Video proposal: «Nvidia no es Cisco… o sí: los datos de 2000 vs 2026»

Content-strategy pitch, 2026-10-06. All figures come from A_laws.md, B_ai_data.md and C_investing.md (sources there).

The pitch in one line

Everyone asks "is AI a bubble?" and everyone answers with opinions. This video answers with a scorecard: the same metrics for 2000 and 2026, side by side. It ends with a verdict that is neither "it's a bubble" nor "buy everything".

Why this video, why now

• The question your audience already searches. It targets "burbuja de la IA", "Nvidia" and "invertir en inteligencia artificial". The title is a question with a twist, and people click to get the verdict.
• A story that sells itself: Cisco.
  • In March 2000 Cisco was the world's most valuable company.
  • It never went bust and it tripled its sales.
  • The stock still took 25 years to get back: it only set a new record on 10 Dec 2025.
• A sequel to the inflation video. AI makes thinking cheaper and electricity more expensive. Returning viewers get a continuation; new viewers get an entry point.
• Built for the documentary style. Lines drawing on, archive photos from 2000 and 1845, a 2000-vs-2026 scorecard and a map of Aragón.

The promise (first 30 s)

«En marzo del 2000, Cisco era la empresa más valiosa del mundo. Nunca quebró. Triplicó sus ventas. Y aun así, quien compró ese día tardó 25 años en recuperar su dinero. Hoy Nvidia vale 5,8 billones de dólares. ¿Es la nueva Cisco?»

The 6 aha moments

1. Great company, terrible investment.
   • Cisco FY2000 → FY2025: revenue $18.9bn → $56.7bn, net profit $2.7bn → $10.5bn.
   • The stock fell 89% and stayed below its 2000 peak for 25 years.
   • Lesson: the technology can win while the investor loses.
2. The formula that inflated 2000 is back.
   • Metcalfe's law (value grows with users²) was used to justify dot-com valuations. Metcalfe himself admits it helped inflate the bubble.
   • Its critics argue real value grows like n·log n. At 1 million users, the two formulas differ by about 72,000×.
   • It also corrects a common myth: Metcalfe's law is quadratic, not exponential.
3. Nvidia is ~4× cheaper than Cisco, but ~3× bigger.
   • Forward P/E: 26x vs 102x (same Goldman measure).
   • Profit margin: Nvidia keeps ~64 cents of every sales dollar as profit; Cisco kept 14 (own calculation from filings).
   • Size: Nvidia is worth 17.7% of US GDP. Cisco at its peak was 5.4%.
4. Nvidia's sales are someone else's spending (the twist almost nobody explains).
   • Big Tech capex: $416bn (2025) → $799bn (2026F) → $1.2tn (2028F).
   • Growth slows: +92% → +37% → +10%.
   • Spending can keep rising and the supplier can still fall. That is what happened to Cisco in 2001.
5. You already bet on AI without choosing to.
   • 53% of the S&P 500 is AI-linked industries.
   • The top 10 stocks are 41% of the index, against ~27% at the 2000 peak.
   • If you own an index fund, this video is about your money.
6. Bubbles are the rule, not the exception.
   • 73% of 51 major innovations since 1825 came with a stock-market bubble. The bubbles were bigger when the technology had network effects.
   • Darwin and the Brontë sisters lost money in the Railway Mania (shares −64%), and the railway still changed the world.

The Spain twist (cliffhanger at ~6:30)

«Hay un sitio donde la burbuja de la IA ya se ve… y está en España.» - Red Eléctrica has granted data centres 6,329 MW of grid access. None of it is in service yet. - AWS has committed €33.7bn in Aragón. - About 22 GW has been requested (press figure). - Bridge to the inflation video: Powell says the data-centre build-out is "probably pushing inflation up". US electricity CPI is +5.9% against 4.2% headline inflation.

The verdict (payoff)

«Nvidia no es Cisco en precio. Pero sí lo es en dependencia.» - Valuation multiples are about half of 2000's, and the profits are real. - Concentration is higher than in 2000, and earnings depend on a spending cycle that has to keep accelerating. - Close with four scenarios: - Ferrocarril: the technology is right, the price is wrong, and the builders crash. - Dinamo: gains arrive slowly, so prices stall for years while profits catch up. - Internet sin crash: profits grow into the valuations. - Commodity: AI gets cheap, users keep the value and margins shrink. - Then list what to watch: - capex growth - the debt share of capex (9% → 32%) - power - AI revenue against Sequoia's "$3tn question" - «Esto no es una recomendación de inversión».

Structure (~9:30)

Signature visuals

• Cisco ×21 vs Nvidia ×13.8 overlay, each starting from its era's catalyst (Netscape IPO / ChatGPT), with the caption «un gráfico no predice nada».
• Cisco's flat price line against 25 years of rising revenue bars.
• A P/E thermometer, and bubbles sized by share of GDP.
• A capex staircase with the growth line bending down.
• A Spain grid funnel: 22 GW requested → 6.3 GW granted → 0 MW in service.

Before recording

• Refresh Nvidia's market cap and P/E on the recording day.
• The 22 GW figure comes from press only.