Macro Inversor / IA Pitch Guion

Macro Inversor · research dossier for the next video · compiled 5 Oct 2026

Metcalfe e IA

AI is spreading faster than any technology on record. Network math explains the speed; history explains the risk: the technology can win while the people who overpaid for it lose.

The video's spine: Metcalfe's law is quadratic, not exponential. What is truly exponential in AI is compute and the collapsing price of machine thinking. The same n² story that sells adoption also inflated the 2000 bubble.

«La tecnología puede ganar y tú perder dinero.»

Get these three right first

The law you remembered is Metcalfe's law.
Bob Metcalfe (Ethernet, 3Com), early 1980s; named by George Gilder in 1993. Value grows with n². His own contrast (2006): “Moore's Law is exponential in time while Metcalfe's Law is quadratic in size.”
The “double exponential” is Kurzweil's.
“The Law of Accelerating Returns”, 7 Mar 2001: the rate of exponential growth itself grows. Contested by measured data.
The network law that is exponential in n is Reed's.
David P. Reed, 1999: groups among n people = 2ⁿ − n − 1. It fits real data worst of all four laws.

«Metcalfe no es exponencial: es cuadrático. Lo exponencial es el tiempo (Moore). Lo doble-exponencial es la tesis de Kurzweil.»

Every new node can talk to every node already there

Interactive · complete graph
12
66 possible links · n(n−1)/2
node 12 added 11 new links

Relative value with every constant set to 1: shapes only, real constants are unknown. Bars use a log scale (each tick = ×10⁶). Reed's exact group count is 2ⁿ − n − 1. “Realistic use” keeps about three links per node bright: Dunbar's ~150 friends, Zipf-weighted contacts, the n·log n reality.

Key numbers

The figures that carry the video

Eight verified figures, each with its source and date. Red marks the alarm figures.

Flags used on this page: Verifysecondary or regional press only; check before air Disputedsources disagree Pressmedia report, not company-confirmed Computedarithmetic by the researchers on cited data Projectionforecast or guidance

The laws

Seven laws, and which one is which

Network laws describe value as a function of size (n). Moore, Wright and Kurzweil describe change over time or over cumulative output. Mixing the two is where most “exponential” claims go wrong.

Also in the brief

Smaller laws worth one line each.

The math, drawn

Constants set to 1. These are shapes, not valuations.

Is AI a network-effect business?

Four different mechanisms. Only some of them are Metcalfe.

Agents talking to agents: n agents give n(n−1)/2 possible pairs (1,000 agents → 499,500 pairs). Three honest adjustments from A: most pairs are useless (Odlyzko, so think n log n for value); pairs are also coordination cost (Brooks); every link is attack surface (Beckstrom's law; Simon Willison's “lethal trifecta”, 16 Jun 2025).

The data · 2023–2026

Adoption, scaling and money, to scale

Ten charts drawn from the research series. Axes that span orders of magnitude use a log scale and say so. Hover a mark for the exact value; open “Data” under a chart for its table.

History rhymes

Right about the technology, wrong about the price

Railways, electricity, fibre and the dot-com boom all changed the world. Many of the people who financed them still lost money.

Where we might be: Carlota Perez

Technological Revolutions and Financial Capital (2002). Two readings, shown side by side.

Base rates and rules of thumb

How often booms come with bubbles, and what tends to follow.

Investor lens

How to think about it, not what to buy

A framework, not personalised investment advice. The video should say so on screen.

The AI value chain

From silicon to the consumer. Who captured the value in past platform shifts, and what 2026 looks like.

2000 vs 2026: the scorecard

Concentration is higher; multiples are about half; the risk moved from “price with no earnings” to “earnings that depend on a capex cycle that must keep accelerating”.

Risks, and the best counter-arguments

Each with the figure that makes it concrete.

Scenarios

Illustrative. The research assigns no probabilities; the bull / base / bust labels are framing only.

What the network math implies

From the laws brief (A §3.4) and the investing frameworks (C §5).

The inflation link

Bridge to the previous video: AI cheapens thinking and makes electricity dearer.

Spain angle

Aragón, the grid queue, power prices and how fast Spanish firms are adopting.

Talking points

Ready-to-say lines in Spanish, each backed by a figure on this page.

    Angles & titles

    Twelve titles with their hooks

    From the investing brief (C §6). Its recommendation for search intent: T3 or T2 as the main title, with T1 as the cold-open line.

    Query phrases to put in the title and the first 30 seconds:

    burbuja de la IANvidia Ciscoley de Metcalfeinvertir en inteligencia artificialcentros de datos en España / Aragón

    Video structure · 9:30

    Six beats, cold open to close

    C's 8–10 minute outline, refined with the laws (A) and adoption data (B). Timestamps are C's; the strip below is drawn to scale.

    Animation ideas

    Forty-seven visuals, filterable

    All 40 ideas from C plus 7 from A (two of A's merge into C's first two). Dark cards are the top 10. Difficulty is our estimate for the HTML render engine.

    Act
    Type
    Show

    Sources

    Every link in the three briefs

    Grouped by topic. Dates are publication dates where the brief gives one; otherwise the access date (5 Oct 2026). Sources C names without a link are listed last.