The Commanding Heights

AI’s Commanding Heights

The authors argue that lasting AI wealth will accrue less to model inventors than to owners of scarce, controllable economic positions through which increasingly commoditized intelligence must pass.
The model-building industry may keep expanding investment even as returns decline: spending that is rational for individual firms can collectively destroy capital, while larger fortunes emerge elsewhere in the economy.
The text presents American growth, household wealth, and credit as increasingly dependent on AI investment, stock prices, and model capabilities rising together, creating pressure for government financial backstops.
The authors predict that only America and China will sustain sovereign frontier AI, with commercial models diverging from classified systems developed for statecraft and governments exerting control without formal nationalization.
AI-driven demand for dollar stablecoins could extend American monetary power, drawing global users away from local currencies and indirectly into financing US Treasuries.

Gutenberg went insolvent, and the great railroad fortunes were captured by men who never built a locomotive.

AI Reshapes Economic Power

The author predicts that AI will redirect gains from developing countries’ back-office work toward wealthy economies, widening inequality while citizenship and licensing protect some rich-country workers.
Durable economic power will belong to owners of the new “Commanding Heights”: scarce, controllable positions that capture a growing share of the activity enabled by increasingly widespread intelligence.
Firms are expected to reorganize around a few operators, compensated and governed like capital, who direct work performed mostly by AI models.
Investors risk mistaking yesterday’s dominant assets or today’s inventions for lasting sources of returns; the author argues that portfolios and investment mandates must change to capture emerging positions of power.
Ravi’s story gives these shifts a personal dimension: an IIT-trained professional who built a comfortable life in Bangalore now manages five North American insurance clients instead of the usual one or two because of AI.

The rest of his career, his life, is laid out predictably before him like the table set he sees in the morning light as he moves through the kitchen, pouring tea. The phone continues to buzz.

Automation Behind the Boom

Ravi’s management role has shifted from supervising people to approving work produced by Chinese AI accounting systems, while hiring freezes and unfilled vacancies shrink his team.
Automation increases company margins and Ravi’s earnings, but relentless monitoring leaves him sleep-deprived and strains his marriage.
Ravi increasingly approves AI reports without reading them, trusting systems that complete in hours work that once took employees weeks.
AI is reshaping his household too: he replaces his daughter’s private schooling with an AI tutor whose curriculum he can oversee through an app.
Record stock prices and official optimism contrast with layoffs, violent unrest, and economic instability across developing countries; Ravi turns off the news to concentrate on his agents.

The agents never sleep. Now, neither can Ravi.

Productivity Without Security

Ravi’s mortgage, family expenses, and household wages consume most of his income, leaving little opportunity to benefit from rising stock prices.
His outsourcing company announces immediate cuts to half his division and plans to eliminate 90% of subsidiary staff, presenting technology-driven restructuring as an exciting path to higher margins.
The Labor Modernization Scheme provides retraining support while enabling more aggressive workforce reductions; managers are offered potential bonuses tied to cuts and productivity gains.
Ravi works through the announcement without time to reflect, while record stock prices contrast with widespread layoffs, darkened offices, and reports of disturbances in the city.
An inexpensive digital tutor accelerates his daughter’s education, but automation makes the professional future he envisioned for her uncertain, while a weakening rupee and tighter visas complicate studying abroad.

She’s going far faster than she ever would have in school, though now he doesn’t know toward what.

India’s Displaced Future

Ravi earns more than ever, but the rupee’s decline makes overseas travel increasingly unaffordable, exposing the limits of his apparent prosperity.
His father’s sacrifices to fund his education contrast with a world of nearly free AI tutoring and dwindling opportunities for intellectual work.
Mass layoffs have pushed unemployment among young men in Bangalore above thirty percent, fueling unrest as India’s modernization leaves workers behind.
The prime minister responds with arrests and restrictions on gatherings, suggesting Pakistani interference rather than confronting the economic causes of public anger.
Insulated from the streets by their guarded apartment complex, Ravi and his wife watch smoke rise over Bangalore and silently reach for each other’s hands.
The section transitions from India’s upheaval to the origins of artificial general intelligence in San Francisco, connecting technological development with its distant human consequences.

India’s future is bright, and it will be built on jobs that do not need these men.

Capital Commands Machine Intelligence

The author portrays AI deployment as a wartime-scale mobilization without a war, fueled by private capital and projected infrastructure spending approaching and exceeding $1 trillion annually.
Investors across the financial system—from family offices and venture funds to dentists investing through private credit—are financing the AI buildout in pursuit of extraordinary returns.
The author rejects the comforting interpretation that AI investment is merely a bubble, arguing that its durability and accelerating financial momentum are the more unsettling possibility.
Investors’ repayment demands will push machine intelligence beyond experimental technology settings into commercially productive work, with New York finance shaping how its value is captured.
Aligning with Leopold Aschenbrenner’s forecasts of improving intelligence, immense expenditure, and geopolitical stakes, the author seeks durable investment positions that control the passage of AI-generated wealth.

The race to deploy machine intelligence is the first wartime mobilization to come without a war.

Invention and Displaced Wealth

As investment in machine intelligence surges and its price falls, the central question is where investors will find returns once the technology becomes widely available.
Gutenberg’s printing press transformed society, but its inventor became insolvent, illustrating how enormous social value can coexist with poor financial rewards for innovators.
Examples from oil, cotton processing, and transistors suggest that wealth generated by transformative technologies often accumulates far from their places of invention.
Antitrust action forced AT&T to license Bell Labs’ transistor cheaply, helping the technology spread globally and enabling others to build immense fortunes.
The authors expect machine intelligence to follow this historical pattern: early investment and invention may not identify the eventual centers of profit.

And so the exquisite technology became profane and fertile. Escaping, as it always does.

Where Technological Wealth Settles

Railroad investment generated enormous wealth despite repeated booms and busts, but much of it went to those exploiting transportation networks rather than building locomotives.
The author argues that transformative technologies reward owners of the infrastructure that carries and distributes their benefits more than the original inventors.
Antitrust action reveals where economic power concentrates; the text presents Standard Oil’s control of transportation as central to its dominance and eventual breakup.
The semiconductor industry illustrates a similar separation between technical difficulty and financial reward: pioneering lithography enables advances, while chip designers capture substantial value.
Applied to emerging machine intelligence, this pattern suggests that the biggest returns may accrue far beyond today’s heavily funded laboratories, in places not yet apparent.

Antitrust is a map of where wealth concentrates, downstream of the direction and size of returns.

Where AI Value Accrues

The author describes AI training and inference as the first stage of a technological lifecycle: proving that machine intelligence can exist and operate, while requiring enormous infrastructure and continuous spending.
Scaling creates a financial treadmill: investors seeking returns must fund further growth, even as operating margins shrink and the costs of building frontier models rise.
Investors focused on familiar patterns and near-term valuations may misunderstand AI’s historical significance and where its lasting economic value will emerge.
The proposed second stage shifts value away from model-building laboratories toward distribution networks and other strategic points where machine intelligence transforms the physical world.
Rapid imitation commoditizes once-exclusive capabilities, pushing prices toward the marginal cost of serving them; the author therefore argues that durable profits are unlikely to reside in the model layer itself.

Capital is an animal that survives on pattern recognition. It is poorly suited to moments, such as the one that we live through now, of true historic disjoint.

AI Abundance, Sovereign Power

The author argues that transformative AI adoption remains concentrated among a small population, especially software engineers, while most of the economy has yet to experience substantial change.
Rapidly falling intelligence costs and enormous infrastructure investment are expected to spread AI throughout the economy, making powerful capabilities widely affordable.
This expansion creates a financial contradiction: investors keep funding capacity even as prices per task collapse, potentially exposing a broadly distributed capital base to heavy losses.
The text distinguishes valuable frontier models from commoditized intelligence priced near compute costs, while questioning whether frontier labs can sustain margins sufficient to justify the investment boom.
The author predicts that only America and China will retain the capacity for truly frontier training, creating a world where closely guarded sovereign intelligence coexists with ubiquitous commercial AI.

