for the United States to “win the competition of the future” with China, emphasizing military posture, alliances, and technology as part of a broader rivalry. (en.wikipedia.org)
   • Major think tanks like Brookings now explicitly describe U.S.–China ties as great‑power competition and debate whether it constitutes a “new cold war,” noting that intense, prolonged competition is now a “defining feature” of the era. (brookings.edu)

2. Deepening and durable economic/technological confrontation

   • In October 2022, the U.S. imposed sweeping semiconductor and advanced computing export controls on China, described as the most expansive export-control action in decades and intended to block China’s progress in critical technologies for economic and military power. (en.wikipedia.org)
   • These restrictions have been repeatedly tightened (e.g., further rules in 2023–2024 on chipmaking tools), explicitly framed as part of a long‑term effort to preserve U.S. technological advantage and constrain China’s military modernization. (theguardian.com)
   • High‑profile corporate clashes (e.g., Nvidia’s AI chips becoming a central battleground in U.S.–China export and investment controls) illustrate that the rivalry is being fought “with computers and money,” just as Chamath described. (apnews.com)

3. Growing geopolitical and security bloc dynamics

   • Analysts and officials increasingly adopt “new Cold War” or similar framing, emphasizing systemic ideological, economic, and security competition between the U.S. and China, including contests over supply‑chain “chokepoints” (rare earths vs. advanced semiconductors). (blog.maxthon.com)
   • Parallel to this, China has tightened strategic coordination with other U.S. adversaries (Russia, Iran, North Korea), sometimes described collectively as a new anti‑Western axis or the “CRINK” grouping, further reinforcing bloc‑like geopolitics reminiscent of a Cold War environment. (en.wikipedia.org)

Why this supports the prediction

• The rivalry is long‑term and structural, not a transient trade spat: it spans military planning, industrial policy, technology, diplomacy, and alliances.
• Both sides increasingly treat the other as the primary strategic competitor and redesign institutions and laws accordingly (export controls, investment screening, task forces, alliance building).
• Expert discourse widely recognizes a “new Cold War” / “Cold War‑like” dynamic—even if some scholars caution against the metaphor, they still agree the relationship is defined by intense, enduring competition rather than partnership. (brookings.edu)

Taken together, events since 2020 closely match Chamath’s forecast of a “modern Cold War” marked by systemic, U.S.–China strategic rivalry across economic and geopolitical domains.

Conclusion Given five years of consistent evidence of entrenched, Cold War–style competition between the U.S. and China, Chamath’s prediction is substantively correct.

Summary
Sacks predicted that in the 2020 election both parties would compete to be tough on China, that Trump would blame China for COVID-19 and brand Biden “Beijing Biden,” that Biden would not defend China but instead adopt a tougher China line, and that U.S.–China economic decoupling would emerge as a bipartisan policy direction. On all major elements, subsequent events match his forecast.

1. 2020 campaign: China became a central, bipartisan attack line

• Trump blaming China for COVID-19. Throughout 2020, Trump repeatedly blamed China for the pandemic, calling it the “Chinese virus” and insisting the outbreak was China’s fault. (cnbc.com)
• “Beijing Biden” framing. Republican messaging explicitly used the label “Beijing Biden” and tied Biden to China. FactCheck.org documented an RNC robocall that repeatedly called him “Beijing Biden.” (factcheck.org)
  The 2020 Republican National Convention coverage notes Donald Trump Jr. deriding Biden as “Beijing Biden,” and Trump allies launched a website, BeijingBiden.com, focused on Biden’s alleged coziness with China. (en.wikipedia.org)
• Biden did not defend China; he tried to be tougher. NPR/GBH reported that Trump and Biden were “battl[ing] over who is ‘weak on China,’” each accusing the other of being soft, with Biden ads saying Trump “rolled over for the Chinese.” (wgbh.org)
  During the campaign Biden called Xi Jinping a “thug” and emphasized that the U.S. “does need to get tough with China,” including over Xinjiang camps and Hong Kong. (bloomberg.com) This is the opposite of “defending” China and fits Sacks’s expectation that Biden would move to Trump’s right rhetorically on toughness.

