Simon Gros

---

In a wide-ranging onstage conversation at the All-In Summit 2024—later published by the All-In Podcast under the title “Peter Thiel: The Coming Collapse No One Is Prepared For”—technology investor and entrepreneur Peter Thiel frames contemporary politics, geopolitics, and technological change through a single organizing preoccupation: the distribution of agency under conditions of accelerating uncertainty. He is introduced as a long-standing Silicon Valley figure associated with PayPal, early backing of Facebook, and subsequent investments in companies built around unusually concentrated technical advantage. The hosts use that biography to position him as an analyst of asymmetric risk: someone who expects discontinuities, distrusts consensus narratives, and looks for the points where institutional habits fail to scale to new strategic environments.

Thiel begins with a maxim meant to justify contrarian action: in a world that changes quickly, the biggest risk can be the refusal to take risk. In his telling, the practical corollary is that rare, durable businesses tend to be “one-of-a-kind” enterprises—ventures that move from “zero to one” by making a disputed claim about how the world works and then building an organization capable of realizing the claim in products, markets, and defenses. The underlying premise is less motivational than methodological: if novelty is real, then there will be propositions about the future that sound implausible to most people; the investor’s task is to identify which implausibilities are actionable rather than merely eccentric.

From that starting point he moves into U.S. electoral politics, describing a stance that is simultaneously partisan in sympathies and pessimistic about governance outcomes. He says he remains broadly supportive of Donald Trump and JD Vance but has chosen not to contribute money to the 2024 campaign, presenting this as a judgment about marginal impact and predictable disappointment. His forecast is that even a decisive victory would likely be followed by “buyer’s remorse,” because elections are experienced as comparative choices while incumbency becomes an absolute evaluation: the winner inherits structural constraints that turn campaign contrasts into administrative compromises. In that segment he also alleges that, in a very close election, procedural “fortification” and outright cheating could determine the outcome—language he offers with the expectation that it is politically sensitive and contested. In a neutral rendering, the key point is that he treats institutional trust as fragile and expects legitimacy crises to recur when outcomes hinge on narrow margins.

Consistent with that diagnosis, he advocates reforms meant to compress discretion and expand legibility: one-day voting, sharply limited absentee balloting, stronger voter identification, and a national holiday for elections. He presents these as approximations of practices in other Western democracies and contrasts them with what he characterizes as a decades-long U.S. drift toward prolonged counting periods and rules that shift close to the election. The claim is not simply administrative; it is a theory of political stability. Shorter time horizons and more standardized procedures, he argues, would reduce incentives for strategic manipulation, restore confidence in results, and lower the temperature of post-election conflict.

Pressed on what would make him “surprisingly not disappointed” by a future Trump administration, Thiel lists two criteria that function as stress tests of state capacity. First, he points to fiscal arithmetic: meaningful deficit reduction without tax increases and without a contraction in GDP would, in his view, qualify as an unexpectedly strong achievement. Second, he depicts the global security environment as moving from “warm-ups” (he cites Ukraine and Gaza as precursors) toward the central systemic risk: a major U.S.–China confrontation triggered by Taiwan. His aspiration is defined negatively but concretely: avoiding a great-power war over the next presidential term would be better than he expects, and therefore a meaningful accomplishment.

On Taiwan, Thiel’s position is careful in form and stark in implication. He endorses the strategic logic of ambiguity—avoiding explicit commitments that would either invite immediate coercion (if the U.S. disavowed defense) or harden Chinese red lines (if the U.S. promised defense regardless of circumstances). He argues that overly precise declaratory policy can accelerate conflict by turning uncertainty into a timetable. At the same time, he states that Taiwan’s forced absorption by the Chinese Communist Party would be catastrophic, while also implying that it may not justify a third world war. The tension is left unresolved by design: the object is to hold two propositions together—grave moral and strategic loss on one side, unacceptable escalation risk on the other—while treating ambiguity as the least-worst instrument for delaying a crisis.