In that world intelligence will be both exquisite and commodified.

Intelligence, Dependence, and Displacement

Sovereign states, rather than private financiers, are expected to control the pace of frontier AI development as it becomes a focus of great-power competition.
Beyond the frontier race, spreading AI globally promises enormous profits while making users dependent on the winning sovereign’s technology stack and distribution networks.
Malena, an Argentine graphic designer, initially benefits from AI tools that let her triple her client list and run her business alone with automated assistance.
Dollar-pegged stablecoin payments protect Malena’s earnings from peso depreciation, illustrating how AI distribution and digital payment infrastructure can become intertwined.
After learning from users such as Malena, the next generation of AI is sold directly to her clients, eliminating her livelihood while model developers and payment providers profit from both her adoption and her replacement.

They test whether they still need her. They quickly realize that they don’t.

The Neo-Dollar Expands

The author argues that AI distributed through widely used apps will consolidate digital infrastructure and give platform owners proprietary data about human activity.
Because these platforms rely on American infrastructure and dollar-based financial rails, their growing use could generate additional dollar demand and lucrative transaction fees.
The “Neo-Dollar”—the network linking USDT, its distribution channels, and its adoption—could extend financial power beyond merely raising technology companies’ valuations.
Despite declining reserve share and diversification efforts, the dollar remains dominant: it accounts for more central bank currency reserves than all other currencies combined and participates in roughly nine out of ten foreign-exchange transactions.
The author’s central distinction is between national and individual adoption: while the dollar’s position among nations remains steady, internet-based access is expanding its reach to people unable to open American bank accounts.

The power to manufacture Dollar demand and intermediate its transactions is power itself.

From Wages to Assets

The text envisions dollar-backed artificial intelligence serving billions outside the Sinomodern economic zone, helping users escape weakening currencies while generating profits for U.S.-registered firms.
Turning to domestic consequences, the author argues that the United States faces a second major financial reordering after an earlier shift undermined the wage-based economy.
The earlier social bargain linked steady employment to rising living standards, middle-class security, and dependable retirement benefits.
The author contends that post-2008 quantitative easing and near-zero interest rates sustained financial markets by raising asset prices rather than wages, disproportionately benefiting asset owners.
The passage argues that the Federal Reserve’s 2020 intervention deepened this divide, citing the top 1% of U.S. households holding roughly $60 trillion—more wealth than the bottom 90% combined.

The result was a miracle recovery for those who owned assets, and a slow strangulation for those who earned a wage.

AI and Concentrated Wealth

The authors argue that monetary policy has deepened a K-shaped economy, enriching asset owners while inflation erodes the benefits of government payments to others.
They predict a second, technological phase of inequality: firms will replace much paid labor with AI models, leaving small groups of operators compensated and governed like capital owners.
As intelligence becomes cheap and widely available, the authors expect value to shift toward scarce assets, concentrating wealth further and dividing even the ownership class.
The authors frame commercial AI as a loss leader rather than the primary product: inexpensive intelligence attracts users and data, while profits come from controlling the conduit.
They anticipate that competitive copying and state intervention will eventually remove frontier AI from public sale, pairing ubiquitous consumer intelligence with tightly controlled, sovereign-backed capabilities.

The first act split the owners off from the earners. The second will split the owners themselves.

Beyond the AI Buildout

The author predicts that AI will shift from enhancing early adopters to replacing workers at scale, bringing widespread displacement and social unrest.
The greatest enduring fortunes may accrue not to model builders but to owners of scarce infrastructure and channels through which AI reaches the broader economy.
AI labs and hyperscalers may keep increasing investment even as returns decline; spending that is rational for individual firms can destroy capital collectively.
Debates over an AI bubble contrast fears of infrastructure overbuilding with expectations of expanding demand, while rapid model advances continue.
A vast financing network—from venture investors to insurers, private lenders, and retirement savings—is funding the AI infrastructure boom, but its focus may obscure opportunities beyond the immediate buildout.

In the world to come, one either owns the rails or pays those who do.

AI’s Industrial Infancy

The author describes AI as still in its infancy: building and serving models is a historic achievement, but remains largely confined to the infrastructure and token-generation stage.
Despite falling costs, AI has penetrated everyday labor only minimally; relatively few people know how to integrate it into workflows or profit from it.
Early adopters are generating substantial revenue for AI labs through growing usage, even while current applications remain modest compared with the technology’s implied potential.
A wide adoption gap separates technically informed enthusiasts from ordinary workers and small businesses, who often lack the knowledge or tools to deploy advanced AI effectively.
Early locomotives offer a historical analogy: their first productive uses improved existing industrial routes rather than creating broad transport networks, suggesting the importance of applying new technology where demand already exists.

For machine intelligence, this is just a summer job.

AI’s Captive Railway Phase

Early commercially viable locomotives improved existing coal-hauling operations, replacing horses to reduce costs rather than creating entirely new markets.
These industrial contracts funded technological development, much as today’s AI customers provide revenue that sustains model builders.
Railways delivered their greatest economic gains when they became open networks connecting independent users, enabling activity their owners had never planned.
The railroad sector captured only 7 percent of railways’ social return in 1890, illustrating how transformative infrastructure can generate benefits far beyond its providers.
AI remains in a comparable captive-route phase: deploying it effectively requires sophisticated customers who can select models, redesign workflows, write specialized instructions, and manage uneven capabilities, limiting wider adoption.

Managers must learn the spiky distribution of their new electronic employees, who talk just like people, but don’t act at all like the personages they’ve replaced.

AI Diffusion’s Scaling Bottleneck

Enterprise AI adoption is increasingly uneven: OpenAI’s highest-usage firms generated 8.3 times as many output tokens per active user as middle-tier firms in June 2026, versus 2.6 times in January.
Deployment companies embed engineers in businesses, while AI-driven services roll-ups acquire traditional firms to gain the workflow access and authority needed to implement AI.
These approaches may create durable, profitable businesses, but hiring specialists and integrating acquisitions cannot scale fast enough to match the growing abundance of intelligence.
Investors’ fixation on usage, revenue growth, margins, and valuations risks narrowing their attention to today’s successful AI businesses rather than the broader economic opportunity.
Strong revenues and positive inference margins demonstrate demand for the current AI market, termed “Genesis,” but the author argues that the eventual intelligence economy will be orders of magnitude larger.

Slowly, the blinders form to block anything but those metrics, and situational awareness, however robust its owner believes it to be, contracts around a certain San Franciscan summer of love.

AI’s Unstoppable Capital Cycle

AI revenue feeds a reinforcing investment loop: labs finance larger training runs, while high valuations attract new companies and capital seeking profitable opportunities.
Unlike mature businesses that generate surplus cash, the AI ecosystem described as “Genesis” is expected to keep absorbing hyperscaler cash flows and capital-market funding.
The author anticipates enormous demand growth as intensive users become more productive and AI reaches billions of additional users through digital networks.
Infrastructure bottlenecks continually shift, directing investment toward the next scarce resource while AI-related concerns rapidly enter political debate.
The usual capital-cycle brake may fail: labs may keep investing despite weak incremental returns because falling behind the technological frontier could jeopardize the value of everything they have already built.

But in Genesis, there may never be a signal that tells participants to stop building.

AI’s Strategic Spending Imperative

Hyperscalers invest in AI to defend existing businesses and secure future profit opportunities, making continued spending individually rational even when industry-wide overbuilding destroys capital.
As frontier AI gains military, cyberwarfare, and intelligence value, its development increasingly resembles a sovereign arms race rather than a conventional investment seeking financial returns.
The author argues that governments will decisively shape who controls advanced AI, treating frontier capabilities as strategic assets comparable to nuclear deterrents or carrier strike groups.
The AI production system called Genesis is expected to keep expanding regardless of private returns, delivering cheaper, more abundant intelligence while leaving the ultimate distribution of economic value unresolved.
The next section argues that American growth, household wealth, and credit increasingly depend on AI spending, stock prices, and model capabilities rising together, prompting Washington to backstop expansion despite investment impairments.

