Chip Titan

Bubble Talk, Big Orders, and Brave Forecasts

The world’s most valuable public company Nvidia just released blockbuster earnings, cooling down predictions of an AI stock bubble.

Fiscal third quarter results beat lofty expectations with revenue of $57 billion (up 62% year‑on‑year) and net income of nearly $32 billion. Ambitious guidance of $65 billion in revenue for the final quarter also took the market by surprise, sending shares up 5.5% after hours.

Bubble fears: While some AI leaders have admitted to frothy valuations, Nvidia CEO Jensen Huang said he sees “something different”, calling AI “revolutionary” and his company “unlike any other accelerator”.

Why Is Nvidia Worth So Much?

Nvidia doesn’t manufacture chips, yet it’s valued at over $4.5 trillion — roughly the GDP of Germany, the world’s third-largest economy. The reason: design beats assembly.

• TSMC: Taiwanese chipmaker bears the capital costs of producing for others. It must serve clients fairly, limiting pricing power. Margins per chip are far below Nvidia’s.
• ASML: Dutch maker of lithography tools dominates in its niche, etching complex patterns on silicon. But it sells machines on long cycles — steady growth, but not explosive.

Both companies are vital to the overall chip supply chain, but they occupy the lower‑margin side of it. Kind of like Foxconn building iPhones for Apple.

Lesson: In tech, the company with the design and direct customer ties wins the margin war.

The Hidden Advantage Behind Nvidia’s Chips

Historically, the biggest profits go to whoever controls the standard — think Windows in the 1990s or iOS in the 2010s.

In AI, Nvidia’s programming layer CUDA has become the standard, and the company has been quietly building it since 2006. CUDA is the “operating system” for AI hardware. Developers write for CUDA, not for GPUs in general.

That means switching to AMD or Intel would require rewriting code and retraining engineers, a costly slowdown. Nvidia isn’t just selling chips; it’s selling the entire AI infrastructure system. In investing terms, this kind of durable edge is called a moat: the defenses that protect a company’s profits from rivals. Owning the software layer makes Nvidia’s moat deeper than hardware alone.

Nvidia Wins No Matter Who Loses

Big tech firms are locked in an AI arms race, each pouring billions into data centers to power their own AI models or those they back. That spending in high-end chips flows straight into Nvidia’s pocket.

• Wealth transfer: These “hyperscalers” fund the war, Nvidia supplies the weapons.
• Winners vs. Always Winner: If Microsoft gains, Alphabet falls behind, but Nvidia profits either way.
• The payoff: Demand for Nvidia’s chips is so intense that customers are willing to pay almost any price, giving Nvidia unusually high profits. Demand for advanced Blackwell systems is “off the charts”, the CEO said.

Weak Spots Behind Nvidia’s AI Surge

Nvidia’s soaring valuation has investors debating whether its rise is built on a solid foundation. 

Key concerns:

• Customer concentration: 61% of revenue tied to just four clients — likely big cloud providers — leaving Nvidia exposed to their fortunes.
• Circular deals: Heavy investments in customers like OpenAI ($100 billion) and Anthropic ($10 billion) blur independence and sustainability.
• Chip rentals: Spending $26 billion (doubling from Q2) to rent back its own GPUs from cloud partners raises questions about genuine demand.
• Competitive pressure: Alphabet is already designing its own chips, with Microsoft and Amazon preparing their own too, so Nvidia’s dominance may not last.