The Loop and the Black Hole I

Semiconductors, AI, and the Arithmetic of Exuberance

Preamble: The Invisible Geography of the Loop

In 2023, while Nvidia crossed the one trillion dollars milestone for the first time, in Kolwezi, Democratic Republic of Congo, artisanal miners extracted cobalt — a critical mineral in the chip and battery supply chain—by ​​hand, in the open pit, for wages of two to three dollars a day. The cobalt extracted under these conditions travels the supply chain to Asia, is incorporated into devices assembled in Penang or Johor Bahru, and reaches the end user as part of a GPU supporting a stock market multiple of 41.7 times the sector's gross value added. No valuation model accounts for this journey. The value created at the beginning of the chain is not discounted—it is simply ignored.

This is the first layer of social transmission that this article analytically addresses: the extraction of value from territories that lack access to the valorization that this value sustains. In the lithium triangle—Argentina, Bolivia, Chile—the indigenous communities that live above the largest reserves of the most sought-after mineral in the digital transition face increasing drought, aquifer contamination, and forced relocations to make way for extraction projects whose contracts benefit multinational corporations listed on stock exchanges whose capitalization grows in direct proportion to the demand for semiconductors and electric vehicles. The cycle closes far from them.

The second layer is that of the invisible work at the bottom of the value chain. Nvidia's gross margins exceed 70%. In chip packaging factories in Penang and Kuala Lumpur, the average worker's pay is between US$300 and US$500 per month. The worker who performs electrical continuity tests on the latest generation chips will never, in her entire professional life, touch a company's RSU (Restricted Stock Unit) whose product passes through her hands. She doesn't have access to the mechanism described in this article—the one by which market cap pays salaries with overvalued stocks. What you receive is denominated in ringgit, subject to currency devaluation induced by monetary wars between superpowers, and doesn't grow when Nvidia hits quarterly records.

The third layer is that of silent displacement. In the Philippines, approximately 1.3 million people work in customer service centers—the largest concentration of contact centers of the world, based on fluent English and time zones convenient for Western markets, and one of the archipelago's main sources of remittances. The language models trained on the GPUs that Filipino workers indirectly helped manufacture are progressively replacing these functions.

The structural irony is complete: the chain that begins with cobalt in the Congo, passes through the work of Asian factories, and ends in an AI model valued at hundreds of billions of dollars, returns to the starting point a process of smooth but persistent deindustrialization that is not compensated for by any transition policy worthy of the name.

The fourth layer is monetary and structural. The decade of near-zero interest rates after 2008 — deepened by the pandemic QE of 2020–2021 —It created the liquidity conditions that made the technological loop possible: with real interest rates at zero, capital migrated en masse to long-term assets, compressing the discount rate and inflating multiples. The loop didn't create this environment—it was its main beneficiary. When the 2022–2023 inflation forced central banks to reverse the cycle, its causes were different: the abrupt rebound in post-COVID demand, persistent disruptions in global supply chains, and the energy shock of the Russian invasion of Ukraine. The loop was a privileged spectator of this storm, not its author. What matters is the transmission mechanism that followed: the rise in US interest rates, regardless of its causes, withdrew liquidity from the global system with the speed that digital financial integration allows. In 2022 and 2023, countries like Ghana, Sri Lanka, and Pakistan faced sovereign debt crises not because the loop had directly impoverished them, but because the architecture of the international monetary system transforms any tightening by the Fed into a vacuum for capital from emerging markets—regardless of what caused that tightening. It is a structural asymmetry: the peripheries pay the adjustment costs for disruptions they did not generate. The link to the loop is indirect but real: the same liquidity that financed Nvidia's valuations had previously flowed into emerging market sovereign debt in search of yield. When the tide receded, it receded from all sides at once. In the near future, the energy pressure from data centers introduces a second contagion route—the electricity consumption of the AI ​​industry doubled between 2020 and 2024 and is expected to double again by 2026—which will compete with domestic electrification in economies where grid access is not yet universal. This is, for now, an emerging tension, not an established cause. But its direction is unmistakable.

The following article does not start from these geographies—it starts from the numbers. It builds its analysis from market capitalization, added value, and payroll indices in six American industries; it explores the arithmetic that underpins a 200-fold Price/Earnings ratio in semiconductors; it identifies the closed-loop system that connects Nvidia to Big Tech to artificial intelligence and back to Nvidia; and it unfolds three disruption scenarios before attempting the wide-angle view that integrates supply chains, currency wars, escalating sovereign debt, climate stress, and the risk of state disintegration.

What this preamble aims to add is a coordinate that financial analysis rarely includes in its axes: the geographical distribution of consequences when the promised value doesn't arrive, or arrives too late for those who were at the beginning of the chain to sustain it. The market doesn't register Kolwezi. Gordon's model doesn't discount Penang. The Fed's discount rate doesn't include Manila. But these cities are in the same system. And when the loop falters, they will feel the shock before any index reflects it.

There is a question that financial analysts rarely formulate clearly: For today's price to make sense, what kind of economy needs to be built tomorrow? - Schumpeter would ask. In the semiconductor and artificial intelligence sectors, the answer to this question is so demanding that it requires a new category of analysis—not of markets, but of physics. The market is literally betting on building a reality different from the one that exists.

The numbers that follow are not opinions. They are arithmetic. And what they reveal is that current stock market capitalizations do not reflect generated value—they reflect a promise of value that, in order to materialize, requires a reorganization of the global economy unprecedented in the history of capitalism. The central question is not whether this promise will be fulfilled. It is to understand what mechanism sustains it, what happens when it falters, and why central banks seem to be held hostage by the fiction they are supposed to regulate.

Methodological note:

¹ Market capitalization figures refer to the leading US-listed companies in each sector (mid-2024): Nvidia, Broadcom, AMD, and Intel for semiconductors; Alphabet, Meta, Microsoft (partially), and Netflix for internet/software; Eli Lilly, Pfizer, and Merck for pharmaceuticals; ExxonMobil, Chevron, and NextEra for energy; ADM, Coca-Cola, PepsiCo, and Tyson for agri-food; UPS, Union Pacific, and Delta for transportation. Gross Value Added (GVA) and wage figures are derived from the Bureau of Economic Analysis (US, 2022, current prices), using the corresponding NAICS codes (3344 for semiconductors; 519+5112+5415 for internet/software; 3254 for pharmaceuticals; 211+324+22 for energy; 11+311+312 for agri-food; 48-49 for transportation). All figures are rounded.