Verdict on the campaign piece: Correct. China policy became a bipartisan campaign issue, Trump ran heavily on blaming China, Biden was attacked as “Beijing Biden,” and Biden adopted explicitly tough rhetoric rather than defending Beijing.

2. Post‑2020: bipartisan policy direction toward decoupling

Sacks’s deeper claim was not just about rhetoric but that actual U.S. policy would move toward economic decoupling, with support from both parties. Evidence since 2020 supports this in key sectors:

• Trump-era tariffs preserved and expanded under Biden. Analyses from CNN, NPR and others note that Biden kept Trump’s Section 301 tariffs on roughly $300–370 billion of Chinese imports and, in 2024, added higher tariffs on strategic products such as EVs, batteries, solar cells, steel, aluminum and semiconductors. (amp.cnn.com) Republican leaders even praised Biden for keeping Trump’s China tariffs, underscoring the bipartisan consensus. (waysandmeans.house.gov)
• CHIPS and Science Act and tech supply-chain reorientation. The 2022 CHIPS and Science Act—explicitly aimed at reducing reliance on Chinese semiconductors and boosting U.S. production—passed the Senate 64–33 and the House 243–187, a clearly bipartisan vote. (aflcio.org) Guardrails in the law restrict recipients from expanding advanced-chip capacity in China, pushing production back to the U.S. or allied countries. (democrats-science.house.gov)
• Export controls and sanctions on Chinese high tech and human‑rights abusers. In October 2022 the Commerce Department imposed sweeping export controls limiting China’s access to advanced chips and manufacturing equipment; these controls have since been expanded, explicitly framed as countering China’s tech rise. (en.wikipedia.org) Separately, the Uyghur Forced Labor Prevention Act—passed with overwhelming bipartisan support—presumes goods from Xinjiang are made with forced labor and has led to bans on imports from scores of Chinese companies, forcing firms to shift supply chains. (en.wikipedia.org)
• Institutionalizing competition with China in Congress. The House established a bipartisan Select Committee on the Strategic Competition between the U.S. and the Chinese Communist Party in January 2023, with members from both parties, to coordinate economic, technological, and security policy toward China; Reuters notes that this China-focused committee is being continued across Congresses, signaling durable cross‑party commitment. (en.wikipedia.org)
• Measured but real decoupling of trade and critical sectors. Empirical work and policy analysis find that U.S.–China economic ties are being restructured rather than fully severed. A St. Louis Fed study shows China’s share of U.S. imports in communications equipment fell from 62% in 2016 to 44% in 2023, and in information technology goods from 46% to 27%, consistent with deliberate diversification away from China in high‑risk sectors. (stlouisfed.org) A Journal of International Economics paper similarly finds U.S. trade policy has produced “real” U.S.–China decoupling in bilateral trade, with China’s share of total U.S. imports dropping from 22% to 16% between 2017 and 2022 as imports shift to other countries. (sciencedirect.com) Consulting and policy reports from Bain and others characterize U.S.–China tech and supply-chain decoupling as an ongoing, defining trend. (bain.com)

Biden and G‑7 leaders do explicitly say they seek “de‑risking, not decoupling,” to signal they don’t want a total rupture. (cnbc.com) However, the concrete policies—tariffs, onshoring subsidies, export controls, forced‑labor import bans, and congressional actions—collectively amount to partial, sectoral decoupling from China, especially in technology and sensitive supply chains, and they have broad bipartisan backing.

Verdict on the policy piece: While the U.S. has not fully severed economic ties with China, a sustained, bipartisan policy push has clearly moved the relationship toward selective economic decoupling and de‑risking in strategic sectors, which is what Sacks was pointing to.

