Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
InfrastructureInsight · On RadarGuardian→ original

Nvidia Shed 13% After Huang's Exit: What Holds Up the AI Market

Nvidia shares fell 13% at the start of trading after Jensen Huang stepped down as CEO, the Guardian reports. The company rests not only on its GPUs but also on Huang's personal ties with governments, TSMC, and AI labs. No one yet knows how those arrangements will hold up under new leadership, since the contracts are closed. Teams that depend on Nvidia's chips should plan a fallback in advance.

AI-processed from Guardian; edited by Hamidun News
Nvidia Shed 13% After Huang's Exit: What Holds Up the AI Market
Source: Guardian. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceGuardian ↗
Editor-in-Chief
Key metric #1Nvidia promised South Korea 260,000 chips, but timing and payment have not been disclosed
Key metric #2TSMC manufactures nearly all AI chips, and the yield on the most advanced ones is about 60%
Key metric #3Through Megaspeed, $7.5 billion in chips reportedly went to China

When the head of the world's second most valuable company steps down, the shares fall. That part is expected. In Nvidia's case, though, the question runs deeper: almost the entire industry depends on its chips, from training models like ChatGPT and Claude to running them day to day. After Jensen Huang's resignation, which the Guardian reports, Nvidia's stock lost 13% at the start of trading. The paper tries to work out what was actually holding this market up and what could give way. Note that, according to the paper, the resignation is rooted in allegations about personal relationships, which remain the subject of press reporting and claims.

Not Just Chips

On paper, Nvidia's value is simple to explain: the company designs graphics processing units (GPUs), without which modern AI cannot be trained or run. But, as the Guardian notes, the chips themselves are replaceable. Microsoft, Meta, and Google build their own accelerators, and Google's TPUs, by the paper's assessment, are in some cases no worse than Nvidia's solutions and for certain tasks are better.

What makes the company irreplaceable is its network of relationships, much of which Huang built personally. In 2025 he met with Mohammed bin Salman, Keir Starmer, and Emmanuel Macron, arguing for "sovereign AI" built on Nvidia chips. The meetings were followed by supply announcements. Only the terms of these deals were almost always left vague. South Korea, for example, was apparently promised 260,000 chips, but when they would arrive, who would pay for them, and on what terms was unclear.

The AI "Central Bank" and Circular Money

The second pillar is customer financing. Nvidia has backed so-called neoclouds: companies that buy its chips, install them in data centers, and rent them out to the largest AI companies. This supported demand and moved the risks of data center construction onto Nvidia's own balance sheet.

In addition, Nvidia invested in startups that could become its customers and offered large sums to OpenAI to buy its own chips. All of this is measured in hundreds of billions of dollars. Nvidia's money, in the Guardian's phrasing, keeps circulating again and again among a limited set of accounts. That is why the company is called the "central bank of AI."

A simple analogy: a supplier lends money to a buyer so the buyer can purchase from that supplier. As long as everyone keeps paying and growing, the arrangement works. If confidence falls, demand itself comes into question.

TSMC and the Dark Spots

The third pillar is Taiwan's TSMC, which makes the vast majority of AI chips and has no real competitors. The process is extraordinarily complex: circuit lines a few nanometers wide are patterned with lasers in extreme ultraviolet. Even after two decades of refinement, the yield on the most advanced chips is about 60%, meaning 40% of output is defective. According to legend, Huang wrote a personal letter to TSMC's chairman as early as 1997. Such trust, the Guardian observes, is something competitors cannot replicate.

It is almost impossible to say whether anything would change without Huang: contracts are closed, export records are tangled, and the queue for chips, which is known to stretch for years, is opaque. Who is in it and in what order is not known.

A telling case is Singapore's Megaspeed, which received many top-tier GPUs while others waited. According to the paper's data, its head, Alice Huang, reportedly had a relationship with Nvidia's CEO and is allegedly a Chinese spy; as a result, Megaspeed obtained chips worth about $7.5 billion, which, the claims hold, were smuggled into China. These are allegations, not facts established by a court.

What It Means in Practice

A new CEO may bring more transparency. But that same transparency could also expose complex debt structures and verbal agreements that no one knows about right now.

For developers and founders, the conclusion is down to earth. If your strategy relies on a single GPU source, add a fallback: clouds with TPUs or other accelerators, rentals from several providers, optimization for different platforms. If you raise money from AI startups financed by Nvidia, keep in mind that their purchases partly depend on the same circle of money. A limitation of the piece: this is the Guardian's analytical text, which relies on opaque data rather than disclosed financial reporting, so the scale of the consequences remains an estimate.

Story

Key Questions Answered

What are the key details on Not Just Chips?

On paper, Nvidia's value is simple to explain: the company designs graphics processing units (GPUs), without which modern AI cannot be trained or run. But, as the Guardian notes, the chips themselves are replaceable. Microsoft, Meta, and Google build their own accelerators, and Google's TPUs, by the paper's assessment, are in some cases no worse than Nvidia's solutions and for certain tasks are better. What makes the company irreplaceable is its network of relationships, much of which Huang built personally.

What are the key details on The AI "Central Bank" and Circular Money?

The second pillar is customer financing. Nvidia has backed so-called neoclouds: companies that buy its chips, install them in data centers, and rent them out to the largest AI companies. This supported demand and moved the risks of data center construction onto Nvidia's own balance sheet. In addition, Nvidia invested in startups that could become its customers and offered large sums to OpenAI to buy its own chips. All of this is measured in hundreds of billions of dollars.

What are the key details on TSMC and the Dark Spots?

The third pillar is Taiwan's TSMC, which makes the vast majority of AI chips and has no real competitors. The process is extraordinarily complex: circuit lines a few nanometers wide are patterned with lasers in extreme ultraviolet. Even after two decades of refinement, the yield on the most advanced chips is about 60%, meaning 40% of output is defective. According to legend, Huang wrote a personal letter to TSMC's chairman as early as 1997.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Private AI Builders Community · Vibe Club

Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews

Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.

Join Vibe Club →Explore Hamidun Academy
What do you think?
Loading comments…