NVIDIA quarterly report fraud, inflated stock price? Professional Wall Street analysts refuted: Don't go online blindly
Recently, conspiracy theories about NVIDIA's "record-breaking" earnings report for the second quarter are circulating on social media and investor forums, causing shares to fall by about 3% on September 6 and 7, erasing all gains since the release of the second quarter report after hours on August 23.
Nvidia's poor performance dragged down the chip industry benchmark, the Philadelphia Semiconductor Index, fell more than 3% at the beginning of the session on Thursday, its competitor AMD in the GPU field fell nearly 4% at one point, and large technology stocks such as Apple, Microsoft, and Google that benefited from the AI boom also continued to be sluggish.
The reason for NVIDIA's price decline, in addition to the logic of "oil prices rose to the highest in nearly ten months, and the US economic data remained resilient, which raised the market's expectations of another Fed interest rate hike in November, which in turn drove US bond yields up continuously, putting pressure on risk assets such as the stock market", and the long-short battle on Wall Street, which hit a record high after its earnings report and rose more than 200% during the year, was also particularly fierce.
Nvidia's costs did not increase, but revenue doubled year-on-year? Channel surveys show that companies have not significantly increased AI
Capital expenditure?
According to research and investment firm Macrotips Trading, although NVIDIA's second-quarter report far exceeded expectations and gave stronger guidance for future revenue, there are reasons to be concerned about "questionable disclosure and management practices," especially as an AI startup backed by the company's direct investment participated in driving revenue growth, and red flags from gross margin expansion, odd timing of share buybacks and insider sell-offs.
In the second quarter, Nvidia's total revenue doubled 102% year-over-year to $13.5 billion, beating analysts' expectations by $2.4 billion, the agency wrote. Data center revenue increased 171% year-over-year and 141% sequentially to $10.3 billion, $2.3 billion higher than market expectations, given strong demand for GPUs that support large language models and generative AI:
"What's even more incredible is that NVIDIA has achieved phenomenal growth with almost no cost addition, with its cost of revenue increasing by only 6.8% year-over-year and gross margin reaching an unheard of 70.1%.
Nvidia also set an ambitious target of $16 billion for third-quarter revenue, well above analysts' expectations of $12.5 billion. With a market capitalization of $1.2 trillion, Nvidia's quarterly reports of such a large number of companies have never exceeded market expectations by such a large margin.
It wonders why Wall Street's expectations have not bet on its rapid growth at all. It stands to reason that the more than 40 analysts covering the company had conducted and analyzed every conceivable channel check in at least 10 ways before the earnings report.
For example, before the earnings report, Denmark's Saxo Bank analyzed the second-quarter capital expenditures of major AI players and found no signs of high capital spending that would indicate such strong demand for NVIDIA GPUs. ”
This raises questions about how NVIDIA's data center revenue boom, based on the surge in demand for generative AI, did it. Some market participants immediately set their sights on Nvidia-backed AI startup Coreweave.
Conspiracy theorists question Coreweave, which Nvidia invests
Assist in thickening the second quarter revenue, and more large manufacturers may also conspire!
According to the article, there is reason to suspect that Coreweave, which is directly funded by Nvidia, is the main reason behind Nvidia's surge in revenue.
First, Coreweave sold $421 million to a number of investors, including Nvidia, which could give Nvidia control over the former's purchase decisions. At the same time, Coreweave completed a debt financing of $2.3 billion by pledging NVIDIA GPUs in early August, which is exactly the scale of Nvidia's data center second-quarter revenue including AI chip business that exceeded market expectations:
"While this may be perfectly legal, Nvidia funded the startup, which in turn purchased the GPUs through a loan from mortgaged Nvidia GPUs, which required Nvidia to justify better disclosure of its relationships with its customers."
There are also skeptics who believe that Coreweave, the AI technology company that has placed billions of dollars in orders to Nvidia, may not even be a real company, but a shell institution created and supported by Nvidia's largest shareholder, asset management giant BlackRock, in partnership with Nvidia.