Overall assessment

Taken together:

1. The 2020 election did in fact feature a bipartisan competition over who was tougher on China, with Trump blaming China for COVID-19 and branding Biden “Beijing Biden,” and Biden responding by hardening his own stance rather than defending Beijing. (theguardian.com)
2. Since then, U.S. policy has moved in a bipartisan way toward partial economic decoupling from China, especially in technology and critical supply chains, via tariffs, industrial policy, export controls, and rights‑based import bans. (aflcio.org)

Given the prediction’s level of generality (“decoupling is going to happen” and will be “bipartisan”), and the clear alignment of both campaign dynamics and subsequent policy with that description, the forecast is best judged as right in substance, even though what has emerged is targeted, sectoral decoupling rather than total economic separation.

Assessment Sacks predicted that in the years after May 2020 the U.S. would bring home production of strategic medical goods—antibiotics, medicines, PPE—so that these would be manufactured domestically instead of relying on China.

1. Medicines and antibiotics are still largely import‑dependent

Multiple recent analyses show that the U.S. continues to rely heavily on foreign—especially Chinese and Indian—suppliers for both finished drugs and active pharmaceutical ingredients (APIs):

• A 2024–25 risk assessment finds the U.S. now imports about 75% of its essential medicines; China and India dominate API production, and China supplies about 90% of the antibiotic APIs used in the U.S.(exiger.com)
• A 2025 industry study reports that China and India together provide roughly 70–80% of U.S. generic drug supply (counting both finished drugs and APIs), with China controlling most global production of key antibiotic ingredients.(prosperousamerica.org)
• Policy and think‑tank work in 2024–25 concludes that more than 60% of APIs used in the U.S. come from India and China, and only around 10% of APIs are made domestically.(5g.wilsoncenter.org)
• Academic and regulatory reviews note that over 92% of facilities making generic-drug APIs for the U.S. market are located overseas, primarily in India, China, and parts of Europe.(pmc.ncbi.nlm.nih.gov)

Taken together, these data indicate that dependence on foreign—often China-linked—supply chains for antibiotics and generic medicines has persisted or even deepened since 2020, rather than being broadly repatriated.

2. PPE production: domestic boosts, but China still dominant

PPE shows a similar pattern: a temporary domestic surge followed by renewed reliance on imports, especially from China.

• Industry groups note that before Covid more than 90% of PPE used in the U.S. was made in Southeast Asia, mainly China; they report that after an initial pandemic-era buildup of U.S. factories, many of those plants have since downsized or shut, while China’s market share has increased again.(ammaunited.org)
• By 2025, U.S. PPE manufacturers’ association data describe a collapse of the pandemic startup wave: out of 100+ new PPE makers, only a handful remain in operation, while China now produces the vast majority of medical protective gear used by U.S. healthcare workers.(ammaunited.org)
• Government oversight and trade reports find that the U.S. remains highly dependent on foreign suppliers—particularly for items like nitrile gloves and gowns—and that higher tariffs on Chinese medical imports are likely to shift sourcing to other low-cost countries more than to large-scale U.S. production.(files.gao.gov)

So although some U.S. PPE capacity was added, the overall system is still structurally reliant on imported gear, much of it from or tied to China.

3. There have been onshoring initiatives—but they are partial

The federal government has taken steps in exactly the direction Sacks described, including:

• A major BARDA contract with Phlow Corp in May 2020 to build domestic capacity for essential medicines and APIs at risk of shortage.(pharmajournalist.com)
• Dozens of Defense Production Act and related projects from 2020–21 onward to expand U.S. production of N95s, gloves, masks, and other critical supplies.(files.gao.gov)
• 2024 BARDA contracts—over $500 million—to source U.S.-made isolation gowns for the Strategic National Stockpile.(reuters.com)
• A 2025 FDA pilot to fast‑track approval of generics tested and manufactured entirely in the U.S., explicitly aiming to reduce reliance on foreign drug manufacturing.​(reuters.com)

These actions show meaningful movement toward greater domestic capacity, but they represent pockets of reshoring, not a wholesale return of “all that stuff.”

