The Algorithm of Two Empires

A classic study analyzing the 'hub and spoke' structure of Carthaginian trade networks. It shows how Phoenician colonies in the western Mediterranean built commercial hegemony, suggesting structural parallels with modern technology platform network effects.

A landmark work tracing technological change and industrial development in Western Europe. It introduces the concept of the 'penalty of the pioneer,' explaining the paradoxical mechanism by which technological leaders can be overtaken by latecomers.

A classic theorizing the paradoxical advantages of late industrialization. Its argument that economic backwardness can serve as a driver for rapid industrialization provides a key framework for understanding China's catch-up strategy in the US-China tech competition.

The foundational text for infant industry protection theory. It systematized the logic that late-developing nations must protect strategic industries through state intervention to survive free trade with advanced nations, serving as the intellectual origin of today's semiconductor and AI industrial policies.

A work defining the concept of 'finance capital' — the fusion of banking and industry. It analyzes how financial institutions drove industrial investment during German industrialization, providing a historical precedent for modern Big Tech capital concentration and finance-technology convergence.

A grand strategic study analyzing the time lag between economic shifts and military power. Its core thesis that great power rise and fall is determined by relative changes in economic strength provides a framework for examining the long-term trajectory of US-China hegemonic competition.

A comparative analysis of British 'personal capitalism' and German 'cooperative capitalism.' It demonstrates diverse pathways of industrial capitalism, useful for understanding the historical roots of the two models: US market-driven innovation and Chinese state-led innovation.

A study analyzing the gap between Britain's education system and industrial innovation. It traces how the absence of technical education accelerated Britain's industrial decline, historically underscoring the importance of human capital investment in the AI era.

A work establishing the concept of the 'entrepreneurial state' centered on the DARPA model. It demonstrates that government investment was behind breakthrough technologies like the internet, GPS, and touchscreens, challenging conventional wisdom about the state's role in innovation.

A study tracing how the Soviet Union's national computer network project (OGAS) failed due to bureaucratic resistance. It demonstrates the gap between technological possibility and institutional reality, proving that innovation success depends on institutions, not technology.

An analysis of why Russia's brilliant inventions fail to translate into commercial success. Its core argument — 'it's not the quality of ideas but the institutional pathways of diffusion that matter' — illuminates the institutional conditions of innovation ecosystems.

A systematic three-volume analysis of the economic, social, and cultural transformations of the information age. It presents the US-Soviet computer gap of 1985 (30 million vs. 50,000 units), demonstrating the overwhelming disparity created by institutional differences in technology diffusion.

An empirical analysis of the approximately 80,000 km Roman road network. It shows the mechanism by which physical infrastructure contributed to imperial integration and market expansion, suggesting structural parallels with modern digital infrastructure (data centers, communications networks).

An analysis of the economic mechanisms by which Roman roads contributed to single market formation. It suggests that the pattern of infrastructure investment driving market integration applies to modern AI infrastructure investment competition.

Key data quantifying industrialization levels across nations from 1750 to 1980. It shows Britain's share of world manufacturing plummeting from 31.8% in 1870 to 13.6% in 1913 while Germany caught up, providing quantitative evidence for hegemonic transition.

A comprehensive historical statistics collection of national steel production. The figures — Germany's 17.6 million tons vs. Britain's 7.7 million tons in 1913 — pinpoint the exact moment of industrial production reversal.

A monumental dataset estimating the world economy over a millennium in 1990 international dollars. It quantitatively confirms the moment when Germany's GDP surpassed Britain's in 1913.

A comparative study of European education systems. The gap between Germany's 77,000 university students and Britain's 26,000 in 1913 illustrates differences in human capital investment, providing a historical precedent for the AI talent competition.

A report estimating the annual economic value of GPS at approximately $1.4 trillion. It quantifies the scale of economic value created by military-to-civilian technology transfer, demonstrating the returns of the DARPA model.

An ancient source recording Rome's ability to mobilize approximately 700,000 troops in 225 BCE. It demonstrates the overwhelming manpower mobilization capacity of the Roman alliance system, illustrating how institutional advantage translates into military superiority.

A source recording Carthage's annual revenue of approximately 12,000 talents from Iberian mines. It reveals both the scale of the Carthaginian economy and the limitations of a resource-extraction growth model.

A technical document recording Rome's 11 aqueducts spanning approximately 500 km. It demonstrates that large-scale infrastructure construction and maintenance capabilities reflect imperial technological prowess.

A study analyzing the 'distributed pool' structure of the Roman alliance system. It reveals how decentralized manpower pools — rather than single centralized control — enabled Rome's military resilience, providing a historical precedent for distributed innovation ecosystems.

A study introducing the concept of Rome's 'systemic resilience.' It analyzes Rome's institutional robustness that prevented collapse despite repeated defeats, suggesting that institutional adaptability may be more important than technological capability.

A study analyzing the prototype of 'national innovation systems' through the German chemical industry. It demonstrates that organic linkages between universities, industry, and government were key to industrial competitiveness, providing a historical model for modern AI innovation ecosystems.

A seminal paper theorizing the Triple Helix model of university-industry-government interaction. Its framework — that innovation arises from the interplay of three actors — provides the theoretical foundation for analyzing the differences between Silicon Valley (US) and state-led innovation (China).

A classic analyzing Silicon Valley's 'region-based industrial system.' It identifies the mechanisms of the innovation cluster created by open labor markets, venture capital, and informal information exchange, serving as the academic foundation for the Chapter 3 analysis of the Silicon Valley ecosystem.

A study analyzing Soviet computer diffusion data and the IBM-compatible clone strategy. It reveals the paradox where technology catch-up strategies actually weakened indigenous innovation capabilities, providing a historical warning for China's semiconductor self-reliance strategy.

A study analyzing the 'single-objective optimization' mechanism of the Soviet economy. It explains how concentrating resources on military and space programs sacrificed civilian innovation, illuminating the structural limitations of state-led innovation.

A study analyzing the institutional causes of innovation failure to diffuse in Soviet industry. It identifies the structural problem where invention was possible but diffusion was not, showing that innovation success depends on diffusion pathways.

A study analyzing how 70-75% of Soviet R&D spending was concentrated in the military sector. It demonstrates that the failure of military-civilian technology transfer was a fatal weakness of the Soviet innovation system, emphasizing the importance of institutional design through comparison with the DARPA model.

An annual report providing data showing the US accounts for approximately 40% of the top 1% most-cited AI papers. It quantifies the qualitative lead in AI research and serves as a key indicator of US AI foundational research competitiveness.

Talent tracking data showing approximately 60% of the world's top AI researchers are affiliated with US institutions. It reveals the geographic distribution of human capital, a critical variable in AI competition.