Some people say that NVIDIA also conspires with "big manufacturers" such as Microsoft, Amazon AWS and the world's third-ranked large-model startup Inflection, and major industry players have placed more orders for NVIDIA, and these orders are problematic in order to cooperate with NVIDIA to do revenue:
"There is an incredible theory gaining traction on X/Twitter that NVIDIA's amazing earnings performance this year was driven by selling fake GPUs to fake and non-existent customers.
Some have begun to draw parallels between the incident and the FTX pyramid scheme fiasco, fearing that such a scam could bring down the entire AI/chip market. ”
Analysis: Conspiracy theories prove that the NVIDIA multi-short game is very glued, and the bottom layer is worried about AI
The bubble burst and brought down the entire market
But Stacy Rasgon, an analyst at brokerage Bernstein, expressed strong dissatisfaction with the conspiracy theory, saying bluntly in its latest research note this week: "Customers, please don't believe the naïve investment arguments that appear casually on social media." ”
As for the skepticism that Nvidia's revenue doubled year-over-year and cost of sales increased by only 7%, Rasgon said that this was because Nvidia's second-quarter expenses included about $1.22 billion in inventory reserves, which were included in cost of sales in the second quarter of last year, and "when these expenses are excluded, the cost of sales in the second quarter actually increased by about 70% compared to a year ago, which is completely normal." ”
The second mainstream question, involving Coreweave, which transitioned from North America's largest Ethereum miner to GPU cloud provider, Rasgon, cautioned that the company mortgaged NVIDIA H100 chips in exchange for Blackstone-led debt financing, not Blackrock, NVIDIA's majority shareholder, and that this misinterpretation of the conspiracy theory is comical enough, and:
Nvidia didn't need help from Coreweave (or anyone else) to spur second-quarter sales, and Nvidia's products were already allocated.
Moreover, the debt agreement was announced after the end of the second quarter on August 3, and the debt issuers hinted that the deployment of funds may not have taken place yet. (How can this falsely thicken Nvidia's earnings report in the second quarter?) ）”
Some analysts pointed out that before the August financing, Coreweave's own funds were only $371 million, how can a company of this size have the ability to jointly operate with NVIDIA, whose data center revenue exceeded $10 billion in the second quarter:
At present, there is no direct evidence pointing to the two games, and the only criticism is that NVIDIA specifically supported the companies it invested in when production capacity was scarce.
Supporters of Nvidia generally believe that the conspiracy theories prevalent on social media only illustrate the fact that the current long-short game against Nvidia is very glued. The underlying logic may be a deep fear and concern about the idea that "AI fuels the next tech bubble burst".
Rob Arnott, a well-known figure in quantitative investing, said this week that Nvidia is the leader of the AI stock bubble that is about to burst, and that it could drag down the entire market after being punctured, and that NVIDIA's surge this year is a textbook example of a "major market illusion":
"Keep in mind that whenever a bull market emerges in a strong narrative that drives only a small subset of popular companies' share prices up, those stocks can disappoint in the coming years. This paradigm played out after the tech bubble of 2000 and the financial crisis of 2008.
There is no doubt that artificial intelligence will change our world. But Nvidia's share price suggests that the market's belief that it will not be challenged in the future by new entrants in the industry or regulators questioning its dominance is beyond reasonable reality. ”
What are the doubts of those who question NVIDIA's falsification of second-quarter earnings? The timing of share buybacks and high valuations are unsustainable
Going back to the questioning article at the beginning of the article, unlike the skepticism that deliberately exaggerates anxiety on social media, this analysis deeply analyzes NVIDIA's second-quarter earnings report and puts forward several points worth considering.
First, NVIDIA's gross margin over the past few quarters "expanded incredible, jumping from 43.5% in the second quarter of last year to 64.6% in the first quarter of this year, and expanded to 70.1% in the second quarter of this year, while the gross margin of competitor AMD remained stable year-over-year:
"Perhaps NVIDIA's gross margin expansion simply reflects the high demand for its GPUs in the hot AI and generative AI chip market, and customers are willing to pay double the price for NVIDIA chips, doubling their revenue without increasing costs too much."
But NVIDIA's revenue recognition rules for license and development arrangement need to be considered. Unlike product sales, which recognize revenue as a result of a transfer of control as a result of product shipments, a portion of data center revenue may come from software licensing, which is recognized in advance when the software is available to the customer, even if the customer actually pays later.