From geopolitics he shifts to the economic mechanics of conflict, predicting that a Taiwan invasion would trigger a rapid severing of U.S.–China commercial ties. He uses an analogy to Europe’s energy rupture with Russia: where Nord Stream represented a single high-value conduit, U.S.–China interdependence resembles “a hundred pipelines,” meaning that decoupling would be vastly larger in scope and more disruptive in second-order effects. Within that frame he offers a tactical business warning using TikTok as an example: if a cross-strait war occurred, he expects U.S. political tolerance for Chinese-linked platforms to collapse almost immediately, rendering “simulation” exercises about continuity irrelevant. This is presented as a practical argument for corporate contingency planning: firms that depend on Chinese operating environments should assume that geopolitical rupture can destroy enterprise value on a short fuse.

He then addresses inflation risk from decoupling. His view is that the effect could be significant but not necessarily as extreme as common models suggest, because production can migrate not only back to the U.S. (expensive), but to other lower-income countries (Vietnam, Mexico, and others) that can absorb relocated supply chains. The result, in his depiction, is a “negative-sum” adjustment: modestly adverse for the U.S. consumer relative to the old China-centered equilibrium, sharply adverse for China, and beneficial for alternative manufacturing hosts. In diplomatic terms, this is an argument that macroeconomic pain will be distributed, and that third countries will have incentives to position themselves as the new nodes of industrial capacity.

Asked whether war can be avoided through leadership-level engagement, Thiel responds with skepticism grounded in historical analogy. He invokes the “Thucydides Trap” idea—popularized in contemporary debate as the recurrent risk of conflict when a rising power threatens to displace an established one—and suggests that Chinese strategic culture may already assume such conflict as historically normal. His proposed first step is psychological rather than procedural: to reduce the chance of collision, U.S. policymakers must start by admitting that Beijing may believe confrontation is coming. He treats “happy talk” as dangerous because it can substitute for serious preparation, while also arguing that dialogue—including with adversarial regimes—often reduces the probability of miscalculation. The net effect is a posture that is neither pure conciliation nor pure confrontation: it is risk management under mutual suspicion.

The conversation’s technological core centers on artificial intelligence, which Thiel approaches with deliberate resistance to buzzwords. He argues that “AI” has meant many inconsistent things over decades, from speculative superintelligence scenarios to surveillance technologies. He contrasts two influential earlier framings: existential-risk narratives of an intelligence explosion versus geopolitical narratives of state-scale data and monitoring. Against both, he places the recent wave of large language models, which he characterizes as a more classical “Turing test” achievement: systems that can convincingly simulate human linguistic interaction. His emphasis is that language is close to the core of human social life, so automation in this domain is inherently disorienting. It forces questions that typical venture discourse avoids: what remains distinctively human, how labor markets reprice cognitive work, and which social roles prove robust to substitution.

As an investor, however, he returns to a narrower metric: profit capture. He compares the state of AI in 2023–2024 to the internet in 1999—an era when it was already clear the technology would transform society, but unclear which firms would keep the surplus rather than burn capital competing for position. In his telling, the early distribution of profits in AI is unusually concentrated in the hardware layer. He claims that Nvidia is capturing essentially all the industry’s profits while most other participants collectively lose money, because they are buyers of compute rather than owners of scarce compute supply. The strategic question, therefore, is not simply “who has the best model,” but whether chip-level advantage constitutes a durable monopoly and how long it can survive once public attention makes replication efforts inevitable.

Thiel links that profit observation to a broader critique of innovation patterns in the United States. He argues that significant progress has continued in “bits”—software, the internet, crypto, and AI—while progress in “atoms,” meaning applied engineering fields tied to physical infrastructure (energy, transportation, advanced manufacturing), has been comparatively constrained. He offers multiple causal stories: regulatory friction, risk aversion shaped by the twentieth century’s destructive technologies, and cultural or institutional shifts that favored status competition and bureaucracy over ambitious technical programs. He uses the Manhattan Project as a historical counterexample of rapid state-led organization, while acknowledging that contemporary American institutions—government and universities in particular—rarely reproduce that capacity. The implication is that the U.S. can still generate frontier companies, but the frontier is increasingly channeled into domains where iteration is fast and deployment barriers are low.

