Opening
January 2026. Icheon, Gyeonggi Province.
Engineers at the airlock of SK hynix's M16 fab change into anti-static suits. Shoe covers, hairnets, gloves. A ritual repeated before every cleanroom entry. The door opens onto another world. A space where airborne particle density is held below one hundred per cubic meter, hundreds of times cleaner than the air outside. Under fluorescent lights, wafer carriers glide silently along the automated transport system.
This is where the HBM4 mass-production line has begun operating.
A finished chip measures roughly one centimeter across and weighs a few grams. Yet without it, NVIDIA Blackwell accelerators in San Francisco data centers stop computing. High-performance clusters in Shenzhen AI labs lose their speed. G42 infrastructure at Abu Dhabi's hyperscale campus drops to standby.
A small object made in one factory in one country powers algorithms across three continents.
This is the leverage of a "small nation."
The main text focused on two empires — the United States and China. It traced how technology concentrates capital, how capital concentration breeds social instability, and how social instability forces institutional redesign — a formula operating simultaneously in both empires. But the world does not run on two countries alone.
NVIDIA designs. TSMC fabricates. ASML (Netherlands) holds a monopoly on the equipment. SK hynix (South Korea) supplies 57–60% of AI memory. Neither the United States nor China possesses a self-contained supply chain. The U.S.-China power contest is a frame, not the whole picture. Inside that frame — and sometimes exploiting it — "third players" survive and prosper.
The epilogue tracks them.
South Korea, India, the EU, the Middle East. Four nations and regions. Each wields a different form of leverage. South Korea controls the AI supply chain's bottleneck: HBM. India has 1.4 billion people and a homegrown digital infrastructure. The EU has the world's first comprehensive AI regulatory standard. The Middle East has enormous capital in the midst of pivoting away from oil.
All four stand before a single common question.
How long does this leverage last?
Section A: Four Pathways — The Options Available to Middle Powers
Possess What Cannot Be Done Without
In February 2026, Chatham House (the Royal Institute of International Affairs) published a report. Its title: How Middle Powers Can Weather US and Chinese AI Dominance. The document mapped four pathways open to middle powers caught between the two AI superpowers.
The report matters because of its timing. By 2026, the U.S.-China AI gap had narrowed to three to six months. The narrower the gap, the greater the bargaining power of middle powers. Both the United States and China need allies. Forty-four nations participate in the Global Partnership on AI (GPAI), and the annual license-renewal regime has placed supply chain participant countries in positions of structural choice.
The report divides a middle power's AI capability into eight building blocks: data, compute, models, energy, industry, talent, infrastructure, and trust. Which blocks a nation is strong in determines which pathways are viable.
The first pathway is Specialize. Concentrate on a specific bottleneck in the global supply chain and secure an indispensable role. The Netherlands' ASML is the textbook case. A country of 17 million people is home to the world's sole manufacturer of EUV (extreme ultraviolet) lithography machines. Without these machines, neither TSMC nor Samsung nor Intel can produce advanced chips. ASML has not sold a single unit to China since 2019. The company's quarterly revenue share from China peaked at 49% in Q2 2024; after export controls took effect, its annualized China share fell to 20%. Its 2030 revenue forecast is $71 billion. That is the power of specialization.
The second pathway is Align. Enter a full alliance with one of the AI superpowers, trading strategic autonomy for technology access and security guarantees, a ticket into the most advanced ecosystem. The risks are clear: severance from the opposing camp, and exposure to the political shifts of one's allied partner.
The third pathway is Share. Pool sovereignty with like-minded nations to build collective bargaining power. The EU's 27 member states, bound by a single market, produced the world's first comprehensive AI regulation, the AI Act. Slow decision-making is the weakness, but once a standard is set, it becomes a global benchmark. Just as GDPR became the worldwide baseline for data protection within seven years.
The fourth pathway is Hedge. Selectively combine AI capabilities from multiple sources (the United States, China, Europe) without locking into any single camp. Not straddling two sides, but standing on many legs. India hires 32,000–33,000 employees from America's FAANG companies while chairing BRICS and co-authoring G20 AI principles.
The four pathways are not mutually exclusive. South Korea's mapping is Specialize + Align. It holds the HBM bottleneck while maintaining a U.S.-tilted alliance structure. India's mapping is Hedge + Share. It combines American technology with homegrown infrastructure and leverages both BRICS and the G20. The EU is Share + Specialize. The single market's regulatory standard is its foundation, and ASML is its hidden hardware lever. The Middle East is Hedge + Align. Capital is the weapon — buying American technology while partly utilizing Chinese infrastructure.