Tracking data showing the US-China AI model performance gap narrowing from an average of 7 months to 3-6 months as of 2025. It quantitatively demonstrates China's rapid catch-up speed and serves as a key indicator for gauging the dynamic changes in the technology gap.

An analysis estimating the Chinese government's AI capital expenditure at approximately 400 billion yuan. It demonstrates the scale of state-led AI investment and contrasts with the US model of private-sector-driven investment.

An analysis identifying an approximately $500 billion gap between AI infrastructure investment and application revenue. It directly raises the monetization challenge of AI investment and warns of the over-investment pattern typical of early technology revolutions.

Earnings data showing NVIDIA's data center revenue reaching $51.2 billion per quarter (FY2026 Q3). It confirms the explosive growth in AI infrastructure demand and demonstrates that GPUs are the key means of production in the AI era.

Official labor statistics showing the median annual salary of US paralegals at $61,010. It provides baseline data supporting the economic reality of the American character in the Chapter 14 comparative pair (paralegal).

A report analyzing that 69% of paralegal tasks are automatable by AI. It reveals the specific pathway by which AI replaces mid-skilled positions in legal services, empirically demonstrating the formation mechanism of 'the displaced.'

An IMF analysis showing that the negative wealth effect from real estate is structurally suppressing Chinese household consumption. It demonstrates that China's weak domestic demand is a structural rather than temporary problem, explaining the disconnect between AI technology investment and the real economy.

An analysis showing that the real estate downturn is dragging China's GDP growth by approximately 2 percentage points annually. It quantifies the scale of China's structural economic headwinds and raises the question of whether technological innovation can offset the broader economic downturn.

Disclosure data showing Alibaba's full-time employees dropping 51.2% from 254,941 (2022) to 124,320 (2025). It reveals that large-scale restructuring at Chinese Big Tech is proceeding under the dual pressures of AI automation and regulatory tightening.

Disclosure data showing Baidu's workforce declining 21.1% from its 2021 peak to 35,900 by the end of 2024. It demonstrates that even leading Chinese AI companies are not immune to efficiency pressures.

Data showing Meituan has over 10 million delivery riders with high-frequency riders earning 6,650-9,344 yuan monthly. It quantifies the structural shift where a massive gig worker layer replaces traditional employment in China's platform economy.

A statement by Microsoft's AI chief predicting 'automation of all white-collar work within 18 months.' It illustrates both the psychological shock of extreme predictions from AI industry leaders on the labor market and the gap between such forecasts and reality.

A statement by Google DeepMind's CEO assessing the US-China AI gap at 'a few months.' It suggests the actual size of the technology gap from the perspective of a frontier AI researcher, showing that US technological superiority is not absolute.

A study empirically demonstrating the 'Engels' Pause' phenomenon — during 1780-1840 in Britain, labor productivity rose 46% while real wages increased only 12%. It historically proves the time lag before technology revolution benefits reach workers, providing a precedent for AI-era wage stagnation.

An OpenAI study showing approximately 80% of the US workforce is exposed to LLMs in at least 10% of their tasks. It was the first systematic estimate of AI's labor market impact scope and is a key study for gauging the potential scale of 'the displaced.'

A historic document proposing the model of government research funding flowing through university basic research to civilian transfer. It served as the blueprint for US science and technology policy and is the institutional origin of the post-war American research investment system and the Silicon Valley innovation ecosystem.

NSF data showing approximately 70% of US AI/CS doctoral students are foreign-born. It quantifies how heavily the US AI talent pipeline depends on immigration, suggesting the impact of immigration policy changes on AI competitiveness.

Survey results showing immigrants account for 55% of Silicon Valley startup founders. It empirically demonstrates that immigrants are a core driver of the US innovation ecosystem and that Silicon Valley's competitiveness is built on openness.

Data showing $211 billion in global AI VC investment in 2025, with $127 billion concentrated in the Bay Area and $113 billion in 92 mega-rounds. It demonstrates the extreme geographic and scale concentration of AI investment and quantifies Silicon Valley's capital hub function.

Official announcement of Anthropic's Series G at $30 billion, $380 billion valuation, and $14 billion ARR. It shows that AI startup mega-fundraising has reached scales rivaling Big Tech.

Earnings data showing NVIDIA's quarterly total revenue of $57 billion and 12-month cumulative $187.1 billion. It allows gauging NVIDIA's status as the primary beneficiary of AI infrastructure demand and the growth scale of the entire AI industry.

Data showing DARPA's annual budget of approximately $4 billion and the US Department of Defense's official AI budget of $1.7 billion. It reveals the scale of government-led innovation investment relative to private investment and demonstrates the linkage between national security and AI technology.

Data showing Stanford HAI's annual budget of approximately $100 million and faculty of over 200. It demonstrates the scale of university-based AI research and the density of academia-industry collaboration, representing the academic pillar of the Silicon Valley ecosystem.

Benchmark data showing 9 of the top 10 global open-weight AI models are Chinese-made. It reveals the quantitative leap and open-source strategy achievements of Chinese AI models, calling for a reassessment of US technological superiority.

A BVP report showing AI coding tool Cursor surpassing $1 billion ARR. It demonstrates the rapid commercial success of AI-native software and serves as a concrete example of 'the discerning' leveraging AI as a force multiplier.

Media coverage analyzing the latest trends and investment flows in the AI industry. It captures real-time changes in the Silicon Valley ecosystem discussed in Chapter 3.

Fortune coverage of AI industry and corporate trends. It captures changes in Silicon Valley's investment environment and corporate strategies from a mainstream business media perspective.

Coverage from a media outlet specializing in non-Western technology ecosystems. It illuminates the diverse patterns of global AI diffusion beyond the Silicon Valley-centered narrative.

CNBC coverage of AI industry investment and corporate earnings. It provides real-time analysis of Big Tech AI investment scales and monetization strategies.

Coverage from a Silicon Valley regional media outlet on the AI industry. It conveys the direct impact of technological innovation on the local economy and community from a local perspective.

Coverage from the Bulletin of the Atomic Scientists analyzing the security implications of AI technology. It professionally addresses the darker aspects of technology, including the military applications of AI and potential nuclear risk linkages.

The landmark paper proposing the Transformer architecture. Published by 8 Google Brain researchers, this paper became the technological foundation of the modern AI revolution and is a prime example of an innovation born in the Silicon Valley ecosystem that has transformed the entire world.

Earnings showing NVIDIA's annual revenue of $215.9 billion (+65% YoY), Blackwell cumulative approximately $180 billion, and order backlog of $320 billion. It numerically proves that GPUs are the core infrastructure and physical foundation of technological hegemony in the AI era.