The ouroboros is no longer fed by ROIC , but by the coercive laws of competition.

The Ever-Rising AI Line

The author portrays AI infrastructure spending as a historic industrial buildout, with hyperscaler investment climbing from hundreds of billions toward trillions of dollars.
Rising stock indexes turn ordinary workers’ retirement savings into capital driven by an apparently insatiable demand for continued growth.
AI capability benchmarks form another ascending curve, prompting speculative predictions about machines eventually exceeding human capacities.
The section’s central claim is that infrastructure spending, financial markets, and AI capabilities are manifestations of the same interconnected growth trajectory.
Sustaining belief in exponential growth means accepting increasingly staggering budgets and industrial demands; the author depicts skeptics as repeatedly overtaken by actual spending.
A 2024 Epoch estimate projected frontier-model training costs growing 2.4-fold annually, reaching roughly $1 billion per run in 2027, compared with an estimated $4.6 million in cloud-compute costs for GPT-3 in 2020.

What seemed an absurd number becomes a capital budget, and then an inadequate one. The line continues upward.

AI’s Widening Cost Divide

Frontier AI training costs are estimated to be rising 3.5-fold annually, with reported billion-dollar runs arriving earlier than forecast and projected costs reaching $10 billion within two years.
Training compute demand is growing faster than available chip computing power, while efficiency improvements are being reinvested in larger runs rather than reducing frontier spending.
Escalating training costs threaten to exclude more competitors, concentrating frontier development among organizations able to finance increasingly expensive models.
Meanwhile, the minimum price of achieving fixed benchmark performance has reportedly fallen about 47% per quarter since 2023, making once-premium capabilities cheap commodities.
The combination of soaring capital expenditure and rapidly falling output prices could strain hyperscaler cash flows around 2027, increasing dependence on external financing and potentially constraining American AI growth.

And at each order of magnitude, more participants will be left behind, unable to pay the frontier’s cover charge.

AI’s Expanding Financial Leviathan

Epoch projects that hyperscalers’ capital spending, growing at 70% annually, will overtake operating cash flow, growing at 23%, forcing greater reliance on outside financing.
The author warns that AI infrastructure financing could become a securitized asset class, increasingly exposing ordinary people’s deposits, pensions, and retirement savings to speculative expectations about machine intelligence.
Hypothetical IPOs of major AI labs would deepen this exposure by embedding their shares in market indexes and tying millions of investors’ fortunes to AI’s success.
The cited projections anticipate a recovery after a difficult 2027, with hyperscaler free cash flow reaching roughly $120 billion in 2028, $300 billion in 2029, and $400 billion in 2030.
The text traces the available pool of private capital to late-1970s tax changes and pension-investment reforms, then introduces conglomerates as part of that history: businesses justified by superior capital allocation but ultimately run for management’s benefit.

Their hopes and dreams, held in a checking account.

Private Equity’s Reinvention

Early buyout investors profited by dismantling conglomerates whose individual businesses could command higher valuations separately, helping establish the institutional private equity industry.
Buyout firms recognized that stable, profitable companies could support substantially more debt, allowing buyers to finance acquisitions with very little equity and use the companies’ cash flows to repay borrowing.
Leverage magnified ownership gains: debt repayment increased equity value even without business growth, while banks and the emerging high-yield bond market supplied acquisition financing.
As funds grew and competition intensified, spectacular buyout returns became harder to achieve, prompting managers to seek more dependable sources of revenue.
Private equity firms evolved into financial conglomerates, using recurring management fees, public listings, and privately originated investment-grade credit to expand beyond buyouts and tap investors’ larger bond allocations.

The men who had made their fortunes dismantling conglomerates were building financial conglomerates of their own.

AI Swallows the Economy

The author portrays machine intelligence as an organism driven to grow, predicting that it will absorb major economic sectors and outpace those it cannot absorb.
The text describes a split American economy: AI investment and stock valuations boom while the non-AI economy contracts, leaving professional workers vulnerable despite record market indices.
To illustrate this divergence, the text reports that non-AI equipment investment fell 17 percent in late 2025 while computing-equipment investment rose 61 percent.
The author compares the AI infrastructure boom to railroad expansion, arguing that its pace exceeds the housing and telecom booms and is making financial markets increasingly dependent on AI.
As the intelligence economy overtakes the old economy, the author predicts that political leaders will prioritize its protection, extending to it the concern once reserved for the broader economy.

One economy is dying, the other has both hands around intelligence, and it’s booming.

AI’s Economic Stranglehold

The author portrays AI’s transformation from a niche technology into the central driver of the U.S. economy, making its political influence increasingly difficult to dislodge.
As AI firms dominate markets, growth, and retirement portfolios, Washington’s priorities become increasingly tied to the industry’s success.
An account of a presidential medal ceremony honoring major technology executives illustrates the growing alignment between government prestige and the intelligence industry.
This economic concentration creates systemic vulnerability: Chinese price competition, distillation attacks, or disappointing earnings could destabilize AI companies and the broader market.
Because market valuations depend heavily on anticipated future AI growth, even a moderate slowdown could erase substantial paper wealth and trigger a sustained selloff.

You will long for the days when Washington and its creatures talked of anything but machine intelligence.

AI Debt’s Systemic Risk

Identical investment-grade ratings can conceal very different systemic risks: ratings describe expected losses, not whether defaults will coincide with an economy-wide crisis.
As AI absorbs more capital and permeates the economy, apparently diverse bonds increasingly depend on the same spending cycle to sustain repayments, collateral values, buyers, and lenders.
An AI downturn could sharply reduce creditor recoveries because distressed data centers would need customers and buyers drawn from the same struggling industry; circular financing arrangements would deepen the losses.
AI adoption could keep advancing even as the infrastructure’s financing collapses, with temporary downturns causing permanent investor losses and creating opportunities for distressed-asset buyers.
Job losses and pension damage would generate intense political pressure to intervene. Herbert Hoover’s legacy illustrates how an economic collapse can overshadow a leader’s earlier achievements.

A creditor taking possession of a data center will have to look for a new customer among the companies that just vacated it.

AI, Markets, and Sovereignty

The authors argue that dependence on the intelligence economy will make its failure politically intolerable, pushing the U.S. government to sustain AI development despite popular backlash and falling prices.
Rather than relying only on congressional funding, AI labs may offer the state equity stakes, creating a deeper alignment between government interests and corporate survival.
Expanding stock ownership, including through accounts established at birth, ties ordinary Americans’ financial security to rising markets and strengthens political pressure to keep valuations climbing.
The text predicts that only America and China will retain full sovereign agency through continual frontier-model development, with classified state-oriented AI diverging from commercial offerings.
China’s relative freedom from pressure to protect household retirement investments allows it to tolerate low lab valuations and cheap open-weight models, while intense party-supervised competition drives AI prices downward.

Will you place your faith in the line? If you are a market participant, you already have.

China’s AI Price Challenge

The text argues that inexpensive, open-weight Chinese AI models increasingly match American systems for most consumer tasks, eroding U.S. labs’ competitive advantage.
Price compression threatens the revenues expected to finance continued American AI scaling, potentially undermining plans for frontier computing infrastructure.
The authors distinguish commodity intelligence, premium commercial frontier models, and a sovereign frontier held in state reserve; China sets the commodity floor while America’s commercial lead narrows.
In the text’s account of July 2026, Moonshot’s Kimi K3 release and China’s commitment to open-weight AI provoke American allegations of distillation and restricted-chip access, alongside sanctions threats.
The passage depicts those threats as producing no meaningful regulatory action, with Kimi becoming widely adopted, and concludes that the American state is not yet acting as a sovereign backstop for AI.

What good is a frontier model if only a portion of the expected compute ends up being built?