The four pathways Chatham House outlined are not abstract strategy. They are strategies that four players are executing, right now, each in their own way. The rest of the epilogue shows the results.
Section B: South Korea — HBM Supremacy and the Art of the Tightrope
The Day SK hynix Overtook Samsung
In 2025, a reversal occurred in Korea's semiconductor industry for the first time.
SK hynix annual operating profit: 47.2 trillion won. Samsung Electronics semiconductor division operating profit: 43.6 trillion won. Since Samsung Electronics was founded in 1969, no domestic competitor had ever surpassed Samsung in semiconductor revenue. Samsung held the number-one position in the DRAM market for decades. SK hynix was the perennial number two. At the heart of this reversal was HBM, High Bandwidth Memory.
Understanding what HBM is reveals why this reversal is a geopolitical event.
When training an AI model, a GPU must exchange terabytes of data with memory every second. Conventional DRAM cannot handle this speed. HBM vertically stacks multiple DRAM dies and connects them with TSVs (through-silicon vias), boosting bandwidth by orders of magnitude over conventional memory. The NVIDIA H100, H200, and Blackwell architectures are all impossible to design without HBM.
SK hynix is the world leader in this technology. Its global HBM market share: 57–60%. In February 2026, SK hynix and Samsung simultaneously began mass-producing HBM4. For OpenAI's Stargate project, both companies signed letters of intent (LOIs) to supply 900,000 DRAM wafers per month. South Korea holds the heart of the AI supply chain.
The Structure of the Tightrope
Yet this position is privilege and pressure at the same time.
Start with the security axis. South Korea has formally joined the U.S.-led Chip 4 alliance (the United States, Japan, Taiwan, South Korea). Through CHIPS Act subsidies, Samsung received $4.75 billion. On this axis, South Korea is the critical component supplier for American AI infrastructure.
The trade axis points in the opposite direction. China is South Korea's largest trading partner. Both Samsung and SK hynix operate legacy semiconductor production facilities inside China. Without China, the revenue structure of Korea's semiconductor industry changes fundamentally.
Starting in 2026, export licenses shifted to an annual renewal system. Every year, the political situation in the United States determines the permissible scope of Korean companies' China operations. On one side, cutting-edge HBM ships to American data centers. On the other, legacy memory flows to the Chinese consumer market. Balancing on this tightrope, Korean companies must renegotiate every year.
The difficulty of the tightrope is defined as follows. When U.S. export control pressure intensifies, the business scope of Korean factories inside China narrows. Conversely, abandoning China means losing a significant share of total revenue. Since the Trump administration, U.S. export controls have grown tighter. The H200 export review change that took effect on January 15, 2026, shifted from "presumption of denial" to "case-by-case review" — but this relaxation applies only to allies. South Korea's position is close to allied status. How close is reverified every year.
The AI Basic Act and Sovereign AI
South Korea's response atop this tightrope has three components.
First, the AI Basic Act took effect on January 22, 2026. A comprehensive statute consolidating 19 separate AI bills into one, it is the world's second comprehensive AI regulation to be enforced. Unlike the EU's AI Act, which was born from years of negotiation, South Korea completed drafting and enforcement in two years, dramatically compressing the historical band of institutional adaptation (14–64 years) analyzed in Volume 1.
Second, a sovereign AI strategy. Five consortia (Naver, SK Telecom, LG AI Research, NC AI, and Upstage) are developing independent AI models. According to Naver, HyperCLOVA X has 6,500 times more Korean-language training data than GPT-4. The National AI Computing Center is set to deploy 15,000 GPUs by 2027, with total AI infrastructure investment reaching $65 billion.
Third, technology diversification. Rather than resting on HBM as today's leverage, Korea is investing in next-generation memory architectures. Just as the industry transitioned from 3D NAND to HBM, the strategy is to capture the design layer in post-HBM4 architectures.
Where the Displaced Accumulate
Behind the glow of the HBM boom, a shadow falls.
According to OECD analysis, 38.8% of all jobs in South Korea are "high-risk" positions where more than 70% of tasks are exposed to automation. This ratio places Korea among the highest in the 38-nation OECD. While the semiconductor sector added 2.8% in employment, manufacturing employment overall retreated. KDI (Korea Development Institute) analysis found that when IT investment doubles, total firm headcount drops by more than 60, but per-capita wages for those who remain go up.
