Nvidia

When Jensen Huang Says NVIDIA’s Market Share in China Dropped from 95% to 0 — What Key Insights Are Hidden Behind It? | [Jingwei Low-Key Share]

Honghao Wang

— 4 min read
When Jensen Huang Says NVIDIA’s Market Share in China Dropped from 95% to 0 — What Key Insights Are Hidden Behind It? | [Jingwei Low-Key Share]

Building an “Economics of the AI World”

image

On October 16, 2025, a private conversation video from Citadel Securities in the U.S. was leaked. Nvidia CEO Jensen Huang, in his iconic black leather jacket, addressed Wall Street elites managing trillions—sharing a 30-year computing power journey and revealing a stunning fact: due to U.S. export controls, Nvidia’s market share in China’s high-end chip sector fell from 95% to 0%.

image

The scope and density of the discussion resembled a modern Industrial Revolution retrospective. Huang covered:

  • His 1993 contrarian pivot to GPU development
  • Creation of the CUDA language that turned GPUs into general-purpose computing platforms
  • The 2012 milestone when GPU computing powered AlexNet, sparking the AI revolution
  • His core strategy: the “AI factory” — a system that produces intelligence, not information, akin to steam engine factories generating power, but integrated from chips to algorithms.

His prophetic view: “The future of computation is 100% generated.”

He forecasts a generative era—AI becoming an enterprise’s “digital workforce” and CIOs evolving into “AI HR,” with a warning: blocking China, home to half of global AI researchers, is a strategic mistake.

This wasn’t a status report—it was a roadmap for the AI-powered economy.

---

Highlights of Huang’s Speech

Event Context

  • October 6 — Citadel Securities, New York, private talk
  • Attendees: Wall Street leaders controlling massive global capital flows
  • Format: A decade-spanning narrative—from graphics cards to accelerated computing to AI factories

Tone: Less business report, more philosophical reflection on computing power as civilization infrastructure.

Key takeaway:

> The future of computation is 100% generated.

---

Back to 1993 — Pre-Internet World

At the time, everyone invested in CPUs under Moore’s Law. Huang identified the limitation:

> General-purpose tech often isn’t suited to extremely hard problems.

He chose GPUs—a specialist to the CPU’s general worker—aimed at solving hard, specialized tasks.

---

Nvidia’s Struggle in the 1990s

Challenge: No clear market, standard, or ecosystem.

Solution: Create a market—collaborate with game companies like Electronic Arts to build 3D worlds, proving GPU value.

Lesson:

  • If no market exists, invent one.
  • Build the ecosystem before the customers arrive.

---

The CUDA Turning Point

In the 2000s, Nvidia launched CUDA, transforming GPUs into general-purpose computing platforms.

Before CUDA:

  • GPUs rendered graphics only
  • Scientists hacked GPUs for parallel computing

After CUDA:

  • Programmers could directly write for GPUs
  • Hardware gained an “interface to thought”

---

2012 — "Adam Moment"

Researchers Geoffrey Hinton, Andrew Ng, and Yann LeCun hit computing power bottlenecks.

Nvidia built cuDNN, a library accelerating neural network training.

Result: AlexNet dominated image recognition competitions; AI could “run.”

Huang realized:

> A neural network can learn any function — simulate any intelligence.

---

The AI Factory Concept

Core Strategy:

  • Unlike data centers storing information, AI factories produce intelligence
  • Each training, generation, inference = manufacturing step
  • Full-stack integration: chips, networks, servers, software, algorithms

Industrial Revolution analogy:

  • Steam engines → mechanical power
  • AI factories → cognitive power

---

AI as Digital Workforce

Prediction: Companies will have human + AI employees.

CIO role: HR for AI—responsible for onboarding, training, and managing AI workforce.

Implication:

  • AI needs to learn company culture, knowledge, processes
  • Businesses must manage AI like human talent
  • Marks an organizational revolution

---

The China Question

Huang stated:

> Market share in China: from 95% to 0% due to export controls.

Warning:

  • Losing the second-largest computing market is bad policy
  • Half of global AI researchers are in China
  • Blocking them = strategic mistake

His subtle message: Technology sanctions = exiting the AI industrial order.

---

100% Generative Future

Traditional computation: retrieval-based

AI computation: generative — it creates, not simply finds.

Example: Fully generated video (Sora) and search (Perplexity).

Philosophical core:

  • Industrial Revolution liberated power
  • Generative Revolution liberates imagination

---

Why This Speech Matters

Not Technical — It's Strategic

  • A computational worldview declaration
  • Computing = new productivity
  • Chips = steel; AI factories = steel mills

Citadel Securities Connection

  • Audience: movers of global capital
  • Message: invest in AI factories, generative systems, and computing power

Policy Narrative

  • Offers policymakers an off-ramp to ease export controls without admitting defeat
  • Frames openness as strategic necessity

---

Emotional Tone

Huang speaks as if describing a new city, transforming tech into belief.

Goal: Mobilize capital across four audiences:

  • Capital — investment in computing power ecosystems
  • Policy — nudging a shift towards openness
  • Industry — signaling a strong AI future
  • Public — reshaping human-AI collaboration mindset

---

AiToEarn — A Creator’s AI Factory

Platforms like AiToEarn官网 embody AI factory logic—providing:

  • Open-source, global AI content monetization
  • Cross-platform publishing to Douyin, Kwai, YouTube, WeChat, Bilibili, Xiaohongshu, Facebook, Instagram, LinkedIn, Threads, X, Pinterest
  • AI model rankings (AI模型排名)
  • Integrated generation tools + analytics for creators

---

Reference:

  • Nvidia’s Jensen Huang on AI & the Next Frontier of Growth, Oct 16, 2025 — YouTube Link

---

Additional related talks:

image

Read the Original | Open in WeChat

---

For creators and analysts, AI-assisted multi-platform publishing tools—like AiToEarn—offer a practical application of Huang’s AI factory concept: produce intelligence, distribute it widely, monetize it efficiently.

---

Would you like me to prepare an executive summary version of this improved Markdown for quick reading? It could be condensed into 15 bullet points while keeping the critical insights.

Read more