Fortifying Sovereign AI

The author predicts that financial strain on American AI labs and shareholders will draw the U.S. government into a more protective role.
The text defines the AI “artifact” as a reproducible formula of intelligence, arguing that widely available models will be distilled and commodified, driving prices down.
Hardware export controls are portrayed as inadequate: the author argues that China can develop advanced chipmaking technologies rather than remain dependent on restricted imports.
The author anticipates a shift from controlling hardware exports to protecting model parameters and other components of frontier intelligence, through increasingly intimate and informal cooperation between labs and the state.
A widening divide is predicted between restricted sovereign AI capabilities and commercially available models, leaving the most advanced intelligence inaccessible to both ordinary Americans and foreign rivals.

Export-control decree by whispered insinuation, not via public policy and certainly not via think tank.

AI Value Across Borders

Frontier AI models concentrate enormous capital investments into portable, reproducible files, making their economic value unusually vulnerable to copying.
Broad access enables model distillation and commodification, potentially reducing profits to thin margins before investors recover their costs.
The author argues that protecting a sovereign AI frontier serves both American investors’ valuation expectations and the state’s interest in securing immense financial commitments.
Scenes of everyday Chinese life juxtapose memories of scarcity with ubiquitous digital technology, illustrating modernization beyond headline economic indicators.
China has less incentive to preserve high model-layer valuations: it can pursue AI-driven economic efficiencies while funding commercially competitive models at far lower valuations than American rivals.
This asymmetry lets Chinese AI development threaten the American investment proposition without matching the scale of capital and expected value attached to it.

In the hutongs, old men who remember hungry days will gather after a long day of loitering in public parks. They remember the ration coupons during the 70s.

Wealth Shapes AI Policy

The author argues that differences between Chinese and American AI industries are better explained by financial structures than by cultural stereotypes or sweeping historical narratives.
Household wealth composition shapes state incentives: governments prioritize wealth preservation and creation, influencing how they regulate and support emerging technologies.
Chinese households hold far more wealth in housing and cash than in equities, making the Chinese state less compelled to support stock-market valuations.
China’s property crisis illustrates the political importance of housing: tighter borrowing rules undermined developers, and Evergrande defaulted on more than $300 billion in liabilities.
In 2022, Chinese households held 47% of their wealth in real estate, 24% in cash and deposits, and only 15% in equities; the author sees this allocation as central to China’s willingness to commoditize frontier AI rather than protect lofty corporate valuations.

China is fascinating, not least because it causes the most original minds to lapse into the most unimaginative and incurious formulations.

Retirement Rides the Market

The author contrasts China’s economic priorities with America’s dependence on rising stock prices, arguing that Chinese policymakers need not prop up individual AI companies to protect retirees’ livelihoods.
The Shanghai Composite’s roughly 29% gain over a decade, compared with the S&P 500’s 254%, illustrates the two systems’ different expectations of stock-market performance.
American retirement security has shifted from employer-provided, defined-benefit pensions toward contribution-based investment accounts, transferring market risk to individual workers.
Retirement accounts represented about 32% of American household wealth in 2022, excluding the wealthiest 1%, making sustained investment returns especially consequential for ordinary households.
The author argues that as stock-market fortunes become increasingly tied to the AI industry, Americans’ retirement prospects become entangled with the successes and failures of AI training runs.

Put your money in the market, let it grow and grow, and at the end of forty-five years, you will have enough to retire. That was the nursery rhyme.

AI’s Competing Economic Models

The author argues that American dependence on stock-market wealth makes policymakers especially sensitive to AI companies’ valuations, whereas China’s household wealth and policy priorities are less tied to AI equities.
DeepSeek exemplifies an alternative commercial model: CEO Liang Wenfeng reportedly intends to keep models open-weight and pursue a “reasonable profit” rather than maximize shareholder returns.
Despite their differences, American AI factions increasingly entertain central planning, including proposals to use the Defense Production Act to allocate computing resources between training and inference.
China’s pursuit of commoditized intelligence can undermine American AI companies’ commercial advantage while imposing relatively little downside on China’s domestic economy.
Chinese models establish visible price-performance benchmarks, forcing closed-weight American providers to continually justify their premiums to customers and investors.

Your retirement plan already relies on Claude, even if you can’t see it, or even know what it is.

The Shrinking Intelligence Aperture

The author argues that cheap Chinese AI could undermine American labs’ commercial funding, pushing their most capable models into a restricted, state-backed frontier inaccessible to the public.
China’s dependence on distillation may diminish as it expands computing capacity by mass-producing less advanced chips rather than matching the most sophisticated semiconductor technology.
Widely adopted Chinese open-weight models could gain a capability advantage through accumulated usage data, especially if continual learning becomes practical.
Chinese models reportedly exceeded half of OpenRouter’s token volume by summer 2026, though the text acknowledges that the platform disproportionately attracts users of capable open-weight models.
Securing American frontier models against theft would not eliminate the financial threat: inexpensive Chinese alternatives could capture global demand and erode the revenues needed to fund further American advances.

An American lab’s commercial customers will not be able to finance the next training run at the sovereign frontier. They couldn’t pay for that model if they tried.

AI’s Sovereign Financing Backstop

The author argues that China can sustain an open AI ecosystem while reserving its most advanced models for strategic use, rather than choosing permanently between openness and secrecy.
Growing domestic chipmaking capacity could strengthen China’s ability to release models that compress competitors’ prices while simultaneously maintaining a restricted sovereign frontier.
Chinese-driven declines in inference prices undermine American AI labs’ expected training returns, exposing the mistake of assuming Chinese labs share their American counterparts’ commercial incentives.
American labs are portrayed as compelled to keep funding training despite deteriorating returns, moving from operating cash flow to borrowing and ultimately to sovereign support.
The author predicts that conventional debt will prove inadequate as open-weight releases impair collateral, while governments could secure effective control through subtle mechanisms such as a carefully structured golden share.

What America does, surely China will do. Wrong then, devastating now, and a mistake likely to be made again.

Nationalization Through AI Funding

Government procurement, infrastructure support, and equity stakes could bring frontier AI labs under state control without formal nationalization, creating governance largely outside public view.
Private lenders and investors would continue financing AI contractors, but government contracts and sustained public spending would become the foundation of their survival.
The author argues that funding frontier training costs growing at 3.5 times annually would intensify Treasury borrowing needs and eventually prompt Washington to suppress interest rates.
Invoking wartime yield controls, the author predicts that national-security priorities could lead the Federal Reserve to cap Treasury yields again, subordinating monetary independence and price stability to AI funding.
In both America and China, labs would willingly accept sovereign authority to remain at the technological frontier, much as strategic weapons programs depend on state backing rather than commercial viability.

A funding wave will break against the walls of the central bank.

AI’s Sovereign Bargain

The author uses the Polaris submarine missile program to argue that government should commission the best-positioned private specialists rather than recreate their expertise in-house.
The proposed model for frontier AI similarly relies on government partnerships with existing private labs, where capital has already concentrated talent.
AI companies seek close state ties because sovereign power can mitigate political, regulatory, financial, and infrastructure risks that private investors cannot.
State involvement would also supply political legitimacy as AI’s economic gains concentrate among relatively few owners, making the arrangement more than a conventional public offering can secure.
The envisioned settlement prioritizes national stability and American advantage over investment returns, requiring public assent in exchange for benefits that would privilege American workers relative to the rest of the world.

The state has no true competitors, and those who forget this, or ignore it, will spend their entire lives building something that could be crushed with a phone call from Washington. Lampreys on the Leviathan.

AI Labs Under Sovereignty

The author predicts that leading American AI labs will become national assets rather than ordinary vendors, aligning state and laboratory interests without formally calling the arrangement nationalization.
America must compete with inexpensive Chinese models as well as at the commercial frontier: widespread adoption could become essential to capability gains once continual learning arrives.
The author uses England’s royal peculiars—religious institutions answering only to the Crown—as a model for AI laboratories operating outside ordinary oversight and directly under sovereign authority.
As AI researchers become crucial to national power, threats of capture, coercion, and assassination could drive increasingly restrictive protections, sacrificing their freedom for national security.
The envisioned laboratories are isolated, tightly secured enclaves whose researchers answer only to the sovereign; China, the author argues, already possesses the party-state authority needed for such an arrangement.

