> ## Content Index > Fetch the complete content index at: https://atopicosphere.ghost.io/llms.txt > Use this file to discover other available public pages before exploring further. # The Loop and the Black Hole II - URL: https://atopicosphere.ghost.io/en/articles/the-loop-and-the-black-hole-ii/ - Published: 2026-07-23T18:42:00.000Z - Updated: 2026-10-07T18:31:59.000Z - Description: Published on July 23, 2026 - Author: amorsantos - Tags: #en, #artigo, #news ## **Semiconductors, AI, and the Arithmetic of Exuberance** --- ## **I. The Arithmetic of the Promise** To calibrate the problem, we used the United States as a benchmark, as it has the most complete and accessible national accounts and market capitalization figures. The following table compares six non-financial industries according to three indicators:**MCap - Market Capitalization** (market value of listed companies) **GVA - Gross Value Added** (real contribution to GDP excluding financial intermediation) and **SAL - Salary Volume** (total remuneration, including social security contributions). The figures are approximate, based on 2022–2024 data, and the ratios are derived directly from these estimates.¹ | INDUSTRY (USA) | MCap (B$) | GVA (B$) | SAL (B$) | MCap/GVA | SAL/GVA | MCap/SAL | | ------------------- | --------- | -------- | -------- | -------- | ------- | -------- | | semiconductors | 5 000 | 120 | 86 | 41,7× | 72% | 58,1× | | Internet & Software | 9 000 | 800 | 557 | 11,3× | 70% | 16,2× | | Pharmaceutical | 2 000 | 176 | 93 | 11,4× | 53% | 21,5× | | Energy | 3 000 | 736 | 197 | 4,1× | 27% | 15,2× | | Agri-food | 1 000 | 580 | 164 | 1,7× | 28% | 6,1× | | Logistics | 1 200 | 958 | 461 | 1,3× | 48% | 2,6× | The three indices reveal distinct dimensions of the same phenomenon. The **MCap/VAB ratio** measures how many times the stock market value exceeds the real contribution to GDP. In semiconductors, the multiple reaches 41.7×; in software and internet, 11.3×. In industries intensive in physical capital — energy, agri-food, transport — the same ratio oscillates between 1.3 and 4.1×. The latter are worth what they produce. The former are worth what they promise to produce. The **SAL/VAB ratio** maps the functional distribution of income. In semiconductors and software, wages absorb \~70% of the value generated, reflecting the intensity of human capital. In energy and agri-food, this share falls to 27–28%: most of the GVA is captured in the form of profits, rents, and depreciation of physical capital. The **MCap/SAL ratio** is the most revealing example. For every dollar paid in wages in the semiconductor industry, the market assigns $58 in value to the company. In transportation, that same dollar in wages is worth only $2.6 in the market. This difference is not merely quantitative—it's civilizational. It separates industries of *ideas* from the industries of *things*. Regarding artificial intelligence: it doesn't exist as an autonomous category in national accounts. It's a cross-cutting technology whose impact is diluted across semiconductors (hardware) and software (platforms and services). If it were possible to isolate a "pure AI" sector, the MCap/GVA multiple would be even more extreme—because companies like Nvidia or large cloud providers are valued with a growth premium that reflects not the present, but an implicit certainty about the future. --- ## **II. Market Cap as a Wage Currency** There is an interpretation of the figures above that rarely appears in financial literature: in sectors with higher multiples, market cap is not merely reflecting value—***it is paying for it.*** A significant portion of compensation in the semiconductor and internet sectors consists of stock rather than cash. For a senior engineer at Nvidia or Google, the compensation package might comprise 50% base cash salary and 50% RSUs (Restricted Stock Units). The value of these shares depends directly on market capitalization. A high market cap literally funds a substantial share of salaries, because the company issues securities with a value dictated by the share price. > **The 58.1x ratio (market cap to personnel costs in semiconductors) means that for every dollar reported as a personnel cost—including the fair value of shares granted—the market assigns $58 in value to the company. The "currency" used to pay talent is extraordinarily efficient: a fraction of the equity funds a payroll that a company with standard valuation multiples could not sustain using cash.** Yet this efficiency drives a self-reinforcing cycle—one that contains within itself the seed of its own disruption. The logic follows this chain: the market anticipates massive future profits, inflating market capitalization; this high valuation allows current salaries to be paid using high-priced stock; these salaries attract and retain the talent that will generate those future profits; and as those profits begin to materialize, they validate the narrative and further bolster market capitalization. ***It is a self-feeding cycle that functions only as long as growth expectations remain intact.*** If the multiple contracts—for example, if semiconductors go from 41.7x GVA to 10x—the stock value plummets. The compensation package loses purchasing power overnight. The company loses its ability to recruit and retain talent. Since salaries represent 72% of GVA in semiconductors, the company is viscerally dependent on these people. A stock market correction very quickly turns into a human resources crisis. There is no buffer. --- ## **III. The Equation that the Market Demands** The 41.7x VAB multiple for semiconductors requires translation: for the current market cap of $5 trillion to make sense, what future earnings need to be discounted? Using approximate data from the US industry: the sector's GVA is approximately $120 billion; after deducting 72% for salaries, the gross operating surplus is approximately $34 billion; the estimated net profit, after depreciation and taxes, is approximately $25 billion (net margin \~21%). **Implied Price/Earnings ratio* is from *5 T / 25 B = 200×*.* A P/E ratio of 200 is not sustainable with moderate growth. Using the Gordon model (perpetual growth) with a discount rate of 10%: > ***P/E = 1 / (r – g) → 200 = 1 / (0,10 – g) → g = 9,5%*** Profits would have to grow by 9.5% **indefinitely** A more realistic two-phase model imposes even more concrete requirements: - Current global net profit (USA + Asia): \~$60 billion - Phase 1 (20 years): 20% annual growth - Phase 2 (terminal): 5% growth - Discount rate: 10% At the end of 20 years, the annual profits would have to be: 60 × (1.20)²⁰ ≈ 2,3 trillion dollars.In two decades, this industry would have to generate profits **38 times superior** to those of today. | Indicator | Current value (2024) | Required Value (2044) | | ------------------------------ | -------------------- | --------------------- | | Global "semis" Revenue | \~0,6 T | \~9,2 T - (×15) | | Global GDP (nominal) | \~105 T | \~190 T - (+3%/year) | | Share of "semis" in global GDP | +0,6% | +4,8% | The isolated percentage wouldn't seem impossible. But semiconductor revenue doesn't exist in a vacuum: it requires downstream industries—AI, cloud, automotive, IoT—to generate even greater added value. Conservatively estimating that every dollar of chip revenue requires three dollars of value generated in the industries that incorporate them, the value chain implied in the current market cap would imply, by 2044, an ecosystem of \~$28 trillion of GVA— approximately 15% of the projected global GDP. ***To put the scale into context: the entire global manufacturing industry today represents \~16% of GDP*.** The financial and insurance sector, \~7%; health and education, \~10%. > **Investors are betting that the "semiconductors + AI + automation" complex will replace and absorb a share of economic activity comparable to that of the entire global manufacturing industry. This requires large-scale Schumpeterian creative destruction: automated administrative and legal services, transport fleets replaced by autonomous vehicles, and entirely robotized production chains with a drastic reduction in the wage component.** The arithmetic may add up—but the political, social, and temporal cost of that equation could become unsustainable for the investors financing it today. --- ### **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. ---