Industry data showing NVIDIA's GPU market share at 92% (2025 H1). It reveals how a single company's overwhelming market dominance becomes a core leverage point for technological hegemony and supply chain control.

Patent and market analysis showing NVIDIA's AI chip market share at 90%. It confirms the monopolistic structure of the GPU market as a core strategic asset for the US in the US-China tech competition.

A visualization showing NVIDIA's AI data center revenue share at 86%. It demonstrates how NVIDIA's dominance extends into the data center domain of the AI infrastructure market.

Coverage showing Big Tech's combined 2026 Capex of $635-665 billion (Amazon ~$200B, Alphabet $175-185B, Microsoft ~$145B, Meta $115-135B). It reveals the astronomical scale of AI infrastructure investment and empirically demonstrates the convergence of financial and technological hegemony.

IMF data showing the dollar's share of global foreign exchange reserves at 56.32% (2025 Q2) and the yuan at approximately 2%. It quantifies the overwhelming gap between the dollar and yuan in financial hegemony, showing one axis of America's dual hegemony structure (financial + technological).

St. Louis Fed analysis showing that while the dollar's share fell from its 2001 peak of 72%, the real decline is only 0.12 percentage points when exchange rate effects are removed. It provides an important counter-argument to dollar hegemony decline narratives.

IEA projections showing data center power consumption reaching 800 TWh to 1,000+ TWh in 2026, increasing 160% by 2030. It reveals that AI infrastructure expansion's energy constraints are emerging as a new strategic variable.

FERC projections showing US data center power demand growing from 19 GW in 2023 to 35 GW by 2030. It demonstrates that energy infrastructure is emerging as a new bottleneck for AI competitiveness.

SemiAnalysis data showing Huawei Ascend 910C yield at approximately 40% and DeepSeek's total server Capex at approximately $1.6 billion. It reveals that China's indigenous chip development faces yield issues, demonstrating the practical difficulty of technological self-reliance.

Coverage of Huawei Ascend's 2026 production target of 1.6 million dies. It demonstrates China's strategic commitment to scaling indigenous AI chip production despite US export controls.

Analysis showing DeepSeek V3's GPU training cost at approximately $6 million. This contrasts with Western AI companies' hundreds-of-millions training costs, demonstrating that an 'efficient catch-up' strategic pathway is feasible.

Barclays analysis showing Meta's free cash flow (FCF) could decline 90% if current Capex levels are sustained. It quantifies the pressure AI investment competition places on Big Tech's financial health.

IEEE Communications Society analysis showing approximately 75% of AI-related Capex is directly invested in infrastructure. It reveals that the majority of AI investment is concentrated in physical infrastructure such as GPUs and data centers.

Data showing China added approximately 430 GW of net power capacity in 2024, 14 times the US (~30 GW). It demonstrates China's physical advantage in the energy foundation for AI infrastructure and illuminates the energy dimension of US-China competition.

Coverage of NVIDIA H100/H200 chip smuggling seizures exceeding $160 million (Oct 2024-May 2025). It reveals the enforcement reality of export controls and the scale of the black market, exposing the limitations of technology control policies.

Historical record of Britain's Machinery Export Ban enacted in 1774 and repealed in 1843. It provides a historical precedent of a technology hegemon's export controls ultimately failing long-term, suggesting implications for modern semiconductor export controls.

The case of Samuel Slater transferring British textile technology to America in 1789 to build the first water-powered spinning mill. As a historical precedent for technology leakage, it demonstrates the pattern of core technology transfer despite export controls.

A classic exposing the miserable conditions of the working class during the Industrial Revolution. It documented the prototypical paradox where technology revolutions increase productivity while deteriorating workers' lives, historically illuminating the conditions of AI-era 'displaced.'

Data showing manufacturing employment in 5 Rust Belt states declining 35% since 2000 (steel -60%, auto assembly -40%). It quantifies the specific scale of deindustrialization and provides background for the additional shock AI may bring to these regions.

MIT/Boston University joint research projecting AI and robots will eliminate 2 million manufacturing jobs by 2026. It forecasts the scale of additional job losses from AI automation in a manufacturing sector already hit hard.

MIT research showing only 12% of workers displaced by automation successfully transition to better jobs. It quantifies the structural difficulty of recovery for 'the displaced' and demonstrates the limits of labor market mobility.

Harvard Business Review reporting 55,000 direct AI-caused layoffs in 2025. It demonstrates that AI's direct impact on the labor market is no longer hypothetical but reality.

CNN reporting 32,000 tech company layoffs in January-February 2026. It demonstrates that restructuring is accelerating even within the tech industry in the AI era, suggesting the boundary between 'the discerning' and 'the displaced' is forming within tech itself.

Harvard analysis projecting 15-25% AI job destruction with net losses of 5-10%. It distinguishes between gross job changes and net losses, providing a balanced outlook that counters both excessive fear and excessive optimism.

World Economic Forum projection that AI and robots will displace 92 million jobs while creating 170 million by 2030. While net job creation is expected, the time lag and skills mismatch between displacement and creation remain the core challenge.

Data showing the national average COBRA premium at $584/month. It concretizes the economic burden displaced American workers face in maintaining health insurance, demonstrating the mechanism by which the absence of a social safety net amplifies the pain of displacement.

Dallas Fed analysis showing AI hits high-skill cities, unlike past automation that hit the Rust Belt. It reveals that the geographic distribution of AI displacement differs from previous automation waves, showing that even highly educated regions are not safe.

Oxfam data showing global billionaire wealth at $18.3 trillion (+16% YoY) and the wealth growth gap between the top 1% and bottom 50% at 2,655x. It quantifies the extreme capital concentration of the AI era and demonstrates the structure by which technology revolutions deepen inequality.

Goldman Sachs projection that AI will begin contributing to US GDP growth from 2027. It reveals the time delay before macroeconomic-level AI effects materialize, suggesting the time gap between individual suffering and aggregate growth.

McKinsey projection that AI will create $2.6-4.4 trillion in additional value annually by 2040. While it demonstrates the scale of long-term value creation, it simultaneously raises the question of how those gains will be distributed.

Fortune coverage of employment decline data by industry in the Rust Belt. It details the specific industry-level patterns of manufacturing decline and the impact on local economies.

Coverage showing NAFTA and US-China trade offshoring eliminated 1.2 million jobs. It demonstrates the cumulative impact of the dual shocks of globalization and technological change on the US labor market.

Ford CEO's statement that 'AI will replace half of white-collar workers.' As a manufacturing CEO's forecast on AI's white-collar job destruction, it illustrates the trend of displacement expanding from blue-collar to white-collar.

Salesforce CEO's statement that '50% of work is handled by AI' and 2025 engineering hiring freeze. It demonstrates a concrete case where AI adoption leads to actual employment decisions, revealing the paradox of 'discerning' companies accelerating displacement.