America a fortress, and inside its walls is yet another stronghold occupied by the labs’ sovereign research teams, packed away in their SCIF s. The work is silent and air-gapped.

Sovereignty and AI Diffusion

China governs generative AI through registration and continuous oversight, giving the party-state enduring authority over developers without requiring outright state ownership.
The author portrays China’s party-state as an ever-present sovereign whose concern about instability reinforces its determination to retain control.
Chinese AI firms’ accountability to the state may hinder international adoption and value capture, since global diffusion depends on user trust and widely accepted currencies.
America benefits from deeply established commercial infrastructure—including English, Delaware incorporation, New York finance, and dollar-based private platforms—but its advantage remains politically precarious.
The emerging state absorption of AI laboratories is functional rather than necessarily legal: sovereign control need not resemble the conspicuous property seizures and public decrees of historical nationalization.

A workman would change the sign above the door. Private property was abolished by a very, conspicuously, public decree.

The Neo-Dollar Sovereign

The author argues that modern sovereignty depends less on overt conquest or nationalization than on precise control exercised through ownership, technology, and financial infrastructure.
AI models are predicted to generate enormous demand for a global dollar stablecoin, the “Neo-Dollar,” extending American monetary influence to people outside the U.S. banking system.
Neo-Dollar users would indirectly hold U.S. Treasuries, helping finance the American state, while currency debasement would support government costs and channel owners would collect inflation-linked tolls.
Digital surveillance and intelligent machines promise increasingly predictable, orchestrated societies, restricting ordinary individuals while expanding the power of those at the top.
Despite the state's formidable military and surveillance capabilities, the author presents financial engineering—and especially the Neo-Dollar—as the central instrument and ultimate expression of future American power.

In the past, you had to seize the Commanding Heights with soldiers and rifles. Today you need only own them.

The Neo-Dollar’s Global Reach

Invoking Christ’s instruction to render unto Caesar, the authors frame money as an enduring expression of sovereign authority, even as its physical form changes.
The “Neo-Dollar” is digital ledger money rather than paper currency, backed ultimately, in the authors’ account, by American military power.
The authors predict that access to American AI and other productive capabilities will deepen international dependence on dollars, sustaining the currency’s centrality.
Even Chinese AI services could remain embedded in American financial infrastructure, with foreign users paying in dollars through American intermediaries.
The projected expansion of dollar use abroad would help absorb growing American monetary claims, while domestic purchasing power falls and most goods become more expensive even as AI becomes cheaper.

The image may have changed, but the world is still beneath the yoke of superscription.

AI Wealth and Dollar Power

Cheaper AI, including Chinese offerings, could expand economic activity while American financial infrastructure continues collecting fees, even when sovereign AI revenue ambitions falter.
The author invokes biblical teachings to argue against hoarding money, presenting passive saving as especially vulnerable in the emerging economic order.
Domestic cash savers, pensioners, and investors tied to familiar assets risk losing ground as economic value shifts toward intelligence-driven businesses.
Foreign retail investors face additional constraints: they earn in local currencies while seeking American assets and accessing intelligence priced in an emerging, largely invisible “Neo-Dollar.”
The predicted outcome is sharply unequal wealth accumulation: many investors might earn positive returns, but owners of the economy’s “Commanding Heights” would capture disproportionately greater gains.
The section closes by introducing a contrast between America’s East and West Coasts, beginning with the Eastern view that capital alone offers only limited power.

The agglomeration of digital dollars traveling blindly from all corners of the earth before coming to rest on the books of a Delaware corporation.

Sovereignty Over Wealth

The author argues that wealth ultimately depends on coercive political power, a reality understood by America’s East Coast but often overlooked by its technology-rich West Coast.
China makes the hierarchy explicit: commercial cities generate wealth and innovation, while Beijing determines what businesses may do with their creations.
Jack Ma’s criticism of financial policy, followed by the cancellation of Ant’s IPO and a broader industry downturn, illustrates the party-state’s supremacy over private capital.
The author portrays regulation as sovereign discretion rather than merely written law, suggesting that American technology companies face an increasingly opaque relationship with political authority.
Both countries are described as extracting wealth indirectly: China suppresses deposit returns to subsidize borrowing, while American inflation erodes cash and fixed-payment claims while favoring holders of repriced assets.

Honesty grows out of the lug of a bayonet.

Currency Power, Democratic Constraints

The author portrays currency debasement as a hidden means of financing government, citing America’s postwar reduction of debt through returns below inflation.
The dollar’s reserve-currency status lets Washington shift some costs of debasement abroad, making preservation of that status central to its protection of capital.
American democracy is expected to constrain machine-intelligence development through protections against labor displacement, more generous allowances, and resistance to admitting new citizens.
Public fears about machine intelligence are becoming material barriers to expansion, as local officials can block the data centers and energy infrastructure it requires.
The passage contrasts America’s fragmented, federal system with China’s overriding sovereign authority, arguing that American national ambitions must contend with local opposition.

America’s last scraps of federalism will enable the Loudoun County Board of Supervisors, and many more like it, to stand in the way of machine intelligence reordering the physical world.

Financial Chokepoints and Taxation

Control of physical and financial infrastructure creates chokepoints that allow those controlling access to extract payments and exercise power.
The author frames taxation as a means of taking wealth, distinguishing methods by their targets, mechanisms, and visibility to those affected.
California’s proposed one-time billionaire wealth tax illustrates, in the author’s view, how visible taxation can provoke capital flight when wealth can move across borders.
China’s capital controls are presented as making wealth extraction easier to administer, while less visible methods are described as especially effective.
Roosevelt’s 1933 gold surrender order restricted an escape from the dollar and enabled its subsequent devaluation against gold, which the author interprets as a successful wealth tax.

Physical infrastructure is expensive to build and monopolistic by nature, so whoever controls the switch also controls a tollbooth.

Washington’s Invisible Wealth Tax

Hoover and Roosevelt sharply raised top income-tax rates, while Roosevelt unsuccessfully sought a 100% tax on net income above $25,000, framing income limits as fairness to poorly paid wartime soldiers.
The author presents Roosevelt’s policies as exercises of sovereign power and argues that today’s U.S. government exercises comparable power through a more dispersed, less visible monetary system.
The dollar’s global reserve status allows Washington’s monetary decisions to affect wealth worldwide, including the savings of people using dollar tokens to escape unstable local currencies.
Dollar-token holders receive no interest on their tokens, while issuers collect income on the underlying reserves at market rates shaped by Washington.
Inflation affects businesses unevenly: durable economic channels can increase revenue with nominal activity, while capital-intensive firms must spend rising receipts on increasingly expensive replacement equipment.

This is the secret to Washington’s taxation of wealth; not only is it difficult to protest because it’s difficult to point fingers, it’s invisible to the taxpayer.

Digital Empires, Dollar Rails

The author portrays digital networks as the rails carrying dollars worldwide, enabling infrastructure owners to collect tolls on an expanding share of economic activity.
Digital platforms become nonterritorial empires: their users share no national identity, yet their daily activities enrich owners who can financialize population-scale participation.
The anticipated “Neo-Dollar” will underpin AI-powered apps while continuing to finance physical infrastructure, resource extraction, and global shipping.
China may pioneer new forms of digital adoption, but the author expects even Chinese AI services to accept dollar-backed stablecoins because Chinese capital controls constrain alternative monetary circulation.
American dominance in large-scale AI computing, combined with dollar-denominated payments and balance sheets, is presented as the foundation of continued U.S. influence regardless of which country's models users choose.

Not empires that anyone would die for, but countless lives will be peacefully expended in the furtherance of their designs.

