La revolución de la IA es más GRANDE de lo que pensábamos: NVIDIA crecerá otro 70% @ElInformeK

Company / Market Context (AI Capex Cycle)

The video centers on NVIDIA (NVDA) as the dominant supplier behind the AI infrastructure buildout, arguing the AI spending cycle can be bigger than expected due to:

• Supply being the bottleneck (manufacturing capacity), not (just) demand
• A potential shift from “chat” AI to AI agents, implying much higher compute demand

---

Key NVIDIA (NVDA) Financial Claims / Valuation Points

Revenue growth and near-term expectations

• Latest-quarter revenue: $96.2B
  • +106% YoY (“doubled income in 12 months”)
• Next-quarter revenue expectation: $108B
• Scale claim: NVDA expected to pass $100B revenue in a quarter (described as the first time in its history)
• End-of-year revenue level (claim): around $400B

Growth forecast (framed as supply-limited)

• Next fiscal year growth forecast: ~70% (“supply-limited”)
  • Interpreted timeline: fiscal year 2027 (narrator framing), aligned with calendar year 2027
  • Implied revenue range after growth: ~$680B–$700B

Profit / EPS scenario (explicit estimates)

• “Profit above $360B”
• EPS ~ $16
• Implied multiple: “< 15x expected earnings” at the “current share price”

Longer-term, very optimistic scenario

• Revenue target: $1T in fiscal 2029 (calendar 2028, per subtitles)
• EPS: could “easily exceed $3 in 2 years”
• Share price reference: “around 230 euros” (likely regional reference; not necessarily an official USD valuation)

Market pricing vs expectations

• Stock performance: >20% up over the last year, but “far from its growth rate”
• Valuation claim: ~15x expected profits (next year)
• Narrator’s framing: “has the market gone crazy?”

---

Guidance / Demand vs. Supply Framework (Methodology Called Out)

The subtitles emphasize a recurring interpretation of NVDA’s guidance:

• Guidance is “supply-limited,” not demand-limited
  • NVDA expects ~70% revenue growth because they can manufacture enough GPUs/inputs
  • If capacity could be expanded, growth could be higher than the stated 70%

The narrative uses NVDA’s forecast as evidence that:

• AI demand is overwhelming, and
• supply constraints persist, making the growth potentially more durable

---

Macro / Sector Spending Figures (AI Infrastructure Race)

Escalating capex by major tech

• Major companies mentioned: Microsoft, Amazon, Google, Meta, Oracle
• Spending framing: could rise from ~$800B in 2026 to $1.3T next year (as stated)

Customer deployment examples

• Amazon: committed to 2 million additional Nvidia GPUs
• OpenAI: secured ~12 GB of Nvidia capacity by 2030 (subtitles wording unclear, but clearly references long-dated capacity access)

---

“Ecosystem Broadening” and Customer Concentration Shift

The narrator argues NVDA is reducing reliance on only a few hyperscalers by expanding toward:

• AI-specialized “clouds”
• Governments seeking sovereign cloud
• Industry players, etc.

Additional stated claims:

• A “block” generating >$40B per quarter for NVDA
• Growth for that segment: >130% annually
• Expected contribution: about half of NVDA’s data center business “today” (as stated)

---

Business Model / Margin Drivers (Why Revenue Can Outgrow Chip Unit Sales)

The subtitles describe a framework for how NVDA can grow faster than chip counts:

• From selling components to selling “full systems”
  • Hopper / Blackwell / Rubin-era (verbatim phrasing includes “Vera Rubín”)
  • “~$18B per gigawatt of data centers” (Hopper)
  • “~$25B per gigawatt” (Blackwell)
  • “~$40B per gigawatt” (Rubin generation)

Why it rises

• NVDA increasingly sells GPUs + embedded processors + networking + complete systems + software
• The video also claims NVDA earns more in networking alone than Cisco (company referenced, without clear quantified figures)