A key study empirically demonstrating that the surge in Chinese imports in the 2000s eliminated over 2 million US manufacturing jobs — the 'China Shock.' It shows that trade shocks' labor market impacts were far larger and more persistent than economists predicted, providing a comparative benchmark for gauging the potential scale of an 'AI Shock.'

BVP data showing Cursor's ARR surpassing $1 billion, used by more than half of Fortune 500 companies, with 35% of PRs generated by AI agents. It demonstrates that AI tools are revolutionizing software development itself, serving as a concrete example of 'the discerning' leveraging AI.

Official data showing Anthropic's ARR rapidly growing from $1 billion (end of 2024) to $14 billion (Feb 2026) with a $26 billion target. It demonstrates the explosive growth trajectory of AI companies, empirically proving the economics of 'the discerning' — AI-native companies rapidly overtaking traditional ones.

Oxfam data showing 8 of the top 10 AI companies are controlled by billionaires. It reveals that the fruits of the AI revolution are concentrating among an extreme few, exposing the exclusivity of 'the discerning' class and the deepening of capital concentration.

Cross-reference to the Crunchbase VC investment data discussed in Chapter 3. It connects how the scale and concentration of AI investment form the capital foundation of 'the discerning' ecosystem.

An interview where Anthropic's CEO expressed 70-80% confidence that 'a one-person billion-dollar company will emerge this year.' It envisions an era where AI dramatically amplifies individual productivity, enabling a tiny minority to create outsized value.

Data showing 7 major tech companies spent $50 million on lobbying (Jan-Sep 2025), with Meta at $19.7 million (87 lobbyists) and OpenAI at $2.1 million (+68% YoY). It quantifies the scale of Big Tech's political influence and suggests the possibility of regulatory capture.

Bloomberg Government data showing total AI-related lobbying income at $92 million. It demonstrates the aggregate scale of interests at stake in AI regulation and quantifies capital's influence in democratic policymaking processes.

Data showing 3,570 AI lobbyists (26% of total), a 168% increase since 2022. It reveals the rapid surge in lobbying activity around AI regulation as one cause of institutional rigidity.

University of Maryland polling data showing 79% overall, 84% Republican, and 81% Democrat support for AI regulation. Despite bipartisan high support, the failure to legislate reveals a 'representation gap' between public will and political output.

Legal analysis showing zero comprehensive federal AI bills passed, with only the TAKE IT DOWN Act enacted. It demonstrates how severe the US AI regulatory vacuum is and empirically demonstrates the lag in institutional adaptation.

Data showing 38 states adopted approximately 100 AI-related measures. While state-level measures partially fill the federal vacuum, it demonstrates how regulatory fragmentation can actually compound confusion.

FEC disclosure and press coverage of Elon Musk's $250+ million support for Trump's campaign. It demonstrates that Big Tech leaders' political influence has reached unprecedented scale, suggesting the formation of a technology-politics complex.

CNBC reporting on the dispute between Anthropic and the Pentagon. It reveals the tensions between AI companies' ethical positions and national security demands, documenting real conflicts over military use of AI.

Washington Post coverage of the Pentagon designating Anthropic as a 'supply chain risk.' It demonstrates how conflicts between AI companies and national security agencies are materializing at the supply chain level.

CNN coverage of Anthropic setting red lines against autonomous weapons and mass surveillance of US citizens. It concretely demonstrates the tensions between ethical boundaries self-imposed by AI companies and government demands.

Biden administration's AI safety executive order issued October 2023, revoked by the Trump administration on January 20, 2025. It reveals the instability of US AI policy direction with administration changes and exposes the absence of institutional consistency.

The Trump administration's 'Promoting AI Innovation' executive order. It prioritizes innovation promotion through deregulation, demonstrating the policy pivot from Biden's safety-centered approach.

Legislation federally criminalizing deepfake sexual exploitation material — the only AI-related federal law passed in the US. It reveals the structure of US politics where legislation is possible only on specific issues with clear harm, amid an overall AI regulatory vacuum.

The Trump administration's national AI policy framework executive order issued December 11, 2025. It focuses AI policy direction on industrial competitiveness rather than regulation, reflecting strategic positioning in the US-China tech competition.

California's AI safety bill SB-1047 defeated by Governor Gavin Newsom's veto. It demonstrates the difficulty of AI regulatory legislation even at the state level and the challenge of achieving political balance between innovation and safety.

The EU AI Act enacted in 2024, establishing a risk-based 4-tier classification with fines up to 35 million euros. It demonstrates a 'rules-based' approach distinct from the US and China, providing a benchmark for comparing three regimes' AI regulation.

The 2010 Citizens United Supreme Court ruling that relaxed Super PAC campaign finance regulations. It opened the legal door for corporate political funding and laid the legal foundation for Big Tech's expanding political influence.

The world's first generative AI regulation, enacted August 15, 2023. As a case of China proactively moving on AI regulation, it demonstrates China's rapid institutional response in contrast to the US regulatory vacuum.

Stanford HAI data showing China's AI papers ranked first globally in 2023 (23.2%) with over 60% of global AI patent share. It quantifies the quantitative lead of Chinese AI research and stimulates discussion about when quantitative growth converts to qualitative breakthroughs.

Data showing China's semiconductor self-sufficiency at approximately 50% overall but only 19-23% for domestically designed and manufactured chips, with equipment self-sufficiency at 13.6%. It reveals how the reality of Chinese semiconductor capabilities varies dramatically depending on the definition of 'self-sufficiency,' exposing the practical limits of technological self-reliance.

Data showing over 2,100 Chinese government guidance funds with a target total of approximately 13.5 trillion yuan. It demonstrates the overwhelming scale of state-led venture investment and reveals the structural differences from the market-driven VC model.

Tsinghua University report showing China's core AI workforce growing from under 10,000 in 2015 to approximately 52,000 in 2024. It reveals the rapid expansion of the talent pool, while also suggesting the qualitative gap relative to the US should be considered.

Official State Council data: core AI industry at 1.2 trillion yuan, 6,000+ AI companies, and 1,590 EFLOPS of smart computing power. It shows the official scale and strategic direction of China's AI industry.

Projection of China's smart computing power reaching 2,781.9 EFLOPS by 2028. It quantifies the computing power expansion plans for AI infrastructure and shows the trajectory of infrastructure competition with the US.

CNNIC data showing 515 million generative AI users, accounting for 47.4% of internet users. It demonstrates the rapid mass adoption of AI technology and suggests the potential of data generation and AI utilization at a population scale of 1.4 billion.