4. Conclusion

Sacks’ prediction was not just that there would be policy efforts or incremental shifts; he said that antibiotics, medicines, PPE, and similar survival‑critical products would “come back home,” ending U.S. dependence on China. Five-plus years later, the empirical record shows:

• The majority of essential medicines and APIs used in the U.S. are still imported, with China and India at the core of those supply chains.
• U.S. PPE production expanded temporarily but then contracted, while China again accounts for most of the medical protective gear used in U.S. healthcare.

Given that overall dependence on foreign—and especially China-linked—production remains high, the core outcome Sacks predicted has not occurred within the “coming years” after 2020.

Therefore, this prediction is best classified as: wrong.

Why this prediction is rated “wrong”

1. How TikTok’s U.S. status has actually been handled

   • In August 2020, Donald Trump issued Executive Order 13942 targeting TikTok/ByteDance under U.S. national‑security and ICT‑supply‑chain authorities, handled as a unilateral U.S. executive action rather than part of a broader trade bargain with China. (en.wikipedia.org)
   • In 2024, Congress passed the Protecting Americans from Foreign Adversary Controlled Applications Act, explicitly aimed at TikTok and other ByteDance apps. It conditions TikTok’s continued U.S. operation on divestiture from its China‑based parent, again on national‑security grounds, not as a negotiated trade concession with Beijing. (en.wikipedia.org)
   • Federal courts later upheld this law as constitutional, emphasizing its national‑security purpose (Chinese government access to data and influence), and framing the issue as sale or ban, not as an element in a reciprocal U.S.–China commercial deal. (apnews.com)
   • Subsequent U.S. executive orders in 2025 implement and sequence enforcement and potential divestiture under this statute; they continue to treat TikTok as a regulated “foreign adversary controlled application,” not as a chip in a broader market‑access swap. (whitehouse.gov)

   Across this timeline, the formal resolution path for TikTok in the U.S. has been: targeted legislation, CFIUS/national‑security review, and court challenges—domestic legal/regulatory processes, not a quid‑pro‑quo trade negotiation with China over unrelated U.S. firms.

2. Tesla’s China access was established independently and earlier

   • Tesla’s Shanghai Gigafactory was agreed with local government in July 2018, ground was broken in January 2019, and production began in late 2019, making it China’s first wholly foreign‑owned auto plant—well before the May 2020 podcast and before TikTok became a major U.S. political issue. (en.wikipedia.org)
   • Chinese policy changes allowing 100% foreign ownership for EV makers, plus local tax breaks and subsidies, explain Tesla’s market access and favorable treatment; these are documented as part of China’s industrial strategy to promote EVs, not as a concession traded for U.S. treatment of TikTok. (appen.media.gov.cn)

3. No evidence of the specific quid‑pro‑quo mechanism Sacks described

   • Sacks predicted TikTok’s ability to operate in the U.S. would “probably” be resolved inside a larger U.S.–China negotiation, where TikTok’s U.S. market access is swapped for something like Tesla’s right to sell cars in China.
   • Public reporting and official documents on both TikTok policy and Tesla’s China operations contain no indication that TikTok’s U.S. status has been explicitly bargained against reciprocal commercial access for Tesla (or a similar single firm) in China. Instead, they are treated as separate issues: TikTok as a national‑security/ownership problem; Tesla as a beneficiary of China’s EV‑industrial policy.

Because TikTok’s U.S. fate has been driven by unilateral U.S. national‑security law and litigation rather than a documented U.S.–China quid‑pro‑quo trading it against Tesla’s China access (which was already secured before the prediction), the specific mechanism Sacks described did not occur.

Conclusion: The prediction about how TikTok’s U.S. access would be resolved—via a Tesla‑style quid‑pro‑quo negotiation with China—has not come true; it is therefore wrong.

Evidence since 2020 shows some movement toward reciprocity-based arrangements in U.S.–China and related commerce, but not clearly enough to say Sacks’ vision has broadly materialized, nor clearly failed.

