The Worst Is Yet To Come – How And Why Nvidia Lost $279 Billion In 1 Day, The Biggest Loss In Wall Street History

The Worst Is Yet To Come – How And Why Nvidia Lost $279 Billion In 1 Day, The Biggest Loss In Wall Street History


September 6th, 2024 by financetwitter


Artificial Intelligence (AI) chip maker Nvidia has been minting tonnes of millionaires among its 30,000 employees. The nearly US$3 trillion company generated US$27 billion in revenue in its fiscal 2023. Then, something spectacular happened. Nvidia revenue for the 12 months ending July 31, 2024 skyrocketed to US$96.31 billion – a 194.69% increase year-over-year.

The 30-year-old tech company, whose specialized chips now account for 90% of AI-related chip sales, seemed unstoppable and invincible. Since 2019, Nvidia’s stock has soared 3,000%, thanks to Jensen Huang, co-founder and CEO of Nvidia. He received a hero’s welcome during a visit to his birth land of Taiwan to attend an industry event recently.

Born Jen-Hsun Huang in Taipei in 1963, Huang moved to Thailand at the age of five. He lived there for about 4 years before migrating to the United States. After graduating from Stanford University, he co-founded Nvidia in 1993 at age 30. He had “no idea how” to start a company, but his master’s degree in electrical engineering would make Nvidia the largest company in the world by market capitalization in June 2024.


At a time when Intel was synonym with CPU (central processing unit), the young engineer was toying with GPU (graphics processing unit), a technology that few believe in. Known simply as “graphic card”, the GPU was seen as a niche product – good for nothing except to improve video game graphics. Huang and Nvidia were well-known only among the gamers and computer graphics experts.

The company was 30 days from going out of business, before its RIVA 128 – a consumer GPU created in 1997 – saved the day. It was the first to integrate 3D acceleration in addition to traditional 2D and video acceleration. It may look incredibly obsolete at today’s standard, but at resolutions higher than 640×480, the RIVA 128 offered superior quality when played with games like Quake II.

Fast forward to today, Huang’s early vision – GPU could become the backbone of a new computer era – has become a reality. First, Nvidia became a powerhouse selling cards for video games when people were spending more time at home playing games during the 2019 Covid-19 pandemic, leading to demand for computer parts, including GPUs. Gaming revenue jumped 41% in 2021.


Then, came the artificial-intelligence (AI) boom, where Nvidia helped in powering OpenAI’s ChatGPT and other applications. From Tesla’s self-driving feature to major tech companies’ cloud computing services, and from streaming content on Netflix to cryptocurrency mining, Nvidia is essentially the Birkin bag of AI chips. Selling like hot potatoes, some clients chasing for Nvidia’s AI chips had to wait as long as 11 months.

As Jensen Huang unveiled his new vision of a computing era where AI empowers everyone to become a programmer, his net worth rose rapidly along with the value of Nvidia’s stock – from US$3 billion in 2019 to US$107 billion in June 2024. The company even surpassed Apple when its valuation reached US$3 trillion. Mark Zuckerberg wrote on Instagram – “He’s like tech’s Taylor Swift“.

At Nvidia’s Santa Clara office, the parking lots were packed with luxury cars like Porsches and Lamborghini, bringing back the good old memories of dot-com boom in the 1990s. In its last financial quarter, Nvidia reported a gross margin of 76%, which means it cost them just 24 cents to make a dollar in sales. Nvidia’s fans were arguing that AI would take over the world, pushing the stock price even higher.


But on Tuesday (Sept 3) this week, the stock suddenly crashed, dragging global semiconductor and even other technology stocks along. Nvidia market capitalization lost a whopping US$279 billion overnight – the biggest in Wall Street and the U.S. history (the previous record was held by Facebook, which suffered a $232 billion wipe-out in a day in February 2022).

Exactly how could Nvidia, whose graphic cards happen to be perfect for handling the computing load AI requires to perform its tasks so much so that Microsoft and Facebook-parent Meta both now spend more than 40% of their budgets on Nvidia gear, suffer such punishment despite having posted great earnings – record revenue, raised guidance and strong future prospects?

The stock took a hit precisely because the company has become a bellwether for the global economy. People were worshipping Nvidia and investors have the highest expectation from the stock. Heck, even Nvidia AI chips are being smuggled and sold in China (at 60% discount). However, because the company is priced for perfection, any sign of potential weakness will trigger nasty reactions.


During the earnings call, Mr Huang discussed a 3-month delay in the Blackwell chip’s production. Even though the CEO insisted it was not due to design flaws, investors and traders believed it was, and dumped the stock like a plague. Adding salt to the injury, research firm SemiAnalysis asserts Blackwell GPUs have been delayed due to advanced packaging problems at TSMC, Nvidia’s chip manufacturer.

It didn’t help that data center operators too are concerned, because they have been busy since March revising specs to get ready for liquid cooling systems required with Blackwell. The most technically advanced Blackwell chip called B200 requires higher power density of about 125 kilowatts compared to a standard for data center racks of up to 20 kiloWatts.

The design flaw was apparently discovered by manufacturer Taiwan Semiconductor Manufacturing Company (TSMC). Nvidia will now have to go back to the drawing board and re-design the chip before conducting further production tests with TSMC before it can move on to mass production. To address the delays, the workaround involves extending the life span of its former flagship line of chips, Hopper.


But even if there hadn’t been any design flaw to begin with, there have been doubts about the return on AI investment. Analysts at Goldman Sachs if the US$1 trillion investment in Artificial Intelligence by big boys like Microsoft, Google and Meta over the next few years will “ever pay off”. It was estimated that tech companies will need to earn US$600 billion to pay back their AI investments.

Sure, in addition to reporting 122% annual revenue growth to over US$30 billion, Nvidia said sales in the current period will jump about 80% to roughly US$32.5 billion, beating analysts’ expectation of US$32 billion. However, some were expecting US$34 billion because a powerhouse like Nvidia would have to dramatically surpass analyst estimates in its guidance in order to impress investors.

Assuming investors could close one-eye over the missing revenue expectation, Nvidia’s gross margin, which slipped a bit in the latest quarter to 75.1% from 78.4% in the prior period, has provided an excuse for some investors to dump the stock. What goes up must come down, therefore, some analysts think the company could have reached its peak already.


The fourth factor has to do with the U.S. Department of Justice, who reportedly had sent subpoenas to Nvidia and other companies to investigate whether the chipmaker violated antitrust law. The U.S. authorities are concerned with complaints that Nvidia has made it difficult for customers to switch chip providers due to its dominance in artificial intelligence chips.