The pattern identified in Volume 1 is repeating with precision. Productivity explodes, wages rise for those who stay, and the displaced grow in number. Engels' Pause — a period in which productivity rises while wages stagnate or fall — is operating in South Korea too.
Engineers on the HBM line saw their salaries increase. Subcontractors on the factory's periphery found fewer places to go. Two South Koreas are being created simultaneously — one inside the cleanroom, one outside it.
The country that holds the AI supply chain's bottleneck may also be the country most exposed to automation shock. The boundary between the discerning nation and the displaced nation is drawn not along national borders, but along industry sectors and skill levels.
Section C: India — A Third Way Through Scale
February 4, 2026, Afternoon. Bangalore.
Lunchtime had just ended at TCS Manyata Tech Park. More than 50,000 people work on this campus, one of the largest in Bangalore. Employees filed out of the cafeteria and headed back to their desks.
A notification lit up Ravi's phone. A software tester with eight years of experience. The Nifty IT index had plunged 6% intraday — the steepest drop since March 2020. Infosys, TCS, and Wipro fell 5-8% in lockstep. News of Anthropic's Claude Cowork launch had swept through the market. Some 2 trillion rupees ($23 billion) in market capitalization across India's leading IT companies vanished in a single day.
Ravi knows what he does for a living. He generates software test scenarios, logs bugs, and writes reports. Repetitive, structured work. He also knows he is training the AI to do that work. Filling spreadsheets, labeling patterns, documenting edge cases. And Ravi knows the AI he is training is the AI that will replace him.
India's $283 billion IT outsourcing industry has been hurtling toward this moment. TCS cut 12,000 jobs in 2025. India's four largest IT firms laid off more than 25,000 in the first half of 2025 alone. Forecasts emerged that up to 500,000 IT jobs could disappear within several years. The most vulnerable: mid-level workers with four to twelve years of experience. The layer AI absorbs first and fastest.
Projections also surfaced that if IT exports collapse, the rupee will weaken further against the dollar. The pathway from one industry's crisis to a currency crisis.
This is India's version of Engels' Pause.
India Stack: The Infrastructure the World Is Trying to Replicate
India's paradox lies here.
At the very moment AI is eating into India's largest export industry, India possesses a digital infrastructure the rest of the world is trying to copy.
India Stack is a three-layer Digital Public Infrastructure (DPI). The first layer, Aadhaar, is a biometric-authenticated digital identity for 1.4 billion people. Opening a bank account, receiving government subsidies, linking hospital records — all done through a single number. The second layer, UPI (Unified Payments Interface), is a mobile payment network enabling instant, free transfers. Money can be sent with nothing more than a phone number, no bank account required. The third layer, DigiLocker, is a platform for digitally storing and authenticating every official document — from academic transcripts to driver's licenses to tax filings.
The effects produced by this three-layer structure are material. In 2014, 53% of India's adult population had a bank account. By 2022, the figure exceeded 80%. A transformation in eight years.
By February 2026, 23 countries had signed DPI cooperation memoranda of understanding (MoUs) with India. A growing number of nations are adopting UPI-style instant payment systems, and governments seeking to introduce Aadhaar-style biometric identity systems are consulting India. These are the first large-scale cases of digital infrastructure becoming an export commodity.
This is India's second lever. At the moment the execution-layer leverage of IT outsourcing is shaking, a design-layer leverage, digital infrastructure architecture expertise, is rising.
Sovereign AI and the Population Paradox
In sovereign AI, too, India charts its own course.
BharatGen Param2 is a 17-billion-parameter multimodal AI model supporting 22 Indian languages — Hindi, Tamil, Telugu, Marathi, and more. A linguistic space that English-language frontier models barely cover. Of India's 1.4 billion people, only 10–12% speak English fluently. To serve the remaining 1.2 billion with AI, Indian-language AI is necessary. The country added 20,000 GPUs to its existing base of 38,000 while cutting the subsidized GPU access cost to 65 rupees (roughly $1) per hour — lowering the barrier for startups and researchers to reach frontier compute.
Diplomatically, India's position is unique. The country that simultaneously chairs the G20 and BRICS. While FAANG companies hire 32,000–33,000 employees in India, India hosts the AI Impact Summit 2026 and positions itself as mediator of global AI governance. The strategy: belong fully to neither camp while becoming indispensable to both.