This policy of recognizing revenue early even before payments have been received could explain why NVIDIA's accounts receivable jumped $3 billion sequentially to $7.1 billion in the second quarter. Moreover, accounts receivable in the second quarter of the year benefited from $1.25 billion paid by customers before invoice due dates. All of this drove NVIDIA's overall quarterly revenue up $6.3 billion sequentially. ”
Second, the article also questions the timing of Nvidia's share buyback. The company's board of directors approved an updated $25 billion buyback program, building on the remaining $4 billion previously approved repurchase program pending execution at the end of the second quarter:
"The choice of NVIDIA's management is puzzling. It did not buy back any shares in the first quarter, 7.5 million shares in the second quarter for $3.28 billion, and 2 million shares for $998 million between July 31 and August 24 this year.
The calculation shows that 7.5 million shares are worth $3.28 billion, which equals an average repurchase price of $437, and 2 million shares worth $998 million, equal to an average share price of $499.
In the second quarter of Nvidia's fiscal year ended July 30, only a few days traded above $437, bringing the average price to $437, meaning the company had to buy back shares at the end of the second quarter, when the stock was near an all-time high. Similarly, between July 31 and August 24, the stock price exceeded $499 on August 24 alone, meaning that most of the buybacks must have occurred on this day.
(Ironically, Nvidia almost chooses to buy back when the stock price is close to its all-time high), but most companies buy back shares through plans such as volume-weighted average price over a longer period of time to avoid overly affecting the stock price. ”
Third, the article also questioned that NVIDIA insiders have been accelerating the sell-off at a time when the company continues to buy back shares at all-time highs:
Over the past six months, insiders have sold $234 million in shares, including CFO selling $2.3 million on August 28 and $2.5 million on May 30, and CEO Jensen Huang recently exercising stock options to sell $117 million on September 1.
Finally, the article claims that NVIDIA's current valuation is "sky-high," trading at a whopping 16.7 times the future enterprise sales value (Fwd EV/Sales) and 117 times the rolling P/E ratio:
"Even if analysts' predictions are correct that NVIDIA can increase revenue to $111 billion in fiscal year 2027 (calendar year 2026), the company's current enterprise value of $1.2 trillion is still more than 10 times EV/sales.
While AI and NVIDIA GPUs are likely to change the world as we know it, valuing NVIDIA at at least 10 times its projected sales over the next three years would make its bubble even more shocking than the dot-com bubble of 2000.
As the insider sell-off accelerates, the risks of owning Nvidia stock also appear to be rising. I personally don't feel comfortable holding a company whose valuation is built at more than 10 times expected revenue in three years. I wouldn't short Nvidia stock, but current holders should sell. ”
Nvidia executives and industry insiders aren't afraid: AI
The market size can reach 6000
Billion dollars, the trend is unquestionable
But Nvidia executives seem unafraid of skeptical voices in the market.
At a technology conference at Goldman Sachs this week, Manuvir Das, Nvidia's vice president of enterprise computing, predicted a long-term potential market for AI worth as much as $600 billion.
That will consist of $300 billion in chips and systems, $150 billion in generative AI software, and $150 billion in omniverse enterprise software, aided by Nvidia's "accelerated computing."
Das said Nvidia is capitalizing on an inevitable industry trend that eventually business operations will go digital and improve efficiency in ways that were previously unimaginable:
"The easiest way to look at Nvidia is that we made a big bet (and won) that we've been making for decades."
Tom Traugott, senior vice president of digital infrastructure strategy at digital infrastructure solutions provider EdgeCore, also said that critical infrastructure is changing under the AI boom and may have far-reaching impacts:
"Enterprises will have to improve existing systems to meet the demand for AI, which in turn will drive demand for NVIDIA chips that support the transformation of data centers to more efficient GPUs."
A company may have AI chips to distribute, but the next key question is whether they have the data center capacity to make those chips workable. AI is driving data center buildings to be bigger and denser than ever before. ”
Unlike online conspiracy theories, according to FactSet, of the 51 Wall Street analysts covering Nvidia, 47 have a "buy" rating, 4 have a "hold" rating, and no one has a sell rating, with an average price target of $649.22 and 42% room for gain.