On macroeconomics, Thiel describes an economy he believes is being propped up by large deficits and government spending, with recession risk obscured by fiscal stimulus. He points to the political difficulty of reducing deficits without either tax increases or growth slowdowns, and he frames this as part of a longer stagnation narrative: if compounding technological progress is weak, then debt and redistribution become the default tools of political management. In his account, the last half-century contained two “one-time” growth accelerants that substituted for deep innovation—1980s deregulation and tax cuts, and 1990s globalization and labor arbitrage—but these levers are now largely exhausted, leaving fewer non-technological ways to expand real output.

His critique of higher education functions as a concrete case study in institutional fragility. He argues that universities have become more expensive, more bureaucratic, and less accountable to labor-market outcomes, with federal loan underwriting removing the underwriting discipline that a private credit market would impose. He presents student debt dynamics as compounding and cohort-dependent, and he connects these dynamics to the rationale for the Thiel Fellowship: paying young people to pursue entrepreneurial work rather than defaulting into credential accumulation. He also adopts a mixed ideological position: he endorses significant reform and a reduction of government involvement in student lending, while also suggesting some broad debt forgiveness on the grounds that many borrowers were sold a bad bargain. In his framing, however, forgiveness must be structured so that universities and financial stakeholders bear part of the cost, or else the system’s incentives remain intact and tuition inflation continues.

He closes with an explicit rejection of extreme optimism and extreme pessimism, treating both as forms of passivity: pessimism because it claims nothing can be done, optimism because it assumes things will work out without deliberate action. What he proposes instead is a posture of bounded agency: the future is neither fixed nor self-correcting, and therefore strategy consists in identifying where choices still matter, where institutions still have degrees of freedom, and where risk can be converted into advantage rather than simply endured. In the overall logic of the discussion, “collapse” is less a single forecast than a name for cumulative unpreparedness: electoral procedures that invite legitimacy disputes, fiscal trajectories that constrain policy, geopolitical entanglements that can sever supply chains quickly, and technological systems that transform labor and identity before governance and markets fully understand how to allocate benefits and costs.

Because the conversation was recorded in 2024 in the immediate run-up to the U.S. presidential election, it is also possible to situate one of Thiel’s central political claims against subsequent outcomes: the 2024 election took place on November 5, 2024, and Donald Trump ultimately won the presidency with JD Vance as vice president, defeating Kamala Harris and Tim Walz. That historical fact does not validate or invalidate Thiel’s broader institutional concerns; it does, however, clarify that his forecast of a potentially decisive Trump victory did align with the eventual result, even as his remarks about electoral manipulation should be understood as allegations and apprehensions rather than established findings.

Thiel returns repeatedly to the same underlying analytic move: he distinguishes between systems that still generate decisive outcomes and systems that have become proceduralized to the point that outcomes are delayed, contested, or diluted. In politics, he locates that problem in what he portrays as an expanding interval between voting and resolution, alongside rule changes that—whether justified as accessibility measures or criticized as vulnerabilities—create the conditions for durable suspicion. When he uses charged language about “cheating” or “stealing,” the substantive content, stated in more neutral terms, is a belief that the U.S. electoral system has accumulated discretionary layers (mail voting, extended counting, ballot handling practices, and administrative heterogeneity across jurisdictions) that make it difficult to persuade losing constituencies that results are both clean and final. His proposed remedy is procedural compression and standardization, presented as a legitimacy technology: fewer moving parts, fewer exceptions, faster closure, and clearer constraints on who can vote and how.