Huawei annual report showing revenue of 862.1 billion yuan with 179.7 billion yuan (20.8%) invested in R&D and cumulative 10-year R&D exceeding 1.249 trillion yuan. It demonstrates the scale of Huawei's strategic investment in pursuing technological self-reliance amid US sanctions.

Baidu FY2025 earnings showing AI cloud revenue of 30 billion yuan with AI-related revenue growing at triple digits for 9 consecutive quarters. It demonstrates that monetization of Chinese AI companies is now in full swing.

Alibaba's announcement of investing 380 billion yuan (~$53 billion) in AI and cloud infrastructure over the next 3 years. It shows Chinese Big Tech's AI infrastructure investment is approaching US Big Tech scales, suggesting deepening infrastructure competition.

Caixin Global coverage quoting Huawei founder Ren Zhengfei: 'AI invention creates a company, but application strengthens a nation.' It symbolically reveals the core of China's AI strategy — focusing on application and diffusion rather than basic research.

OpenAI CEO Sam Altman's assessment of DeepSeek R1 as 'an impressive model, especially at that price.' This moment of a US AI leader acknowledging a Chinese competitor's efficiency is a symbolic case of the narrowing technology gap.

The 2006-2020 Medium- and Long-term Science and Technology Development Plan that formally established 'indigenous innovation' (zizhu chuangxin) as national strategy. It represents the origin of China's technology self-reliance strategy and became the starting point for all subsequent S&T policies.

Made in China 2025, announced in 2015, including AI, semiconductors, and robotics among 10 key industries. This core document of China's manufacturing upgrade strategy triggered US alarm and became a catalyst for technology decoupling.

The New Generation AI Development Plan (AIDP) announced in July 2017, setting a target of 10 trillion yuan in AI industry by 2030. The most important policy document for China's AI strategy, it systematically presents the roadmap and goals for state-led AI development.

Cross-reference to the world's first generative AI regulation discussed in Chapter 7. It demonstrates China's dual strategy of simultaneously pursuing innovation promotion and regulation in AI governance.

The Qinglang algorithm governance campaign implemented in November 2024. Targeting algorithm recommendation transparency and user protection, it demonstrates how Chinese AI regulation intervenes at the level of technical details.

The 'Accelerating AI+ Action' opinion issued in August 2025. It sets AI diffusion across all industries as a national goal and represents the implementation document for China's AI strategy that 'application strengthens a nation.'

AI-generated content labeling rules enacted September 2025, requiring both explicit and implicit labels. It demonstrates China's specific regulatory approach to AI transparency.

The AI Global Governance Action Plan released July 2025. It demonstrates China's strategic intent to secure leadership in international AI governance and presents an alternative to Western-centric AI norm formation.

Data showing 538 services completing the bei'an registration as of September 2025. It shows the operational status of China's AI service registration system and quantifies the concrete scale of regulatory enforcement.

2025 estimate showing approximately 700 million surveillance cameras in China. It reveals the dual nature of the world's largest surveillance infrastructure simultaneously functioning as the world's largest data collection infrastructure, illuminating the relationship between data and control.

Carnegie data showing Chinese AI surveillance technology adopted in 75+ countries. It demonstrates that Chinese AI surveillance technology is spreading globally beyond domestic borders, quantifying the export of AI-based digital authoritarianism.

Cross-reference to Chapter 4 data on China's power capacity additions (~430 GW in 2024). It connects how China's power expansion capability as an energy foundation for AI infrastructure serves as the physical basis for data processing and AI training.

Cross-reference to Chapter 2 Epoch AI data on the US-China AI model performance gap. It connects how China's execution capabilities leveraging 1.4 billion people's data are contributing to the narrowing performance gap.

DigiTimes interview where Kai-Fu Lee assessed that the 'DeepSeek moment' restored Chinese AI confidence. It captures the psychological turning point in the Chinese AI ecosystem that DeepSeek R1's success represented.

NCUSCR podcast where Kai-Fu Lee analyzed the US-China AI division of labor structure. It presents the perspective that the US lead in basic research and China's lead in application and diffusion could be complementary rather than competitive.

Tsinghua professor Xue Lan's academic speech pointing to 'proliferation of 130+ LLMs' and computing power bottlenecks. It provides an insider expert's diagnosis of inefficient resource dispersion in the Chinese AI ecosystem.

Chinese Academy of Sciences researcher Zeng Yi's concept of 'Harmonious AI.' It presents a distinctly Chinese AI governance philosophy distinguished from Western AI ethics frameworks, demonstrating civilizational diversity in AI norms.

National Bureau of Statistics data showing approximately 240 million gig economy participants and youth (16-24) unemployment at 18.8%. It reveals the structural severity of the dissolution of traditional employment and youth unemployment, quantifying the scale of China's 'displaced.'

Zhaopin survey showing 47.7% of graduates prefer state-owned enterprises and only 12.5% prefer private companies. It reveals that the stability-seeking psychology of Chinese youth is tilting toward state-sector dependence over innovation ecosystem participation.

An IMF working paper linking China's high household savings rate to the negative wealth effect from real estate. It analyzes how the structural cause of weak domestic demand lies in the decline of real estate asset values.

Cross-reference to the Chapter 2 data on real estate downturn's GDP impact. It connects how the economic suffering of China's 'displaced' sits atop the macro backdrop of real estate asset decline.

Data showing secondary housing prices in top 100 cities falling 7.60% year-on-year (Oct 2025). It demonstrates the continued downturn in China's property market and exposes the mechanism by which middle-class asset decline accelerates displacement.

Newsweek estimate of 20 million unemployed urban Gen Z. It reveals the absolute scale of Chinese youth unemployment and illuminates the structural suffering of a generation unable to enter the labor market in the AI era.

Statistics showing China installed approximately 270,000 industrial robots in 2023 (70% of world total) with the 3rd highest robot density globally. It reveals that automation was already accelerating before the AI wave, quantifying the physical foundation for labor replacement.

Meituan's official statement denying claims of 'tens of thousands of master's degree delivery drivers.' It reveals that the influx of highly educated youth into delivery platforms has become a social controversy, serving as a symbolic case of credential inflation and job mismatch.

Caixin reporting flexible work (gig) postings increased from 8.4% in 2019 to 15.2% in 2024. It quantitatively tracks the dissolution of permanent employment and the spread of precarious work.

Paper News reporting the average age at internet giants is 30-32. It demonstrates that China's peculiar age discrimination phenomenon — the '35-year crisis' (35sui weiji) — operates even within the technology industry.

Statistics showing 3.41 million applicants for the 2024 national civil service exam at an average competition ratio of 77:1. It demonstrates how private sector instability is driving extreme crowding into the public sector, suggesting a talent drain from the innovation ecosystem.