AI’s Unequal Global Future

The author argues that AI-driven trade will reinforce American dollar dominance, concentrating economic power in countries and institutions that control capital and intelligence.
Automation threatens the outsourced office work that helped poorer countries develop, converting lost livelihoods abroad into higher corporate margins and stock valuations in New York.
Ravi’s dismissal illustrates this transfer: by reviewing AI agents’ output, he unknowingly trained his replacement, then received a machine-written termination letter and six months of severance.
Financial markets may reward automation long before official economic indicators register its human costs, driving up productive asset prices while leaving wages behind.
Citizenship and professional licensing may shield workers in wealthy countries, while displaced workers elsewhere face weaker protections and narrowing migration opportunities, deepening global inequality.

Ravi’s job disappeared to the benefit of a margin accrued in a country he had never stepped foot in. Now he never would.

Automation Breaks Development Promises

The text predicts that AI-driven job losses will strike least-protected workers and economies first, particularly clerical and back-office employees serving wealthy countries.
India’s technology sector illustrates the threat to upward mobility: despite employing nearly six million people, it added only about sixteen net jobs for every hundred engineering and technology graduates.
Corporate profits are becoming detached from workforce growth. In fiscal 2026, TCS reportedly cut headcount by 23,460 while adjusted operating profit increased 7.5%, highlighting owners’ incentives to reduce hiring.
The author warns that displaced, educated, digitally connected workers could organize rapidly, destabilizing governments that have little authority over the foreign firms and technologies driving their losses.
Both office employment and emigration—the traditional routes to advancement for educated workers in poorer countries—are presented as collapsing, threatening a reversal of middle-class aspirations.

The people were told that studying hard would result in an office job and the corresponding middle-class life. They’re headed back to subsistence labor in a single generation.

Closing Borders, Surplus Labor

The author argues that tighter immigration restrictions and automation will strand surplus workers in their home countries, frustrating hopes of upward mobility and increasing unrest.
The EU–India trade and migration agreements illustrate a widening gap between elite commitments to international labor mobility and public resistance in destination countries.
The passage predicts that anti-immigration electoral gains across Europe could undermine labor agreements, framing voter opposition as a response to perceived economic and social costs.
Labor-exporting countries such as the Philippines and Nigeria face threats to migration arrangements that have benefited employers and elites while providing workers with routes to better livelihoods.
The author portrays displaced workers as a lasting humanitarian challenge and warns that capital flight and self-serving governing elites could deepen hardship in poorer countries.

Too close for distance to drown out their cries. Children of God, much like yourself.

From Outsourcing to Cybercrime

The author predicts that elites and multinationals will extract remaining profits and leave developing countries as AI displaces outsourced work, abandoning workers and infrastructure.
Former back-office facilities, computers, and technical skills could be repurposed for fraud, making cybercrime a substitute industry where legitimate businesses withdraw.
Southeast Asia illustrates the scale of this possibility: estimated 2023 cyberscam revenues equaled 30.2 percent of Cambodia’s GDP and 68.5 percent of Laos’s GDP, while Myanmar’s reached $15.3 billion.
An imagined Cambodian worker’s transition from accounting support to scamming shows how criminal employment can preserve the routines and physical comforts of legitimate work, with AI handling translation and messaging.
The author portrays criminal cartels as embryonic states and argues that weakly governed border regions allow scam operations to flourish outside stronger states’ legal reach.

He liked sitting in a chair at his old job. Easier on his sore feet.

AI Wealth and Enclosure

Countries will capture AI-driven economic gains unevenly: strong digital infrastructure, English proficiency, and favorable regulation may matter more than hosting a frontier-model laboratory.
Large labor-exporting economies risk losing service-sector jobs before domestic businesses can absorb displaced workers, widening international economic disparities.
New wealth from the intelligence economy is expected to drive sharp price increases for scarce luxury goods, elite education, and exclusive access whose supply cannot readily expand.
The authors predict that AI could shift income from labor to capital within a decade on a scale comparable to the previous forty years, pushing inequality in leading OECD economies beyond that of Brazil and Mexico.
Growing inequality could turn formerly non-market benefits, such as privacy and safety, into costly private services, insulating wealthy owners while local communities bear disruption.

In the near future, privacy might only be available with a hefty subscription fee.

Private Wealth, Invisible Upheaval

Wealthy households are expected to retreat into protected enclaves, buying private security, concierge medicine, and personalized education as public provision weakens.
Membership-based private firms will increasingly assume functions once performed by the state, creating exclusive networks of access, protection, and patronage.
Traditional measures such as GDP will inadequately capture the transformation: asset wealth remains largely invisible until spent or invested, while free AI output lacks a market price.
The author proposes tracking unconventional signals—including offshore staffing cuts, youth unemployment, private-jet backlogs, and premium land prices—to reveal displacement and concentrated prosperity.
As AI accelerates economic activity beyond institutions’ capacity to measure it, gaps between public and private information may become more valuable, rewarding a small number of well-positioned actors.

For signs of the abundance to come, look not to a revival of Main Street and a swelling of the common prosperity but to the Gulfstream backlog and the price of certain plots of land.

Licensing Shields Labor Insiders

Exposure to AI-driven job displacement depends not only on technological capability but also on formal labor protections, licensing rules, and professional influence.
The text argues that protected American professionals may retain jobs even as overseas workers performing similar tasks are replaced, reinforcing an unequal distribution of automation’s costs.
Licensed workers retain value as legally accountable humans who certify automated output, allowing them to remain employed even after AI absorbs much of their work.
These protections disproportionately benefit established, experienced workers while making it harder for younger entrants to gain a foothold in their professions.
A cited Stanford study found employment among 22- to 25-year-olds in AI-exposed occupations was 19 percent below its comparison trajectory by mid-2026, largely because firms hired fewer young workers rather than dismissing existing employees.

The knowledge workers form a labor aristocracy, protected by lobbies and professional-class wings in both parties.

The Commanding Heights

The authors argue that a protective institutional “paper belt” shields many American workers, making upheavals on the scale seen elsewhere less likely in the United States.
They define “Commanding Heights” as scarce, controllable positions that durably capture economic gains, even as intelligence becomes more capable and widely available.
Unlike temporary bottlenecks, these positions grow more entrenched as the economic activity they enable expands.
Lenin’s seizure of power illustrates the importance of controlling strategic infrastructure and government centers, while symbolic acts help establish the public image of a revolution.
Civil war and agricultural disruption forced the Bolsheviks to accommodate private enterprise; Lenin justified this retreat by arguing that the state retained the economy’s vital Commanding Heights.

The near future is a mirror. You’ll see the estate of the sons of men in its fullness.

The New Commanding Heights

Lenin defended the New Economic Policy as a necessary transitional embrace of state capitalism because Russia lacked the productive capacity to sustain socialism alone.
Allowing peasants to trade surpluses and permitting market activity served the overriding goal of preserving Communist Party rule while the state retained control of vital industries.
Lenin’s strategy rested on occupying the economy’s “Commanding Heights”: essential centers of power whose control mattered more than ownership of every economic activity.
The author argues that advancing AI is creating new Commanding Heights—scarce, controllable economic positions that allow their holders to capture lasting gains from increasingly widespread intelligence.
These modern centers of power are larger, more interconnected, more expensive, and more sophisticated than their industrial predecessors; their emerging ownership will reshape economic arrangements as AI spreads.

And in sophistication, they represent power that Lenin could only have dreamed of, or perhaps they would haunt his nightmares.

Scarcity Versus Durable Power

There is still an opportunity to acquire or build the economy’s future Commanding Heights before their strategic value becomes widely recognized; their owners could gain growing cash flows as AI expands economic activity.
Machine intelligence will sharply reduce labor costs, enabling new businesses and economic arrangements while forcing established bargains to be renegotiated.
Personal AI agents could automatically move household savings toward higher yields, undermining banks’ valuable deposit franchises and increasing their funding costs.
Scarce inputs can capture the profits of otherwise well-equipped businesses: a completed data center without grid access may have to surrender its margin to secure electricity.
Temporary scarcity does not establish enduring economic dominance. Margins can migrate upstream or disappear as bottlenecks are resolved, whereas true Commanding Heights depend on durable control.

