After breaking through the 700 mark, NVIDIA finally fell. "Can't go up every day?"

After breaking through $700, can NVIDIA continue to rise?

Overnight, after a wave of profit-taking, NVIDIA's stock price fell back during the session after breaking through the historical high of $700, ultimately closing down 1.6%.

Previously, benefiting from the positive news of target price upgrades by major banks such as Goldman Sachs and Bank of America, as well as the partnership between NVIDIA and Cisco, NVIDIA's stock price rose more than 4% on Monday, marking the third consecutive day of reaching a new all-time high.

As of Monday, NVIDIA's stock price has risen by about 30% in the past month, far exceeding the monthly gains of the S&P 500 and Nasdaq.

Downward correction after "severely overbought"

However, along with the surge in stock price comes the risk of high premiums.

According to FactSet data, NVIDIA's 14-day relative strength index (RSI) has reached 85, falling into the "severely overbought" category. RSI measures the magnitude and speed of price changes, and any value above 70 indicates that the stock is overbought, pushing the stock price to a level that is not supported by fundamental factors. Generally, after being overbought, stock prices tend to undergo a downward correction.

The surge in stock price has also resulted in NVIDIA's valuation being much higher than other chip stocks. As of this Monday, NVIDIA's trailing P/E ratio reached 31.4 times, far exceeding the industry average of 22.9 times.

Therefore, Rob Ginsberg, an analyst at Wolfe Research, pointed out in a letter to clients that investors should not "chase a 90-degree rise". In this situation, he would rather choose to take partial profits.

Dominant position facing challenges

Barclays stock analyst Sandeep Gupta issued a "bearish report" on NVIDIA last week.

In the report, Gupta pointed out that the booming artificial intelligence arms race will eventually come to an end, and once the database is established, user deployment will not require much computing power. This means that the hardware market for artificial intelligence will undergo a "reshuffle", and NVIDIA's 98% market share in GPUs may not be guaranteed.

Gupta stated:

"Once the initial training is completed, the chip demand for artificial intelligence will eventually normalize. The inference stage of artificial intelligence requires less computing power than the training stage; high-performance PCs and smartphones are sufficient to run inference locally, reducing the demand for GPU computing centers.""We believe that the cost of building large language models (LLMs) and the limited monetization opportunities may eventually force Nvidia's small but still significant customers to exit the AI market to alleviate the GPU supply constraints."

The report also mentions that currently, NVIDIA's main revenue comes from tech giants. However, after the latest earnings report, the growth of revenue from tech giants is expected to slow down. Meta and Amazon are also developing their own AI chips, which could pose a potential threat to NVIDIA.

According to the report, in the last quarter, 46% of NVIDIA's revenue came from five of the "Big Seven" tech stocks, with Microsoft and Meta alone accounting for 28%.

Gupta believes:

"Given NVIDIA's sales concentration, the company is highly sensitive to fluctuations in revenue growth from large tech companies and the purchasing decisions of top customers."

"Considering the competition in the chip market mentioned earlier, as well as the fact that NVIDIA's customers are starting to develop their own chips, NVIDIA may lose market share, which is particularly important."

There is also a potential risk that the practice of using NVIDIA chips as collateral for financing is becoming more common among small customers.

For example, CoreWeave, a cloud computing startup supported by NVIDIA, used H100 as collateral to secure $2.3 billion in financing while using NVIDIA chips. This financing was used to purchase more advanced chips, which are likely to come from NVIDIA as well.

The question is, how long can this "Ponzi game" based solely on the dominance of H100 chips last?

What is known now is that in the past two quarters, an increasing share of NVIDIA's revenue seems to come from startups in which the company has invested.

However, before accepting this argument, it is important to note that the Sandeep Gupta team has been focused on fixed income rather than equity assets such as stocks for many years. The reference value of the above points needs to be evaluated. Moreover, the team still has a positive outlook on NVIDIA and maintains a target price of $650 in the report.