MW The blogger who helped spark Nvidia's $600 billion stock collapse and a panic in Silicon Valley
By Gordon Gottsegen
Jeffrey Emanuel says Wall Street banks that are bullish on Nvidia 'have absolutely no idea what they're talking about'
Last Friday afternoon, Jeffrey Emanuel sat down in his Brooklyn apartment and started writing a blog post. For hours, he pounded away on his keyboard while his wife kept their young children occupied and brought him food. Emanuel worked late into the night, and by early Saturday morning he had written nearly 12,000 words.
Emanuel's manifesto made the case for shorting the hottest company in the stock market, Nvidia Corp. $(NVDA)$, due to a number of shifting tides in the artificial-intelligence world, including the emergence of a China-based company called DeepSeek. He published his thesis on his personal blog and then shared it with the Value Investors Club website and across Reddit, X and other platforms. When he checked his blog's analytics later on Saturday morning, Emanuel saw that 35 people were reading the post. Not bad traffic for a personal blog built into the website of his YouTube transcription-service side project, he thought.
But then the post started to go viral.
By Saturday night, Emanuel could see that 1,500 people across the world were reading his blog post at a given moment. Well-known venture capitalist Chamath Palihapitiya shared Emanuel's post on Nvidia's short case with his 1.8 million X followers. Successful early stage investor Naval Ravikant shared the post with his 2.6 million followers. Jared Friedman, a partner at venture-capital firm Y Combinator, referred to it in a post that was reposted by the official Y Combinator account. Morgan Brown, a vice president of product and growth at Dropbox, pointed to it in a thread that was viewed over 13 million times. Emanuel's own X post got nearly half a million views. He also quickly gained about 13,000 followers on the platform, going from about 2,000 to more than 15,000 followers.
In an interview with MarketWatch, Emanuel said that at one point the traffic crashed his website, so people started sharing an archive link, which his website-analytics tool couldn't track.
But one thing it did pick up was that by the end of the night, the city with the most concurrent readers was San Jose, Calif. - near where Nvidia's corporate headquarters is located.
Emanuel's argument shook Silicon Valley not because he claimed that the big U.S. technology companies were misleading or deceitful. His main point was simply that they were nowhere near as smart and efficient as Wall Street was touting them as. The big tech companies had built and trained their artificial-intelligence breakthroughs using tremendous amounts of data and advanced compute resources that required them to pay for Nvidia's data-center hardware, which is sold at very high gross margins. Emanuel pointed out that a China-based company, DeepSeek, had recently launched its own top-notch AI product using fewer expensive chips. In other words, DeepSeek had achieved what the big AI companies had, but with far less money. What countless Wall Street firms and investment analysts had seemingly missed was being pointed out by some guy in his apartment.
Then on Monday things got real. Nvidia's stock plummeted about 12.5% at market open and continued falling from there. By the end of the day, the slide had wiped out nearly $600 billion from Nvidia's market capitalization - the largest single-day market-cap drop to date for any company. Matt Levine, the prominent Bloomberg News financial columnist, noted the online chatter that claimed Emanuel's post "was an important catalyst" for the stock-market selloff and said it was a "candidate for the most impactful short research report ever."
Emanuel spent the rest of the week booked solid as hedge funds paid him $1,000 per hour to speak on the phone and give his take on Nvidia and AI.
"I'm so exhausted, I'm losing my voice practically," said Emanuel. "It's been the most surreal experience of my life."
Contrarian call
As Emanuel noted in his blog post, "The Short Case for Nvidia Stock," he has professional experience working in financial markets. A native of New Rochelle, N.Y., Emanuel studied math at Reed College in Portland, Ore., before heading to Wall Street, where he worked at several investment funds, including as an analyst at Millennium Management and Balyasny Asset Management, two of the biggest multimanager hedge funds.
But Emanuel, 42, also told MarketWatch that he's been obsessed with neural networks since 1998 and was an early adopter of both crypto and AI. In 2021, Emanuel ditched Wall Street and started Pastel Networks, a blockchain company that provides decentralized storage, AI solutions and other services for Web3 developers. But he continued to carefully follow developments in Silicon Valley and the stock market. This experience in both the investing and the tech worlds led him to conclude that Nvidia is overvalued.