The value flows through and past the data center, pooling in the scarce resource that powers it.

AI’s Shifting Economic Bottlenecks

AI infrastructure bottlenecks, or “crossings,” temporarily grant owners pricing power, but competing routes and new investment continually shift profits elsewhere.
The state can claim a share of AI-generated wealth through its sovereign authority, potentially sustaining frontier development even when it is financially unsustainable on its own.
Machine intelligence threatens to erase the market positions of many established businesses, shifting their value from future earnings toward proprietary data, installed customer bases, or usefulness as training environments.
The author urges owners of businesses whose main remaining asset is data to consider selling, arguing that some datasets may yield only a one-time payout rather than enduring revenue.
Even strategically positioned bottleneck owners face political limits: regulation, local opposition, and antitrust intervention can constrain their profits and undermine otherwise durable advantages.

Most human enterprise will be absorbed or indifferently crushed underfoot. It’s a fire sale, and you don’t want to be the last to sell.

AI’s Commanding Heights

Investors seek control of enduring economic chokepoints to escape recurring cycles of scarcity, high margins, and decline.
The author argues that the strategic industries enabling political control and economic value capture are essentially the same; owners’ intentions distinguish their uses.
Xi Jinping’s emphasis on emerging productive forces reflects a confident party-state whose authority over strategic industries is unquestioned, a stance the author predicts America will eventually adopt.
Machine intelligence is expected to reshape firms into small groups of operators, compensated and governed like capital, directing work performed largely by models.
As technology becomes central to the economy, AI’s value is expected to migrate beyond its originating labs, while China and America dominate further model development and diffusion.
The author anticipates stronger sovereign oversight of AI labs, alongside the spread of a 'Neo-Dollar' that extends financial inscription across America’s digital sphere.

The smartest money is trying to escape that endless cycle of birth, woe and suffering, and the rapid black punctuation of death.

Intelligence and Corporate Control

The authors predict that machine intelligence will concentrate economic gains among distribution and network owners, widening inequalities between countries, firms, and individuals while undermining poorer nations’ development prospects.
This upheaval may create opportunities to acquire new economic “Commanding Heights”: scarce, controllable positions that durably capture value from increasingly widespread intelligence.
Guinness illustrates how firms reorganize ownership, capital, labor, and control as economic conditions change, beginning as an owner-funded Dublin brewery in 1759.
Its 1886 stock-market flotation separated capital from control; by the 1986 Distillers takeover, professional managers exercised decision-making power, with the ensuing share scandal exposing the dangers of managerial misconduct.
Guinness’s 1997 merger into Diageo completed its transformation from an independent brewery into a brand within a conglomerate, demonstrating how a business can persist while its organizational structure changes radically.

The Guinness firm had gone from a brewery to a joint-stock company and reached its terminus as a brand in a conglomerate.

Technology Reshapes the Firm

A firm is a changing arrangement of capital, decision-making, labor, and obligations, rather than a permanently fixed institution.
Ronald Coase explained that firms exist because using markets incurs transaction costs; work that is difficult to describe or measure can be cheaper to coordinate internally.
Technological change alters transaction costs and enables new organizational forms, reshaping how capital, control, and labor fit together.
The author identifies three historical settlements since the eighteenth century and argues that the emerging era of intelligence is creating a fourth.
The first settlement centered on owner-operated firms, where supplying capital also conferred the right to manage; increasing scale and complexity eventually overwhelmed this model.
By the 1920s, ownership and management had separated: shares dispersed ownership among millions, while managers with little ownership exercised operational control.

But each settlement, by its arrangement, contains in itself evidence of what it lacks.

Capital Disciplines the Corporation

From the 1930s through the 1970s, widely owned corporations were controlled by career managers who functioned as privileged employees and felt obligated to provide generous worker benefits.
The text portrays this managerial settlement as insufficiently ruthless: managers’ solidarity with workers and the erosion of corporate cash by 1970s inflation helped provoke a revolt by capital.
Jensen and Meckling’s agency theory supplied a rationale for aligning managers with owners through shareholder oversight, stock-based compensation, and bonuses.
Under the resulting hierarchy, owners disciplined managers, who shifted labor off corporate balance sheets to outside vendors, generating strong financial returns amid broader stagnation.
The third settlement benefited from a decades-long decline in borrowing costs, which the author argues concealed sluggishness and encouraged profitability over innovation.

They sang for new managers, who would swear fealty to those above, rather than below.

Labor Leaves the Firm

The author argues that a new economic “fourth settlement” began around 2023, as falling interest rates stopped providing a tailwind and transformer-based AI became central to markets.
Technological expertise is again becoming a source of profit margins, favoring operators who control capital rather than a separate professional management class.
This emerging order resembles an earlier era of powerful owner-operators, but with AI models performing much of the work previously done by disadvantaged human labor.
Making work legible—standardized, measurable, and easily transacted—reduces the costs of outsourcing, allowing firms to shrink their headcounts while expanding their capabilities.
The author presents offshoring as a precursor to AI substitution: once human services could be specified and purchased by the task, they became easier to transfer to models, leaving less room for human workers and intermediary managers.

The fourth settlement is one sympathetic to great men and leaves little room for labor as you know it.

When Talent Becomes Capital

The authors argue that corporate payrolls function as political settlements: a few indispensable workers generate disproportionate value, while broadly distributed salaries conceal this imbalance and preserve internal stability.
They predict that AI will reshape rather than eliminate these arrangements, raising the stakes of labor allocation from corporate politics to national concerns.
Remaining labor is expected to divide into three groups: licensed professionals collecting fees for their signatures, a larger salaried class facing declining compensation value, and exceptional operators compensated more like capital.
Public companies struggle to reward and retain exceptional producers because internal pay politics and standardized bureaucratic requirements conflict with those workers’ value and autonomy.
Talent-focused acquisitions offer a workaround by effectively purchasing individuals as businesses; the authors interpret Meta’s 2025 purchase of half of Scale AI as a $15 billion bid to recruit Alexandr Wang.

The kind of person who could make the firm a fortune will not idly sit through a managerial review, or even worse, the online annual training program.

Talent Commands Corporate Power

AI companies are paying extraordinary sums to secure scarce talent, with reported individual compensation packages reaching hundreds of millions of dollars over multiple years.
Large technology firms increasingly acquire access to teams through hiring and technology-licensing deals rather than buying entire companies, blurring the boundary between recruitment and acquisition.
The text argues that exceptional talent, when valued like capital, can demand governance rights as well as financial compensation, challenging conventional employment arrangements.
Anthropic illustrates the separation of economic ownership from control: its seven co-founders reportedly sought 50.1 percent of votes on key matters while each owning roughly two percent of the company.
These arrangements suggest that command over a company—not simply equity ownership—is becoming the decisive measure of power for elite technical founders.

Command is what matters. Talent can make such demands, which were once to be reserved for capital, and capital alone, in the old worlds of the second and third settlements.

Firms Become Market Machines

Trading firms exemplify the author's emerging “fourth settlement”: small, highly productive partnerships that distribute profits among select participants without answering to outside shareholders.
Applying the good regulator theorem to markets, the author argues that successful firms increasingly model and resemble the markets they operate in, driving the “hedge-fundification” of industries.
AI labs may acquire and operate entire businesses to obtain operational data and create reinforcement-learning environments for complex, long-horizon tasks.
Taken to its extreme, this expansion could turn AI labs into sprawling conglomerates whose diverse businesses serve a shared training purpose that outsiders struggle to understand.
The author predicts that greater transparency will transform venture capital from an access-driven business into something closer to trading, forcing capital allocators to adapt.

A shiny smile and a poorly tied tie with a term sheet in hand. The new transparency, characteristic of the fourth settlement, will bend it into a trading business.