---

Supply-Chain Dominance and Working-Capital / Cash Flow Claims

• Supplier commitments: $119B → $279B (over 3 months, per subtitles)
  • Claims: commitments are mainly for memory, locking in supply/pricing into next year
• Payment/financing dynamics (“circular finance” risk framing):
  • NVDA collects payments in about 60 days (vs 45 days previously)
• Last-quarter profitability/cash flow (as stated):
  • Free cash flow: ~$21B
  • Profit: >$50B

---

Risk Discussion: “Circular Financing” / Supplier-Backed Buildout

The narrator highlights a potential structural risk:

• NVDA may be enabling customers using its balance sheet (supplier lending/guarantees)
• This could create circular financing dynamics:
  • supplier finances customers → customers buy more → orders return to supplier

Caution scenario described

• If major data centers/AI buyers stop monetizing:
  • NVDA could face lower sales and increased exposure to unpaid financing/loans

The risk is framed as something that “needs to be monitored.”

---

“AI Agents” as the Next Compute Demand Multiplier

Agents vs. chat

• Chat: user asks, system responds, interaction ends
• Agent: behaves like a digital employee (browser/email/calendar) with continuous task execution

Compute multiplier

• Agent compute demand vs a person using AI: ~15x to 100x greater (depends on problem type)
• One person can create multiple agents → multiplicative effect

Investment implication

• Compute demand could skyrocket while supply cannot keep up for a long time

---

Additional Market Data Points (Examples + Infrastructure Estimate)

Meta app example (demand proxy)

• Meta launched a “personal user” (agent-like product)
• In 10 days, it became the most downloaded app in the US, ahead of GPT chat (per subtitles)

Token consumption claim

• Active users may consume ~300x more tokens than simple chat users

Brookings Institution study (US infrastructure spend estimate)

• US AI infrastructure investment could reach $10.3T between 2025 and 2032
• Framing: 3.6% of the payback period each year
• Positioned as potentially the largest infrastructure investment in history
• Implied inputs: GPUs, CPUs, memory, copper, electricity, etc.

---

Memory Bottleneck and Related Holdings Mentioned

The subtitles argue HBM memory is the next bottleneck:

• Memory is ~one-third of the cost of an AI server rack (as stated)
• Only three companies can manufacture HBM (as stated)

Companies/holdings cited

• SK hynix (spelled in subtitles as SK Hinx)
• Samsung

Narrator framing included:

• SK hynix described as a “latest major addition” after a market correction

---

Portfolio / Fund References (Mentioned, Not Executed)

The narrator references:

• “Baltia Global”
• “Visual Factory investment fund”
• Availability “at Renta 4” (broker/platform)

A further implied theme:

• “Risk the ecosystem becomes profitable later,” implying possible “bubbles and bursts.”

---

Disclosures / Disclaimers

• No explicit “not financial advice” disclaimer appears in the provided subtitle text.
• Promotional content is mentioned for a platform (OD / “Odulu”) including pricing and trial info, but it is not a finance disclaimer.

---

Tickers / Assets / Instruments / Sectors Mentioned

Ticker(s)

• NVDA (NVIDIA) is explicitly referenced by name; no alternate ticker symbol is shown.

Companies used as infrastructure/demand examples

• Microsoft, Amazon, Google, Meta, Oracle, Cisco, OpenAI, Anthropic, Intel, SpaceX

Memory-related

• SK hynix, Samsung

Financial institutions / alliance capital raising

• BlackRock, KKR, Goldman Sachs

Cryptic platform/fund names (no tickers provided)

• Visual Factory investment fund, Baltia Global, Renta 4

Infrastructure / geography / projects

• “megacampus in Ohio,” data centers, sovereign cloud

---

Key Presenters / Sources Mentioned

• Jensen Huang (CEO and founder of NVIDIA; referred to as “Jansen Juan” in subtitles)
• Brookings Institution (macro/market study source)

Other entities referenced (not as presenters) include Renta 4, BlackRock, KKR, Goldman Sachs, Meta, White House, OpenAI, etc.

Note: The subtitles include promotional segments unrelated to investing; the summary above focuses only on the finance/markets content.