[New York Diary] The 'AI Bubble Theory' and the Nvidia Dilemma

The stock price of Nvidia, the biggest beneficiary of the generative artificial intelligence (AI) boom, is faltering. Although Nvidia's revenue and net profit for the second quarter of this year (May to July) both exceeded Wall Street's expectations, its stock price plummeted 6.4% the day after the earnings announcement. Two days later, as overall investor sentiment revived due to easing inflation in the U.S. and expectations of a rate cut in September, Nvidia's stock rose about 1.5%, but it still has not surpassed the $3 trillion market capitalization mark. Why did the formula that better-than-expected earnings lead to stock price increases not apply to Nvidia?

[Image source=Yonhap News]

First, both the second-quarter earnings and the third-quarter revenue guidance fell short of the 'whisper number.' The whisper number refers to the actual earnings estimates within securities firms that are not officially disclosed to the public. Also, although Nvidia announced plans to launch its latest AI chip, 'Blackwell,' in the fourth quarter, it only mentioned expectations of generating billions of dollars in revenue for that quarter without providing specific forecasts, which is also seen as a factor weighing on the stock price.

At its core, concerns about an 'AI bubble' that intensified starting in the second quarter are present. Analysts argue that while AI companies are investing astronomical amounts, there are limits to the profits that can be generated from AI. Venture capital firm Sequoia Capital recently released a report stating that AI companies need to generate $600 billion in annual revenue to recoup their investments in the related industry. Although Nvidia posted over $30 billion in revenue in the second quarter, not all companies are generating meaningful profits. For example, OpenAI, the developer of ChatGPT, is projected to have annual revenue of only $3.4 billion this year. Ultimately, concerns remain about whether the AI arms race among big tech companies will translate into viable AI revenue models. This is why some argue that Nvidia's earnings should have not only exceeded market expectations but significantly outperformed them.

There are also concerns that the business 'moat' Nvidia has built over time could be eroding, which adds pressure on its stock price. Since 2006, Nvidia has provided a software program called 'CUDA' to facilitate AI-related service development using its graphics processing units (GPUs). Developers accustomed to creating AI programs with CUDA complain about the difficulties in designing products when using GPUs from other companies that do not support CUDA. This is why Nvidia dominates the GPU market as the overwhelming number one player. The problem is that cracks are appearing in this Nvidia monopoly. As Nvidia GPU prices have soared several times over, more companies are opting to use GPUs from other firms like AMD, which are cheaper than Nvidia's products, even if it means hiring more developers. Startups like Rebellion, as well as big tech companies like Microsoft (MS), are accelerating their own AI chip development.

If various paid services based on generative AI gain popularity and become essential daily commodities like smartphones, Nvidia could not only reclaim the top spot in global market capitalization but also solidify its status as an overwhelming AI superstar. The important point is that Nvidia's fate is likely to depend as much on how rapidly the AI market grows as on Nvidia's technological capabilities. Despite more than a year and a half having passed since the launch of ChatGPT, there are many criticisms that there has been little change in productivity for both companies and individuals. Although Nvidia is the biggest beneficiary of the generative AI market's emergence, there are limits to its ability to lead AI market expansion, so investors' concerns about Nvidia are likely to continue.