Chinese University of Hong Kong research showing the unemployment risk inflection point dropped from 54.3 years (2018) to 40.3 years (2022). It provides empirical evidence for the '35-year crisis' and demonstrates the rapidly declining age of labor market expulsion in the AI era.

Public lecture by Cai Fang, academician at the Chinese Academy of Social Sciences, warning of structural unemployment from robotization. As a diagnosis from one of China's top economists, it represents the severity of the impact of AI and robot automation on China's employment structure.

SemiAnalysis estimate that DeepSeek possesses approximately 20,000 GPUs (H100+H800). It demonstrates DeepSeek's strategic capability and efficient resource utilization in securing substantial computing resources despite US export controls.

QuestMobile data showing ByteDance's AI app Doubao surpassing 100 million DAU in 2025. It demonstrates the rapid mass adoption of Chinese AI apps and exemplifies how China's 'discerning' demonstrate strength in the application layer.

Cross-reference to Chapter 2 Epoch AI US-China AI performance gap data. It connects through performance data that the technical capabilities of China's 'discerning' are rapidly growing.

Benchmark data showing 9 of the top 10 models on HuggingFace's open-source leaderboard are from Chinese companies. It demonstrates that China's AI open-source strategy has established an overwhelming presence in the global ecosystem.

Huawei's 2024 annual report showing over 8,500 Kunpeng/Ascend ecosystem partners and 6.65 million developers. It demonstrates Huawei's 'discerning' strategy of building an indigenous AI chip ecosystem amid US sanctions.

IPC data showing Temu's global cross-border e-commerce market share at 24%. It demonstrates the model of Chinese platform companies leveraging AI to rapidly expand in global markets as 'the discerning.'

36Kr/ChinaTalk interview with DeepSeek founder Liang Wenfeng on his founding philosophy. It is a key case applying the frame of 'those who built the system are the discerning' to the Chinese AI ecosystem.

Coverage of ByteDance's 2025 AI CapEx reaching 160 billion yuan (~$22 billion). It reveals that China's largest internet company's AI investment is approaching US Big Tech scales.

Lawfare article analyzing DeepSeek's designation as a national high-tech enterprise. It illuminates from a legal perspective the structural characteristics of the Chinese AI ecosystem where the relationship between private companies and the state is fundamentally different from the West.

IC Insights data showing China's domestically designed and manufactured chips at approximately 19.4% (2024). It demonstrates the large gap between overall self-sufficiency rates and actual indigenous capabilities, quantifying the practical limits of China's semiconductor self-reliance.

TechInsights data showing China's domestically designed and manufactured chips at approximately 23.3% (2024). The discrepancy with IC Insights (19.4% vs. 23.3%) reveals methodological variations and suggests the difficulty of accurate assessment.

IDC data showing NVIDIA+AMD at 71%, Huawei HiSilicon at 23% in China's 2024 AI chip market, with domestic AI chip penetration rising from 15% (2023) to 30% (2024). It demonstrates that China's indigenous chip market share is expanding rapidly despite export controls.

IMF Article IV report expressing concerns about China's long-term debt sustainability. It reveals the structural dilemma where technology investment and growth momentum conflict with debt management.

IMF data showing China's government debt at approximately 127% of GDP (expanded definition, projected 135% by 2026). It reveals the scale of the actual debt burden including Local Government Financing Vehicles (LGFVs), suggesting limits on fiscal capacity.

Atlantic Council analysis of China's real estate crisis response, identifying an 'Extend and Pretend' strategy. It structurally critiques the approach of buying time rather than fundamentally solving the property problem.

Morningstar projection of China's real estate market stabilizing by late 2026 to 2027. While it suggests a potential resolution timeline, the cumulative economic losses until then must also be considered.

ASML data showing China revenue share at 36.1% annually, peaking at 49% (Q2), with a 2026 forecast of approximately 20%. It demonstrates the changing dynamics of China dependency in the semiconductor equipment supply chain and quantifies the economic cost of export controls.

Data showing CHIPS Act subsidy allocations (Intel $7.86B, TSMC $6.6B, Samsung $4.75B). It quantifies the fiscal scale and investment targets of the US semiconductor reshoring strategy.

CASS projection showing the active-to-beneficiary ratio in urban worker pension insurance deteriorating from 2.57:1 (2020) to 0.89:1 (2050). It quantifies the structural crisis of population aging and pension financing, raising questions about social safety net sustainability in the AI era.

Financial data showing Evergrande's debt at 2.5 trillion yuan with only 1.8 billion yuan in cash. As a symbolic case of the Chinese real estate crisis, it reveals the concrete scale of structural limitations.

Interview with Hong Hao projecting 'years to a decade' for real estate sector repairs. It represents a market expert's diagnosis of the long-term nature of the Chinese real estate crisis.

Former IMF Deputy Managing Director Zhu Min's remarks on AI optimism and the $7 trillion pandemic output gap. It presents a senior policy expert's perspective on whether AI can offset China's structural economic limitations.

CFR's Huang Yanzhong analyzing the limited effectiveness of China's pro-natalist policies. It demonstrates that the demographic cliff is a structural problem difficult to reverse through policy intervention, raising whether AI automation could serve as an alternative to labor shortages.

Data showing TSMC's foundry market share expanding from 59% (2023) to 66% (2025) with Samsung at 12-13%. It reveals that extreme concentration in semiconductor manufacturing is a key geopolitical risk variable, explaining the structural backdrop of the 'chip war.'

SemiAnalysis data showing DeepSeek's total server CapEx at approximately $1.6 billion and V3 training cost at approximately $5.58 million (2.78 million GPU hours). It quantifies DeepSeek's strategy of building competitive models through efficient resource utilization in an export-controlled environment.

DeepSeek R1 API pricing at $0.55 per million input tokens and $2.19 per million output tokens. Compared to OpenAI o1's $15/$60, the 27-30x cheaper pricing opens a new dimension of tech competition.

OpenAI o1 API pricing ($15 input/$60 output) as a benchmark for comparison with DeepSeek. It exposes the structural difference between American AI companies' premium pricing strategies and China's low-cost strategies.

Data on Tesla and Samsung's $16.5 billion long-term supply contract. It provides a concrete example of decoupling where US companies strengthen relationships with Korean manufacturers in the semiconductor supply chain reorganization.

Cross-reference to Chapter 4 NVIDIA earnings, CHIPS Act subsidies, and IC Insights data. It connects how these key figures constitute the strategic backdrop of the chip war and decoupling.

Coverage quoting ASML CEO Christophe Fouquet from the Q2 2024 earnings call. It illuminates the relationship between export controls and the Chinese market from the perspective of a key company in the semiconductor equipment supply chain.

KFF data showing the national average COBRA individual premium at $584/month. It reveals the specific scale of medical cost burden facing the American comparative pair character (42-year-old paralegal) after layoff, quantifying the economic reality of 'free fall' in the absence of a social safety net.

