The miraculous semiconductor boom sparked by the AI development competition. The reign of Nvidia, Samsung Electronics, and SK Hynix. How long will it last.

Nvidia has once again defeated market doubts with its earnings. (Early morning on August 27, Korean time) However, as you turn the pages of the earnings report, unexpected numbers that still give investors pause begin to appear one by one.

Nvidia's latest quarterly revenue reached a staggering $96.2 billion. That means it is earning more than 1.4 trillion won a day in Korean currency. In particular, the revelation that "next year we will earn 70% better than this" drew cheers from investors. By predicting that it will become a company earning about 2.6 trillion won a day in 2027, it greatly surpassed the market's expected growth rate for Nvidia next year, reigniting optimism. However, my eyes keep lingering on the "two numbers" that follow behind it. What on earth are these two numbers?

They are 00 and ㅁㅁ, which increased sharply compared to the previous quarter.

The growth rate promised by Nvidia in this earnings presentation is a level that cannot be achieved by just a few U.S. "hyperscalers." It is true that U.S. hyperscalers account for 55% of Nvidia's data center revenue. Even if the current AI development competition turns out to be self-defeating, they will not slow down that pace. However, for the "AI board" to proceed while justifying its current stock price, the remaining 45% must also be able to stably spend large amounts of money on expensive semiconductors without a single stumble. Yet, looking at these "two numbers," it is hard to deny that aspects stimulating investors' doubts—wondering "whether such a smooth flow will really be possible"—stand out. This means that concerns over "circular financing," where money in the AI ecosystem ultimately flows out of Nvidia and returns to Nvidia, were not dispelled by this earnings announcement.

CEO Jensen Huang has already presented his own "self-solution" to Nvidia's lingering concerns regarding these "two numbers" ahead of the earnings release. He plans to join hands with six major Wall Street financial giants to pull $500 billion of "fresh money" into the AI ecosystem. Larry Pink, the legendary CEO of BlackRock, the world's largest asset manager and one of these financial institutions, vowed that the investment model to be created hand-in-hand with Nvidia would be a "fantastic product." Much like mortgage-backed securities (MBS), which were fully developed in the 1970s. While the collateral for MBS is housing, the collateral for this "fantastic product" will be Nvidia's AI accelerators and the AI infrastructure built using them. Unlike housing, can semiconductor sets—where the spotlight quickly fades from past products every time a new one is released—become a "future financial collateral" that guarantees stable yet lucrative returns?

Nvidia has "proven" itself once again... [Smart E] examines the questions still left behind with the sense of balance needed right now, delivered by Reporter Kwon Aelee.

1. "This Number" Fuels Doubts Over "AI Sustainability"
2. Is "Fresh Money" Finally Flowing Into the AI Development Race?
3. A "Fantastic New Financial Product" That Collateralizes Data Centers?
4. Breaking Away From "AI Circular Financing" VS Risks Shifted to Financial Markets
5. "Turns Out We Don't Really Need Such Expensive AI Models?"

(Reported by Kwon Aelee | Video by Park Woo-jin and Hwang Se-hoe | Produced by Jung Seo-woo | Edited by Lee Ki-eun | Design by Ahn Jun-seok | Intern: Kim Hye-won | Production: Knowledge Contents IP Team)