Points suggesting the prediction is at least partly on track

• U.S. policymakers have elevated “reciprocity” to a formal organizing idea. The True Reciprocity Act of 2023 declares it U.S. policy that any trade or investment negotiations with China should address non‑reciprocal arrangements and secure structural changes in China’s trade practices. (congress.gov)
• The U.S.–China Economic and Security Review Commission explicitly recommends that Congress adopt reciprocity as a “foundational” principle in all U.S.–China legislation, covering market access and regulatory parity. (uscc.gov)
• The draft EU–China Comprehensive Agreement on Investment (CAI) was framed by the EU as rebalancing asymmetries and increasing reciprocity of market access and a level playing field for EU firms in China. (policy.trade.ec.europa.eu)
• A 2025 White House proclamation on the Kuala Lumpur Joint Arrangement with China openly links tariff changes to addressing a “lack of trade reciprocity” and ties U.S. concessions (e.g., tariff modifications) to Chinese commitments on export controls and purchases—an explicitly quid‑pro‑quo structure. (whitehouse.gov)

Points cutting the other way

• The dominant U.S. approach to China under Biden has been framed as “small yard, high fence”: targeted export controls and investment restrictions on sensitive technologies for national security, not a general reciprocity‑for‑access bargain. (en.wikipedia.org)
• Business and policy analyses still describe the relationship as non‑reciprocal and deteriorating; U.S. firms cite tariffs and export controls plus continuing market-access barriers in China, rather than a new stable reciprocity regime. (apnews.com)
• The flagship EU reciprocity project (CAI) remains stalled and, as of 2025, senior EU trade officials say the EU has “no interest” in reviving it, citing worsening market barriers in China. (scmp.com)

Netting this out, reciprocity has clearly become a more explicit goal and features in some specific deals, but the overall structure of U.S.–China and allied commerce is better described as fragmented, security‑driven, and only selectively reciprocal. That mixed, qualitative reality makes the prediction neither clearly fulfilled nor clearly falsified, hence “ambiguous.”

By May 2022 (two years after May 2020), researchers had identified some genetic loci associated with COVID-19 susceptibility and severity, but these explained only a small fraction of outcome variability and did not enable precise, individual-level predictions of clinical course.

Key points:

• A large 2021 Nature paper from the COVID-19 Host Genetics Initiative reported 13 loci associated with infection or severe COVID-19, and explicitly noted that “much remains unknown about the genetic basis of susceptibility to SARS‑CoV‑2 and severity of COVID‑19.” (nature.com) This falls well short of a detailed genotype→outcome map.
• A comprehensive May 2022 review on The human genetic epidemiology of COVID‑19 found that polygenic scores (PS/PRS) for COVID-19 were weakly powered, with most variation explained by a few major signals (e.g., the 3p21.31 locus), and concluded that clinical application of such scores for SARS‑CoV‑2 susceptibility or severity was unlikely in the short term. The paper emphasized that PS might be useful for research but not for routine clinical prediction. (pmc.ncbi.nlm.nih.gov)
• Early and subsequent PRS studies for COVID‑19 severity showed limited discriminative performance: one study reported an AUC ~0.51 (essentially no clinical utility), explicitly concluding they could not create a clinically useful genetic test. (humgenomics.biomedcentral.com) Later work using more sophisticated PRS improved AUC only modestly (e.g., from 0.68 to 0.71 when adding a genome-wide PRS), underscoring that genetics added some risk stratification but still lacked strong individual-level predictive power. (cambridge.org)
• A 2023 analysis of polygenic scores for COVID‑19, drawing on Host Genetics Initiative data, similarly found that while genetic associations were statistically robust, they had limited predictive value for translational/clinical settings, with most predictive power coming from simple epidemiologic risk factors rather than genotype. (pmc.ncbi.nlm.nih.gov)

Together, these sources show that even by and shortly after May 2022, the field had:

1. Only a partial list of risk loci with modest effect sizes; and
2. No clinically reliable, fine-grained mapping from an individual’s genotype and health profile to a precise predicted COVID‑19 outcome.

That matches Friedberg’s prediction that “we’re not going to know that in two years”—i.e., we still would not have a detailed, reliable genotype→clinical‑outcome map for SARS‑CoV‑2 by roughly May 2022.