Last Friday he was chatting with a friend who works at a hedge fund about why he thought Nvidia's days of outperformance were numbered. This was a few days after DeepSeek released its R1 model and nobody on Wall Street seemed to notice.
"Every single bank has a super-bullish buy rating on Nvidia. It's like the blind leading the blind - they have absolutely no idea what they're talking about," Emanuel told MarketWatch. "All of their arguments have become totally divorced from reality."
MarketWatch checked this claim and found that 61 out of 67 analysts rated Nvidia as a buy as of the morning of Jan. 31. Six analysts, including one from Deutsche Bank, gave the stock a hold rating. None gave it a sell rating.
"They try to defend their arguments by saying, 'Well, we talk to industry experts.' But that's like asking the barber if you need a haircut," Emanuel said.
Emanuel's argument ran counter to the bullish Wall Street sentiment that has surrounded Nvidia during its epic run. Nvidia has helped drive the entire U.S. stock market over the past year. In 2024, it became the largest U.S. company by market cap and one of the most actively traded stocks by retail investors, and its 171.2% gain helped lift the S&P 500 SPX by 23.3% that year.
Throughout this time, Nvidia investors were pricing in that it would be one of the - if not the single - main beneficiary of the AI craze. The market saw big tech companies like Microsoft $(MSFT)$, Meta $(META)$ and Google parent Alphabet $(GOOGL)$ $(GOOG)$ spending hundreds of billions of dollars to gobble up Nvidia's hardware to build their AI data centers, and it expected that money to keep flowing as long as Nvidia's GPUs were better than the competition's.
The short case
The gist of Emanuel's argument is as follows:
Some of the most influential tech companies have determined that deep learning and AI are the biggest technological advances since the advent of the internet. In order to integrate that technology into their businesses, those companies have to build and train their AI, which takes a lot of data and compute resources. Nvidia sells the key hardware these companies need, and the margins on its most sophisticated chips are enormous.
But a few things are changing that are proving this might not be sustainable, Emanuel says.
For one, AI companies have been using scaling laws that essentially say the more data that is used to train an AI model, the better it gets. But Emanuel wrote that the industry may be running low on quality data to train that AI - that is, a potential "data wall" is looming that could slow down AI scaling and reduce some of that need for training resources.
Emanuel also posed the question of what happens to the training hardware after the AI is trained. GPUs are constantly getting exponentially better, so after a few years, companies might not want to use old hardware anymore. This puts them on a cycle where they're always spending more to get the best hardware. But eventually, those companies are going to want to see a return on their hefty investment.
Some of these companies, like Alphabet, have also been investing in building out their own semiconductor chips. For a while, Nvidia's hardware has been the best for training AI, but that might not be the case forever as more companies, such as Cerebras, build better hardware. And other GPU makers like Advanced Micro Devices $(AMD)$ are updating their drivers software to be more competitive with Nvidia.
On top of that, some new AI models are proving to be much more resource-efficient. This is where all the drama with DeepSeek comes in. DeepSeek launched its own AI that's on par with the likes of OpenAI's ChatGPT, but the real kicker was that it said it trained its AI in less time using fewer chips.
Read more: Does DeepSeek spell doomsday for Nvidia and other AI stocks? Here's what to know.
Add all these things together - unsustainable spending and data-center building, less training data to work with, better competing hardware and more efficient AI - and you get a future where it's harder to imagine Nvidia's customers spending as much as they currently are on Nvidia hardware.
"By the time I finished writing the article, I said, 'I'm convinced,'" Emanuel told MarketWatch. "It was when I realized that every one of their big hyperscaler customers was literally making their own competitive silicon, all made by [Taiwan Semiconductor] $(TSM)$, and that was already coming out and was imminently going to hit the market. I was thinking, 'Do people realize this?' Because I don't think they do."
If training and integrating AI becomes significantly cheaper, why would these big tech companies keep spending obscenely large sums of money?
For Emanuel, this called into question why Nvidia was trading at such a high price-to-earnings ratio.
"If you know that a company will only earn supersized returns for a couple years, you don't apply a multiple. You certainly don't put a 30-times multiple," Emanuel told MarketWatch.
(MORE TO FOLLOW) Dow Jones Newswires
January 31, 2025 16:56 ET (21:56 GMT)
Copyright (c) 2025 Dow Jones & Company, Inc.