From there, he frames U.S. governance as constrained by a set of hard problems whose solutions are not ideologically obvious and whose failure modes are cumulative. On fiscal policy, he does not argue from a technical budget plan but from a macro-constraint: deficits that remain large even at the top of the business cycle imply either a future fiscal correction, a sustained reliance on debt expansion, or a return to growth that is fast enough to change the denominator in debt-to-GDP dynamics. In this register, he treats “reducing the deficit without raising taxes and without contracting GDP” as a shorthand for a genuinely difficult policy achievement, because it requires either uncommon political discipline, uncommon growth, or both. His broader pessimism about political outcomes is not just partisan skepticism; it is a diagnosis that modern administrations inherit structural conditions that limit the space for visible success and that increase the probability of post-election disappointment.

The geopolitical segment is organized around a risk hierarchy in which Taiwan functions as the decisive stress test of U.S.–China relations. Thiel’s emphasis on strategic ambiguity is presented as an attempt to preserve optionality and deterrence simultaneously: if the U.S. defines a rigid red line, it risks either provoking a test or committing itself to a war over contingencies it cannot fully control; if it defines a rigid non-commitment, it risks accelerating an invasion by lowering expected costs for Beijing. He therefore treats non-precision as an instrument—arguably the least fragile instrument—within a setting where precision can become self-fulfilling. Yet he also states, in plain terms, that Taiwan’s absorption would be catastrophic in political and moral meaning, while also insisting that catastrophe does not automatically justify a global war. The result is a deliberately tension-filled position: deterrence is necessary, war is unacceptable, and the optimal policy is therefore one that reduces the probability of decision while leaving room for decision.

He then translates geopolitical rupture into an economic scenario, arguing that a Taiwan conflict would likely trigger a rapid, politically enforced severing of commercial ties that currently operate as the connective tissue of globalization. The image of “a hundred pipelines” is meant to communicate scale and ubiquity: supply chains, capital flows, platform dependencies, and technical infrastructure that are routinely treated as stable background conditions could be interrupted not gradually but abruptly. Thiel’s TikTok example operates as a concrete illustration of how quickly legitimacy and security framing can transform market access. In his account, corporate strategy that treats geopolitical risk as a low-probability tail event is mispriced; the relevant question is not whether decoupling is desirable, but how quickly it could become mandatory under crisis conditions, and how much value would be destroyed in the transition.

On inflation and relocation, he argues against a simplistic model in which “decoupling from China” necessarily means “repatriating to the United States.” He instead points to the existence of many lower-income countries that could absorb production, suggesting that the real adjustment may be a re-routing of manufacturing networks rather than a reversal of globalization. In that frame, the macroeconomic effect becomes distributive: some cost increases for U.S. consumers and firms relative to the previous China-centered equilibrium, severe loss of export-linked leverage for China, and large gains for alternative manufacturing hubs. He calls this “negative-sum” not because it yields no winners, but because it reduces overall efficiency relative to the prior arrangement, even as it creates new regional beneficiaries.

Thiel’s remarks on avoiding conflict introduce a further layer of realism about the epistemic limits of diplomacy. He suggests that U.S. commentators frequently assume that deterrence and reassurance can be achieved through “happy talk,” while he believes the more relevant difficulty is that Beijing may already interpret the long-term trajectory as conflictual, independent of U.S. rhetoric. In that setting, he treats the admission of adversarial expectations as a prerequisite for stability: if one side believes confrontation is structurally inevitable, then superficial reassurance may be read as either deception or weakness. At the same time, he argues for maintaining communication even with objectionable regimes, because the alternative—no channels, no backchannel capacity, no credible de-escalatory interface—can leave states trapped inside domestic political narratives that penalize contact and reward escalation. The practical core of his point is that crisis management is a capability, and capabilities atrophy when politics makes them taboo.