Venture Capital’s Final Settlement

The author argues that recent venture returns rewarded access to powerful networks and large checks more than original insight, but expects AI-driven transparency to turn investing increasingly into trading.
Institutional seed investing is predicted to fade as operators take on capital-allocation roles and private markets become more liquid and financialized.
Venture is becoming sharply divided: ordinary assets offer direct equity access, while prestigious investments attract buyers through complex SPVs whose fees and indirect exposure can obscure actual ownership.
Lasting venture value will concentrate in elite networks that do more than select promising founders: they actively improve their companies’ chances of durable success.
As AI businesses require enormous capital commitments, fewer providers will dominate, and major investment platforms will shift toward holding companies, buyouts, and operating businesses.
Public equities already challenge venture’s claims to superior opportunity: the author cites Leopold Aschenbrenner’s fund outperforming venture through publicly available stocks while many 2021 venture portfolios remain near cost.

Intelligence creates transparency, and transparency will always turn investing into trading. Too many are ill-suited to the times.

AI Reshapes the Market

The authors argue that the forty-year bond bull market is ending, removing the falling interest rates that previously helped financial engineering compensate for weak scientific and civilizational progress.
They interpret Treasury yields rising above their historical average as a signal of a potentially decades-long monetary regime change and persistently higher capital costs.
Machine intelligence offers a lifeline to private equity and the broader economy, but its benefits require buying and rebuilding businesses around AI models rather than sustaining low-value software projects.
Markets are pricing AI transformation through successive investment themes rather than recognizing its economy-wide implications, creating what the authors see as pervasive mispricing.
The authors anticipate fewer public companies and a wave of acquisitions, roll-ups, and take-private deals, requiring new private-equity structures to spread machine intelligence.

For a narrow set of companies, the market has decided there is no price too high—and for the rest, no price low enough.

Jurisdiction as a Product

The author argues that AI can already transform businesses, but institutional inertia creates an opening for private equity to acquire companies, modernize operations, and restructure workforces.
As intelligence increases transparency, firms increasingly shop for legal jurisdictions, forcing states to compete for mobile capital through attractive laws and institutions.
The text envisions a world dominated by American and Chinese technology stacks, with other countries retaining nominal independence while depending on one of these powers for essential capabilities.
This diminished sovereignty could empower forceful local rulers who combine civil and military authority, producing upheaval and more personalized forms of government.
Corporate jurisdiction-shopping will increasingly center on freedom to reward exceptional contributors; the author presents disputes over executive compensation and acquisition-based payouts as incentives to create more accommodating legal venues.
The frontier being advanced by AI laboratories is therefore presented as jurisdictional as well as technological: greater intelligence could reshape both companies and the legal orders governing them.

There is only one Haiti, but there will be a multitude of Princes.

Capital Meets Sovereign Power

The author argues that AI firms gain an advantage by choosing favorable jurisdictions and organizational structures, making legal mobility a central competitive asset.
As mobile firms escape taxation, governments may shift the burden onto less-mobile salaried workers, including indirectly through American-controlled currency systems.
Cities may demand equity or other concessions from technology firms, adding local pressure to an increasingly fragmented struggle over jurisdiction.
The text envisions firms partnering with accommodating jurisdictions that sell access to stable legal regimes, while AI-driven arbitration and insurance underwriting replace some courts and conventional regulation.
The distinction between corporate capital and sovereign power is narrowing: the author treats post–Cold War business autonomy as a temporary exception sustained by states’ sense of security, not a lasting norm.

Something strange happens at the vanishing point, where from a sufficient distance the lines between capital and sovereign power become indistinct.

The Price of Protection

The text frames financial crisis management as selective sacrifice: regulators preserve the system while powerful firms surrender weaker participants, often strengthening the survivors.
Exceptional success can turn a frontier firm or financier into a public target; Michael Milken’s imprisonment illustrates how punishment can coexist with the wider adoption of the punished figure’s innovations.
Industry alliances help firms survive political scrutiny by sacrificing vulnerable members to protect the core, a recurring pattern across finance, defense, and pharmaceuticals.
AI labs seek federal protection through preemption of state laws, favorable permitting, and copyright rules while facing Chinese competition; the author warns that this protection will demand payment beyond taxes.
The Atomic Energy Commission’s takeover of Manhattan Project facilities offers a precedent for explicit sovereign control, which the author prefers to opaque intervention and views as more publicly accountable in America than in China.

The sovereign will afford protection to its peculiar labs. But protection always has a price, and you are far too small-minded if you believe that it will be paid in taxes.

Sovereignty Meets Frontier AI

Washington’s ideological factions favor different interventions in frontier AI, ranging from compulsory compute allocation to taxpayer equity ownership, informal regulatory agreements, and aggressive taxation.
The authors predict licensing for frontier training runs and some form of nationalization, potentially limited to a government-held golden share.
Their envisioned settlement splits an AI lab into two operations under sovereign oversight: a commercial intelligence business constrained by Chinese competition and a cost-plus frontier research lab directly connected to the Oval Office.
The central unresolved question is whether AI firms remain subjects of an ultimately authoritative state or become partners exercising portions of sovereign power.
An expansion of state capacity housed within private firms could produce a Palantir-style future, concentrating public functions, capital, and managerial authority in a small group of corporate leaders.

The antichrist is in the details.

Investing Beyond Yesterday’s Heights

The author argues that both owning yesterday’s dominant businesses and funding today’s inventions can miss the durable sources of value in the emerging intelligence economy.
The next “Commanding Heights” will be businesses that collect unavoidable, enduring tolls on an expanding economy, rather than necessarily the companies creating its foundational inventions.
Investors must reconstruct portfolios and revise managers’ mandates to identify these future gatekeepers before their advantages become obvious and fully priced.
The Pennsylvania Railroad’s fall from global corporate leadership to bankruptcy illustrates how economic transitions can undermine seemingly impregnable businesses.
Buffett’s newspaper investments show that even exceptional investors can mistake an era’s profitable dominance for permanence: later purchases totaling $344 million ended in a $140 million sale in 2020.

It is finally time again to buy DoubleClick and Android, not Alphabet or Anthropic.

Reinventing the Commanding Heights

Even exceptional investors can remain anchored to yesterday’s dominant businesses, as Buffett’s newspaper investments illustrate.
Google inherited newspapers’ advertising tolls, but AI agents now threaten to reroute search traffic and advertising revenue, shortening the lifespan of its dominance.
Diversification can finance reinvention: Hearst developed a major profit source in Fitch Ratings, while Alphabet’s other businesses give it room to disrupt its own search franchise.
Investors risk penalizing companies that undertake necessary reinvention while praising apparently steady businesses whose competitive positions are eroding.
The author argues that AI will divide asset owners into winners and losers: established holdings face destruction or inflationary erosion, while many fashionable startups have not yet built durable advantages.
The intelligence era’s dominant toll-collecting businesses remain to be established; rare opportunities comparable to buying Android may emerge, but chasing each new launch is no substitute for identifying them.

The smartest companies will disembowel themselves in reinvention; the rest will wither and file. You will de-rate the smartest for taking an uncertain swing, and laud the rest for being steady compounders.

Toward the AI City

The authors argue that technological eras create increasingly powerful dominant businesses, making established market leaders dangerous investments when a new technological order emerges.
Conventional investment vehicles—including indices, venture capital, and buyout funds—are portrayed as structurally ill-equipped to capture the full opportunities of the AI transition.
Cheap access to machine intelligence could accelerate individual learning and achievement, but its consequences depend on whether people merely preserve familiar routines or develop new ways of living.
The authors warn that a small minority who recognize and exploit AI’s possibilities could acquire extraordinary power over a broadly comfortable but less active majority.
The closing passage depicts technological history as an irreversible, accelerating journey toward an unknowable destination, invoking nuclear weapons to frame machine intelligence as another potentially world-transforming creation.
Rather than treating radical disruption as implausible, readers are urged to recognize that their familiar world has already ended and prepare for what follows.

The few who see and act will become a new shape of their own, and we will be their subjects.