India's hedging strategy, however, carries a prerequisite.
The working-age population (15–64) stands at 68% — among the highest in the world. Twelve million people enter the labor market every year. In theory, a demographic dividend. In practice, only 4% of those aged 15–29 have received formal vocational training. McKinsey estimates that 70% of Indian jobs will be exposed to AI risk by 2030. If IT jobs shrink and manufacturing jobs fail to grow fast enough in a labor market absorbing 12 million entrants per year, the demographic dividend becomes demographic pressure.
A demographic dividend is not automatic. Whether India Stack — the digital infrastructure — can serve as a tool to accelerate institutional adaptation, or whether AI will erode employment faster than the infrastructure can respond: on that outcome rests the success or failure of India's third way.
Back to Bangalore's TCS Tech Park. Ravi stares at his screen and runs a single calculation. How fast the AI he is training will replace his work. Whether that speed exceeds the speed at which he can build new capabilities. This calculation is the one India's entire IT industry is running right now.
Section D: The EU — The Regulatory Superpower's Dilemma
Two Levers
ASML headquarters in Veldhoven, in the south of the Netherlands. Moving a single EUV (extreme ultraviolet) lithography machine built here requires 40 trucks. Weight: over 160 tons. Components: more than 100,000. Price: over $300 million. Without this machine, no one can manufacture advanced semiconductors at 7 nanometers or below. Only one company in the world makes it: ASML.
Not a single unit has been sold to China since 2019. The company's 2030 revenue forecast is $71 billion. A bottleneck that grows even after excluding China.
A narrative persists that the EU trails the United States and China in the AI race. What that narrative omits is ASML. A single company in a country of 17 million holds the most decisive lever in the U.S.-China semiconductor war.
In early 2026, ASML invested €1.3 billion as lead investor in the Series C round of Mistral AI — Europe's sole frontier AI contender. A hardware equipment company became the largest external shareholder of an AI software company. For the first time, the EU's two hidden levers — the semiconductor supply chain bottleneck and a European frontier AI firm — joined hands.
The AI Act: Can Regulation Build an Industry?
On August 2, 2026, the EU AI Act's rules on high-risk systems take full effect. The world's first comprehensive AI regulation. AI systems deployed in high-risk domains (medical diagnosis, credit scoring, hiring decisions, educational assessment) must meet transparency requirements and human oversight obligations. Violations can incur fines of up to €35 million or 7% of global revenue.
Yet the EU's regulatory leadership has not always worked in the EU's favor. After GDPR took effect, EU venture capital investment fell 26% relative to the United States. Annual compliance costs run to an estimated €16 billion. The Mario Draghi report, commissioned by the European Commission, flagged GDPR as an issue requiring "strategic recalibration."
The EU's dilemma is this: it has AI to regulate, but insufficient capacity to build the AI that gets regulated. Of the ten largest AI companies by global market capitalization, zero to one are EU-based. While 60% of global AI venture capital concentrates in the San Francisco Bay Area, the EU's share is 13%. Borrowing the Rome-versus-Carthage frame from Volume 1, the EU structurally resembles Carthage: possessing sophisticated rules but lacking market scalability.
Yet 2025 data hints at a different possibility. European AI venture capital investment grew 55% year over year. At the Paris AI Action Summit, France announced a €109 billion AI infrastructure investment plan. Mistral AI's valuation exceeded $14 billion, and its annual recurring revenue (ARR) hit $400 million, a twenty-fold increase in one year.
Whether regulation can create a trust-based ecosystem, and whether that trust can attract companies and capital in a virtuous cycle. The EU's experiment remains unresolved. For American AI companies to enter the EU market, they must comply with the AI Act. Compliance costs are "internalized" for EU firms but "additional" for American ones. This asymmetry could grant EU companies a home-field advantage.
Can regulation build an industry? The EU is writing the answer now.
Section E: The Middle East — From Oil to Algorithms
When Capital Changes Direction
On the wall-mounted screen in a conference room at MGX headquarters in Abu Dhabi, two lines are plotted. One tracks the UAE's remaining proven oil reserves by years of extractable supply. The other tracks the current scale of AI infrastructure investment. At the intersection, an analyst speaks: "When those lines cross, only one of two things will remain. A new form of leverage, or nothing at all."