The second half of the conversation shifts from geopolitics to technology, where Thiel’s stance is both cautious and emphatic. He expresses hostility to buzzwords as a category, treating them as markers of herd cognition and as signals that people are substituting shared vocabulary for shared understanding. “AI,” in his telling, is an especially overloaded term: at different times it has meant general intelligence, automation, robotics, predictive analytics, surveillance, and marketing. He recalls earlier frameworks that dominated elite debate—one oriented around existential risk from superintelligence, another oriented around surveillance and geopolitical competition—and then argues that large language models arrived as an “in-between” reality that did not fully validate either camp’s most definitive claims. The distinctive significance he assigns to language models is not simply technical performance; it is the social meaning of linguistic simulation. If a machine can convincingly imitate human conversation, then a line that many people tacitly used to define human distinctiveness becomes less secure, and the philosophical question of what remains uniquely human becomes harder to avoid.

He also places AI inside a familiar investment-historical analogy: the internet in 1999. The point of the analogy is not that AI will “crash” in the same way, but that a transformative technology can be simultaneously real in its long-run importance and confused in its near-term business models. In such phases, he argues, capital tends to overpay for narratives while under-analyzing profit capture. His preferred analytic anchor is therefore simple and deliberately unromantic: who is making money right now, and why. In the AI ecosystem he describes, the answer is concentrated at the compute layer: the primary beneficiaries are the companies that sell the chips and the infrastructure required to train and run models, while many downstream firms spend heavily without clear pricing power. He highlights Nvidia as the emblematic winner, describing a situation in which the hardware provider captures an unusually large share of the industry’s economic surplus while other participants collectively operate at losses.

This profit-structure observation leads to a more general argument about how Silicon Valley’s habits may be mismatched to the current technological moment. For decades, the dominant success pattern in consumer internet and software involved scalable distribution, network effects, and relatively asset-light expansion. The AI wave, by contrast, makes the cost of compute and the supply of advanced hardware central. Thiel suggests that the cultural training of the tech elite—its instinct to analyze software layers and user growth rather than physical supply chains and industrial capacity—creates a blind spot. His provocative aside about 1993 and the shift away from electrical engineering is meant to dramatize that blind spot: as the internet era accelerated, many ambitious people avoided hardware and “atoms,” leaving fewer entrepreneurs and fewer institutional competencies in the domains that now determine leverage in AI.

When the discussion turns to labor and automation, Thiel avoids a definitive prediction, but he frames the uncertainty in terms that resemble historical industrial transitions. He notes that the Industrial Revolution increased output and eventually created new forms of work, while acknowledging that contemporary AI could concentrate displacement differently. He registers, without endorsing, a recurring social fear: that a world with vast automated capability could generate a population that is economically redundant. His response is not a confident denial so much as a skepticism toward simple linear extrapolations—either utopian or catastrophic—paired with an insistence that distributional effects will be politically decisive. He also uses culturally loaded examples (Hollywood, university administration, contemporary academic writing) to imply that roles built around formulaic language production may be particularly exposed to substitution, while roles built around embodied performance, irreverence, or social risk-taking may be more resilient. Interpreted neutrally, the claim is that AI will differentially pressure occupations depending on how much of their output can be replicated through pattern completion and stylistic conformity.

Thiel’s innovation diagnosis is then broadened into a civilizational critique about “bits versus atoms.” He argues that meaningful progress has continued in software-centered domains, while applied engineering domains tied to physical reality—energy, transport, advanced manufacturing, large-scale infrastructure—have slowed. He offers several overlapping explanations rather than a single cause: regulatory accretion, the exhaustion of easy gains, cultural risk aversion shaped by the destructiveness of twentieth-century technologies, and institutional bureaucratization. His reference to the Manhattan Project functions less as praise for militarization than as evidence that organizational decisiveness, when it exists, can compress timelines that otherwise stretch into decades. The implicit contrast is that contemporary democratic societies often struggle to mobilize comparable coordination for non-military infrastructure goals, even when the economic and strategic stakes are large.

On the U.S. domestic economy, Thiel describes a situation he believes is precarious beneath surface indicators. He points to patterns such as repeated downward revisions in jobs data and to the broader problem of interpreting macro signals in a period where fiscal policy is unusually expansionary. His central contention is that government spending has functioned as a stabilizer that delays recession signals, while the deficit trajectory itself becomes a longer-run destabilizer. He frames this as a structural problem: in a setting where growth is not being driven by sustained technological progress in the “atoms” domain, politics reaches for non-compounding tools—debt expansion, redistribution, or one-off policy shifts—that can create temporary relief without restoring durable productivity growth.