Survey showing 77% of legal professionals responded that AI would have a 'significant impact.' It demonstrates that legal professionals themselves recognize AI-driven change and quantifies the anxiety among legal workers including paralegals.

Official BLS statistics showing approximately 345,800 paralegals with a median salary of $61,010. It provides the total scale and economic level of the American comparative pair's occupational group, quantifying the potential scope of AI automation's impact.

Indeed data showing paralegal job postings declining 18% from 2022 to 2024. It empirically demonstrates that demand for paralegals in the actual hiring market is shrinking following the adoption of AI legal tools.

Robert Half data showing legal job postings requiring AI skills surging from 12% (2023) to 34% (2024). It demonstrates that AI proficiency is becoming a key requirement for transitioning from 'displaced' to 'discerning.'

Alibaba investor report showing AI handles 97% of first-tier Taobao/Tmall customer inquiries. It provides the backdrop for the Chinese comparative pair character's (38-year-old CS manager) situation, demonstrating the concrete scale of AI replacing customer service.

Data showing JD.com's AI customer service system JIMI automatically handles over 90% of inquiries. It demonstrates that CS automation in Chinese e-commerce is already ubiquitous and that large-scale reduction of CS roles is inevitable.

BOSS Zhipin data showing 80%+ of internet/IT job postings include age restrictions. It reveals how systematic age discrimination in China's tech industry structurally blocks the 38-year-old CS manager's re-employment.

Zhilian Zhaopin data showing interview invitation probability for those over 35 is 60% lower than for ages 25-30. It provides empirical evidence for the '35-year crisis,' quantifying how steep the age barrier is in the Chinese labor market.

Ministry of Human Resources data showing only 42% of those aged 35-45 find re-employment within 6 months. It provides official statistical backing for the re-employment difficulty facing the Chinese comparative pair character.

Statistics showing only approximately 2.3 million unemployment insurance recipients — a take-up rate of approximately 1%. It reveals the extremely low coverage of China's social safety net and exposes the structure where unemployment pain is transferred to individuals in 'stasis within control.'

AARP data showing 78% of job seekers over 40 experience age discrimination in the US. It demonstrates that age discrimination is a universal phenomenon not limited to China, revealing the shared suffering of both countries' comparative pair characters.

China Index Academy data showing Shenzhen housing prices falling 30-35% from peak. It provides the backdrop where real estate asset decline in Shenzhen — where the Chinese comparative pair character resides — amplifies the economic pain of displacement through the negative wealth effect.

Coverage of AI legal startup Harvey AI raising $100 million in Series C at a $1.5 billion valuation. It symbolizes the paradox where the AI tools displacing paralegal jobs simultaneously generate massive investment returns — one person's displacement becomes another's discernment.

The case of major law firm Allen & Overy being the first to adopt Harvey AI. It demonstrates the pathway of legal AI tool adoption starting at large firms and spreading industry-wide, concretizing the trigger for paralegal workforce reductions.

Testimony of Lancashire hand-loom weavers before the 1834 British Parliamentary Committee. A historical source conveying the voices of skilled workers during the Industrial Revolution, it reveals striking parallels with the experiences of AI-era 'displaced.'

An empirical study showing that displaced high-income workers experience a long-term 15-20% income decline even after re-employment. It demonstrates that the economic impact of displacement does not end at the moment of layoff but persists as a structural scarring effect.

A landmark study analyzing 101 years of global investment returns. It presents historical evidence that optimists were rewarded over the long term, providing a foundational long-term investment perspective for evaluating the three scenarios of US-China competition.

Epoch AI data showing the MMLU gap narrowing from 17.5 percentage points (2023) to 0.3 (2025), and HumanEval from 31.6 to 3.7 percentage points. It quantitatively demonstrates that the US-China AI performance gap has effectively disappeared on benchmark measures.

Recorded Future analysis showing the US-China AI time gap narrowing from an average of 7 months to 3-6 months. It quantifies the Chinese catch-up speed in time units, serving as key data for gauging the realistic probability of each of the three scenarios.

Data showing China's open-source AI global share surging from 1.2% (end of 2024) to 30% (August 2025). The dramatic share change in just 8 months demonstrates how China's open-source AI strategy is rapidly reshaping the global ecosystem.

BlackRock analysis projecting $5-8 trillion in AI capital expenditure by 2030, with 54% of respondents seeing greater AI opportunity in energy companies. It demonstrates the aggregate scale of AI investment and the investment opportunity in the energy-AI nexus.

Microsoft report showing DeepSeek usage in Africa is 2-4x that of US AI. It provides empirical evidence for the 'divided world' scenario where Chinese AI models spread faster in developing countries.

Goldman Sachs projection that GenAI will boost global GDP by 7% (~$7 trillion). It demonstrates that AI's macroeconomic impact will be enormous across all three scenarios, suggesting that distribution differences are the key variable.

McKinsey projection of $13 trillion in additional AI-driven economic activity by 2030. Together with the Goldman Sachs projection, it forms a consensus on AI's economic potential and allows gauging the economic scale of the three scenarios.

World Economic Forum Global Risks Report ranking geoeconomic risk as #1. It demonstrates that US-China tech competition is recognized as the most important risk globally.

White & Case analysis showing an average of 9 months from conception to implementation of Chinese AI regulations. It demonstrates China's faster institutional adaptation speed compared to the US regulatory vacuum, quantifying the time dimension of institutional competition.

Pantheon Insights projection of hardware bifurcation becoming visible by 2026. It forecasts the specific timing of US-China tech decoupling manifesting at the hardware level, supporting the technical plausibility of the 'divided world' scenario.

Data showing 44 countries participating in the Global Partnership on AI (GPAI). It demonstrates multilateral cooperation attempts in AI governance and suggests the possibility of a third pathway beyond the US-China bipolar system.

Oxford Blavatnik School of Government research comparing AI policy speed across democracies and authoritarian regimes. It analyzes structural differences in AI policy responses by regime type, providing an academic foundation for the core question: 'which system is more adaptive in the AI era?'

Former Google CEO Eric Schmidt's prediction of accelerating US-China AI technology divergence. It supports the plausibility of the 'divided world' scenario from a tech industry leader's perspective.

Kai-Fu Lee's forecast of a bifurcation between countries adopting Chinese apps versus US apps. It presents the specific contours of an AI app ecosystem dividing along geopolitical lines.

Yuval Noah Harari's argument for the efficiency of centralized data processing in the AI era. It presents the controversial perspective that authoritarian regimes may have structural advantages in the AI era, triggering theoretical debate on democracy vs. authoritarianism in AI competition.