This urgency explains the speed of Gulf AI strategy.
Consider the strategy of Crassus. When fire broke out in Rome's wooden insulae, Crassus's fire brigade arrived first. But they did not spray water until the owner agreed to sell the building at a fraction of its value. The fire was the opportunity.
The Gulf states practice a variant of the strategy described in Chapter 15, but here the fire is the impending decline of oil. Before the value of oil declines, buy the assets of the next era in bulk. Oil (material leverage) → data centers (semi-material leverage) → AI services (immaterial leverage). A pattern of "leverage dematerialization." The historical current in which the source of power migrates from land to factories, from factories to API calls — the Gulf states have read it.
The scale of investment reveals the urgency of this pivot. The Stargate UAE campus, a joint venture of the UAE's G42, OpenAI, and NVIDIA, is a 5 GW facility with $20 billion in investment. HUMAIN, under the Saudi Public Investment Fund (PIF), is co-building a 500 MW AI factory with NVIDIA. Ninety-five percent of Gulf enterprises surveyed said they plan to invest in sovereign AI platforms within three years.
Two conditions work in their favor. First, money. Oil revenues remain abundant, making now the optimal window for investment. Second, energy. Saudi Arabia's solar power generation costs are among the lowest in the world, and the UAE operates nuclear power plants. One of the largest cost drivers for AI data centers is electricity. An energy cost advantage is a structural edge that lowers AI infrastructure operating expenses.
But this strategy has structural limits.
The UAE's population is 10 million; Saudi Arabia's, 36 million. Most AI engineers must be recruited from abroad. AI models are primarily purchased from the United States, and Chinese companies often participate in infrastructure construction. The result is simultaneous dependence on both superpowers' technology. Capital is abundant, but the source of capability is external.
G42 divested its China-linked assets under American pressure and aligned with the United States. The price of severing ties with China: entry into an exclusive club with access to the world's most powerful processors. The moment a hedging strategy converges into alignment.
Fast decision-making does not necessarily mean correct decision-making. When the Soviet Union poured vast resources into technological leapfrogging in the 1950s and 1960s, it outpaced the United States on short-term metrics. But because those investments never connected to a civilian innovation ecosystem, long-term technological competitiveness eroded from within. So long as the Gulf's AI investments depend on foreign talent and foreign technology, the same pattern can repeat.
Section F: Implications for Korean Readers — Displaced Nation or Discerning Nation?
Three Conditions
Comparing the four third players reveals a common pattern.
South Korea's HBM, India's talent and digital infrastructure, the EU's regulatory standard, the Middle East's capital. The forms differ, but the underlying structure is identical. Secure something indispensable, convert U.S.-China competition into leverage, and race against time.
Decomposing this common structure yields three conditions.
The first condition is irreplaceability. Securing a role that cannot be done without. South Korea's HBM, the Netherlands' EUV. India exporting DPI to 23 countries falls in this category as well. A nation that is irreplaceable is needed by both the United States and China, and from that need, bargaining power arises. Bargaining power is the ability to refuse a forced choice.
The second condition is bilateral utility. Being needed by both the United States and China. South Korea supplies HBM to NVIDIA while selling memory to the Chinese consumer market. India is FAANG's largest talent pipeline while chairing BRICS. Without bilateral utility, choice is imposed. The moment choice is imposed, leverage disappears.
The third condition is institutional agility. The capacity for institutional responses to keep pace with a rapidly shifting environment. South Korea enacting and enforcing the AI Basic Act in two years is the case in point. The institutional adaptation case study from the Industrial Revolution presented in Volume 1 took 64 years. Can this cycle be compressed in the AI era? Legislation, however, does not guarantee that institutions will function.
Three Clocks
Each condition has an expiration date. And those expiration dates are shrinking at different speeds.
The irreplaceability clock. China's effort to achieve HBM self-sufficiency is underway. Chinese firms including CXMT (ChangXin Memory Technologies) have entered HBM development. China's semiconductor equipment self-sufficiency rate stood at 13.6% as of 2024. Low. But the direction points up. Huawei Ascend 910C production is rising from 600,000 units in 2025 to 1.6 million in 2026. Whether it takes China five years or ten to fully self-supply HBM is unknown. But in a race where the direction is set, "when" is a question of probability, not "whether."