He sketches a historical narrative in which late twentieth-century growth was supported by two large, partially non-repeatable moves. One was the Reagan–Thatcher period of deregulation, tax reductions, and corporate consolidation—measures that can generate a step-change in efficiency but do not necessarily compound indefinitely. The other was the Clinton–Blair period’s embrace of globalization and cross-border labor arbitrage, which lowered costs and expanded markets while producing its own externalities. In his view, both levers are now constrained: further tax cuts are politically and fiscally limited, and further globalization is challenged by geopolitical fragmentation. This creates a strategic vacuum in which the economy’s “next engine” must come from renewed real innovation rather than from rearranging existing incentives.

The education segment operates as an institutional case study of what Thiel calls a bubble: a system where prices rise faster than underlying value, supported by financing mechanisms that mute normal market discipline. He argues that U.S. higher education costs have been inflated by federal underwriting of student loans, by accreditation regimes that function as gatekeeping rather than quality assurance, and by internal administrative expansion that increases overhead without proportionate improvements in teaching or outcomes. He cites the growth of student debt across decades and emphasizes cohort dynamics: earlier cohorts could often pay down debt over time, while later cohorts increasingly carry debt longer and sometimes see balances rise. His Thiel Fellowship is presented as a practical intervention: not a universal replacement for universities, but a signal that for at least some high-ability individuals, the opportunity cost of traditional credentialing has risen relative to building projects and firms.

Importantly, he combines what would typically be labeled “right” and “left” policy instincts in this segment. On the one hand, he argues that the federal government should reduce or exit its role as the primary underwriter of student lending, thereby forcing universities and lenders to reintroduce underwriting discipline based on expected earnings and program value. On the other hand, he says that many students were effectively mis-sold and therefore some broad debt forgiveness is justified—but he argues that forgiveness should not be designed as a pure taxpayer transfer. Instead, he suggests that universities and financial stakeholders should share the burden, so that relief does not simply reset the system for another cycle of tuition inflation and administrative growth. Neutralized, his position is a proposal to pair partial relief with incentive reform: distribute losses to the institutions that benefited from the expansion, and change the financing structure so future cohorts face a more reality-based pricing regime.

In the closing exchange about “who wins AI,” Thiel declines to rank specific leaders and instead underscores his broader point about persuasion and attention: charismatic founders can be compelling in conversation, but the market structure often determines outcomes more than rhetoric does. He returns to Nvidia as the anchor, arguing that the decisive contest in AI may not be among consumer-facing model companies but among those who control the scarce inputs of compute and advanced manufacturing. He also suggests that extreme public attention can be double-edged. Early on, attention is fuel for recruitment and capital; later, attention can attract regulatory scrutiny, competitive imitation, and politicization that erodes strategic flexibility. His remark that Nvidia briefly becoming the world’s largest market-cap company represented a “phase transition” is offered as a way to describe that pivot: once a firm becomes symbolically central, it becomes harder for it to operate as a quietly compounding technical enterprise.

He ends with what functions as the conversation’s philosophical spine: he rejects both fatalism and complacency. Extreme pessimism, he argues, becomes an excuse to do nothing because nothing can be changed; extreme optimism becomes an excuse to do nothing because everything will work out. He prefers a stance oriented around agency—constrained, fallible, but real—where outcomes depend on choices, organization, and the willingness to confront risk without being captured by either panic or denial. In the context of the entire discussion, “collapse” is less a singular prediction than a composite warning about unpreparedness: political systems that struggle to produce legitimate closure, fiscal systems that substitute debt for productivity, geopolitical systems that can sever interdependence abruptly, and technological systems that advance faster than societies can decide how to allocate their benefits and absorb their disruptions.