NYU Stern Professor Damodaran's data showing S&P 500 total return averaging approximately 10.7% annually, roughly 32x cumulative over 35 years. It provides the historical baseline for equity investment returns and serves as a benchmark for AI-era investment strategies.

CME Group data showing Nikkei 225 total return of +54% (yen terms) from 1989-2024 including dividend reinvestment. The fact that long-term investors earned positive returns even during 'the Lost 30 Years' demonstrates that investment strategies can be effective even in worst-case scenarios.

Data showing a global 60/40 portfolio's 30-year annualized return at approximately 8.2%. It demonstrates the long-term performance of diversified investment and provides the foundational principle for portfolio construction amid US-China competition uncertainty.

Cross-reference to Chapter 15 BlackRock AI capex projection ($5-8 trillion). It connects the significance of AI infrastructure investment as key reference data for constructing the investor's map.

Cross-reference to Chapter 4 NVIDIA earnings data. It connects how NVIDIA serves as a core beneficiary and bellwether for AI infrastructure investment from the investor's perspective.

Data showing SK Hynix's HBM (High Bandwidth Memory) market share at 57-60%. It demonstrates the pivotal position of Korean companies in the AI semiconductor supply chain, connecting to the strategic position of Korea discussed in the epilogue.

Projection showing the Korean memory semiconductor market growing 98% year-on-year in 2026 to a market size of $445 billion. It demonstrates that AI demand is driving explosive growth in the memory semiconductor market.

Cross-reference to Chapter 4 IEA data center power consumption projections. It connects how energy infrastructure serves as both a new investment opportunity and constraint in AI investment.

Projection of combined CapEx from 5 hyperscalers reaching $660-690 billion in 2026. It demonstrates the aggregate scale of AI infrastructure investment and the corresponding opportunities available to investors.

Data on Tesla Optimus's 2026 production target of 50,000 units and market projection of $5 trillion by 2050. It demonstrates the potential of the 'Physical AI' market created by the convergence of AI and robotics, presenting its viability as a long-term investment theme.

Data on SEMICON Korea 2026's 'Physical AI' theme. It demonstrates the direction of the semiconductor industry expanding from software AI to physical AI and suggests a new growth driver for Korea's semiconductor industry.

A comprehensive reference on the geopolitical significance of the semiconductor industry. It analyzes key companies like TSMC, ASML, and NVIDIA along with the history and structure of the US-China semiconductor competition, serving as essential background for the book's technology hegemony discussion.

A macro framework analyzing the rise and fall of empires through debt, currency, and military power cycles. This book serves as the departure point for extending Dalio's macro analysis to the specific variable of AI and down to individual lives.

A pioneering work analyzing the early contours of US-China AI competition. It presented the dichotomous frame of US 'discovery' advantage versus China's 'execution' advantage — a framework this book critically updates.

The IMF's flagship publication providing quarterly and annual growth, inflation, and trade data for the global economy. It serves as the foundational data source for US-China economic comparative analysis.

An IMF database tracking the quarterly currency composition of global foreign exchange reserves. It is a key indicator source for monitoring dollar hegemony and the progress of yuan internationalization.

An OECD annual report comparing R&D spending, research personnel, and patent data across nations. It serves as the standard data source for quantitative comparison of US-China science and technology capabilities.

A World Economic Forum annual report assessing global risk types and severity. It is useful for understanding the prioritization of geoeconomic risks, technology risks, and other categories.

IEA annual reports analyzing global energy supply-demand projections and AI's energy impact. They serve as essential references for understanding the energy constraints of AI infrastructure.

A WIPO annual index comprehensively evaluating national innovation capabilities and outcomes. It provides a standard framework for comparing the relative positions of US and Chinese innovation competitiveness.

Stanford HAI's annual report synthesizing major trends in AI research, investment, policy, and ethics. It is the most comprehensive single data source for US-China AI capability comparison and is used as a key reference throughout the book.

Epoch AI benchmarks tracking AI model performance, training costs, and computing power quarterly. It is a key tool for quantitatively monitoring the dynamic changes in the US-China AI model performance gap.

MacroPolo's annual report tracking the origin countries, current affiliations, and research activities of the world's top AI researchers. It is the most useful data source for understanding the geographic distribution and movement patterns of AI talent.

An independent research firm providing technical and economic analysis of semiconductors and AI infrastructure. It is a major source for many of the book's key data points, including DeepSeek's GPU inventory and Huawei Ascend yields.

A specialist media outlet conveying the Chinese AI ecosystem to Western audiences. It provides in-depth analysis of the strategies and technical details of Chinese AI companies like DeepSeek and ByteDance.

A Taiwan-based research firm providing price, shipment, and market share data for semiconductors, displays, and LEDs. It is useful for tracking real-time trends in the semiconductor supply chain.

A global research firm tracking shipments and market share in smartphones, IoT, and semiconductors. It is useful for monitoring structural changes in technology markets.

A specialist research firm analyzing semiconductor industry market size, self-sufficiency rates, and technological capabilities. A key data source for Chinese semiconductor self-sufficiency rates, extensively used in Chapters 8 and 12.

The global association for the semiconductor equipment and materials industry. It provides semiconductor manufacturing equipment market and supply chain data, useful for understanding the industry context of key equipment companies like ASML.

A global research firm tracking IT market spending, shipments, and market share. It provides key data such as NVIDIA vs. Huawei market share in China's AI chip market.

An American defense and foreign policy research institution. It provides strategic analysis of US-China military and technology competition and conducts in-depth research on the military applications and security implications of AI.

A bipartisan foreign policy and security think tank. It is a key institution providing policy reports and expert analysis on US-China tech competition, semiconductor supply chains, and AI policy.

A leading US progressive think tank. It provides policy research on AI policy, labor market changes, and inequality, useful for analyzing social impacts within the United States.

The Carnegie Endowment for International Peace. It is a key institution analyzing the geopolitical implications of technology, including the global spread of Chinese AI surveillance technology and digital authoritarianism.

A foreign policy and security think tank centered on the Atlantic alliance. It covers the economic dimensions of US-China competition, including analysis of China's structural economic issues and real estate crisis response.

NVIDIA's official investor relations site. It directly provides the most important corporate data in the AI infrastructure market, including quarterly and annual earnings, GPU shipments, and data center revenue.

A CNBC reporting series continuously tracking AI-related capital expenditure by Big Tech companies. It is the most timely media source for understanding the investment scales and directions of hyperscalers.

A platform providing global venture capital investment data. It is a key data source for tracking AI startup fundraising scale, round-by-round trends, and geographic distribution.

BVP reports tracking growth in the AI-native software market. Analyzing commercial successes like Cursor surpassing $1 billion ARR, they serve as a key investment data source empirically demonstrating the economics of 'the discerning.'