The bilateral utility clock. The annual license-renewal regime regenerates uncertainty every year. A 25% AI chip tariff applies to advanced AI chips from outside the U.S. supply chain. The deeper the U.S.-China technological divergence, the narrower the middle ground becomes. The space in which South Korea can supply both sides is structurally shrinking.
The institutional agility clock. Even after a bill passes, if enforcement capacity does not follow, institutions remain formalities. The AI Basic Act may be in force, but the problem of 38.8% of jobs exposed to high-risk automation cannot be solved by legislation alone.
Conditions for Being Displaced, Conditions for Being Discerning
Start with the conditions under which South Korea gets displaced.
If overdependence on a single industry (HBM) persists, and China's HBM self-sufficiency materializes, and South Korea fails to enter the design layer in next-generation memory architectures. If, under annual license-regime pressure, Korea rapidly loses the Chinese market while U.S. replacement demand fails to form quickly enough. If the displaced accumulate through the AI automation transition and institutional responses cannot match the speed of the shock.
The conditions under which South Korea becomes discerning look like this.
If it maintains continuity in the design layer from HBM to next-generation memory architectures and secures partial global competitiveness in the AI software ecosystem. If it proactively manages changes in the annual license regime, building trust capital with the United States, while confining the relationship with China to the cooperative space of legacy memory supply — avoiding abrupt severance. And if it reinvests HBM profits into developing the next lever — repeating "Germany's choice," the decision in the 1880s to reinvest manufacturing profits into chemical and electrical industry research.
At the 1900 Paris Exposition, a British inspection delegation recorded: "Germany is making better use of the science we invented than we are." The gap between invention and application became the seed of a hegemonic shift thirty years later. Stopping at having made HBM and designing the next technology layer on that foundation are not the same strategy.
No third player aspires to become one of the two empires. Their strategy is to become — just as the two empires need each other — something both empires need. But no third player believes this condition is permanent.
Closing Image: Within the Window
Back to Icheon.
January 2026. The cleanroom of the M16 fab. Finished HBM4 chips move to the packaging line. Epoxy molding, electrical testing, labeling. Packages that pass inspection are loaded onto carriers. Only chips that pass ship out.
These chips scatter in three directions.
Some head to the San Francisco Bay Area. They will be loaded into NVIDIA Blackwell racks at Stargate data centers and put to work training OpenAI's next model. Some go to Shenzhen AI labs. As long as export licenses hold, these are legacy memory and permitted-grade products. Some head to Abu Dhabi. They will be installed in the Phase 1 infrastructure of the G42 campus.
From one factory, three algorithmic empires begin.
What made this structure possible is technology. 3D stacking, high-bandwidth interconnects, process know-how at precision measured in tens of nanometers. Decisions Samsung Semiconductor made to keep investing during years of losses. Choices SK hynix made not to abandon HBM research. Decades of accumulation built this cleanroom.
But what permits this structure to continue is geopolitics. So long as the United States recognizes South Korea as a trusted partner in the Chip 4 alliance. So long as China calculates that it cannot run its domestic IT industry without Korean memory. So long as both conditions hold simultaneously.
Volume 1 Connection: The Expiration Date of Leverage
Leverage has an expiration date.
In 1769, Richard Arkwright patented the water frame. In 1833, Britain enacted the Factory Acts. Technology changed society, and society changed institutions — a process that took 64 years. How much this cycle compresses in the AI era remains unknown. But investing, choosing a career, founding a company, and shaping policy all happen within that cycle. That is the timeline the reader of this book operates in.
Add twenty years to 2026 and the year is 2046.
Within those twenty years: whether South Korea can build the design layer beyond HBM. Whether India's 12 million annual labor market entrants can learn to work alongside AI. Whether the EU's regulatory standard becomes a foundation of trust rather than a shackle on innovation — nurturing the companies that follow Mistral. Whether the Middle East's capital converts into genuine capability instead of a more expensive form of dependence.
These questions converge into one.
What will you do within the window before your leverage expires?
The final question from Volume 1 returns.
Are you among the displaced, or the discerning?
The epilogue adds one more.
And what about your country?
This book tracked the algorithm of two empires. But algorithms do not belong to empires alone. In the cleanroom in Icheon, the tech park in Bangalore, the factory in Veldhoven, the data center in Abu Dhabi — each runs its own algorithm. In a world where two empires compete, the third players turn that competition into their own computational resource. Hegemony is not an outcome; it is a condition. How to move atop that condition remains an open question.
〈 End 〉