As one of the absolute stars leading the surge in US stocks and one of the core beneficiaries of the generative AI wave, NVIDIA has skyrocketed by 245% year-to-date, with a market value exceeding $1.2 trillion. However, the recent upward momentum of NVIDIA seems to be facing resistance.
After releasing its second-quarter earnings report, NVIDIA's stock price did not continue to soar as it did after the first quarter, reflecting the continuous increase in market expectations over the past period and the market's consideration of many uncertainties in the future.
Analysts from CITIC Securities, such as Chen Junyun and Xu Yingbo, believe that as a typical semiconductor company, NVIDIA will still demonstrate obvious cyclical characteristics in the medium and short term. It will benefit from the tight supply-demand structure caused by TSMC's CoWoS production capacity constraints. In the short term, NVIDIA's stock price is expected to fluctuate upwards. The current market value of $1.2 trillion corresponds to a forward P/E ratio of 30 times, which is expected to become an important bottom support.
<h2>NVIDIA is essentially a cyclical stock</h2>
Analysts pointed out that although NVIDIA's major financial indicators such as FY24Q2 revenue and profit continue to exceed market expectations, with data center revenue reaching $10.3 billion compared to the market's expectation of $8 billion, and the company guiding Q3 revenue to $16 billion, significantly higher than the previous market expectation of $12.5 billion, the stock price performance has been relatively calm.
They believe that NVIDIA's gaming graphics cards and data center combined revenue account for nearly 95% of the total revenue, and both businesses have significant cyclicality. NVIDIA should be regarded as a cyclical stock for analysis. Its stock price performance corresponds to market expectations. Compared with long-term performance forecasts, short-term performance improvement and trend inflection point analysis are more meaningful for judging the stock price.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0102cb64-4986-4762-89bd-b787324f311e.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0102cb64-4986-4762-89bd-b787324f311e.png"/>
Specifically:
<blockquote>
<ol>
<li>
Gaming graphics cards: In the recent cycles of 2018/19 and 2022, the company's graphics card business experienced significant declines, mainly triggered by factors such as the collapse of cryptocurrency market prices and insufficient downstream demand, which ultimately manifested as a rapid deterioration in the market supply-demand structure, closely synchronized with the global semiconductor cycle.
</li>
<li>
Data center: It mainly includes GPU, networking, CPU, etc., with networking and CPU mainly sold in conjunction with GPU. The downstream customers of the GPU business are relatively concentrated, with global cloud computing and technology giants accounting for more than 80% of purchases. GPU demand is mainly influenced by the AI technology cycle, giant capital expenditure cycle, and its own product cycle.
</li>
</ol>
</blockquote>
Among them, the data center business follows three major cycles: technology, cloud service providers' CAPEX, and product cycles:
<blockquote>
Technology cycle: NVIDIA's data center GPU revenue mainly comes from AI training, and it relies on its powerful product capabilities, complete software stack, and good product versatility to achieve market leadership. Therefore, the demand for NVIDIA GPUs depends on the rapid evolution and iteration of AI algorithms, and the innovation cycle of AI technology constitutes the most important variable affecting the demand for NVIDIA GPUs.
Cloud Provider CAPEX Cycle: According to data from third-party organizations such as Trendforce, the four major cloud providers in North America (Microsoft, Google, Meta, Amazon) currently account for approximately 65% of global AI server purchases. When combined with several major domestic internet and cloud computing giants, as well as secondary cloud providers in North America, we expect the overall purchase proportion to exceed 80%. It is known that, excluding factors such as the AI technology cycle, cloud providers' capital expenditures in the data center field typically follow a pattern of two big years followed by one small year, repeating in this cycle.
Product Cycle: NVIDIA's GPU architecture is generally updated every two years. The performance, price, and other factors of the new product are also the main considerations for customers when deciding whether to replace existing equipment or make additional purchases.
</blockquote>
<h2>Future Performance Outlook</h2>
Analysts believe that in the short term, NVIDIA's performance is expected to improve quarter over quarter, but there will be increased disturbances in the medium term.
<blockquote>
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/37f3d2aa-3d71-4a65-a3d8-7b20288eabb9.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/37f3d2aa-3d71-4a65-a3d8-7b20288eabb9.png"/>
<ol>
<li>
Currently, sales of NVIDIA's data center acceleration cards are still mainly dominated by the high-end A100/H100 series cards. The main constraint is the capacity of TSMC's CoWoS process. Considering factors such as the delivery cycle of packaging and testing equipment, we expect substantial improvement in TSMC's CoWoS capacity to occur around 2024Q2 at the earliest. The short-term tight supply is likely to result in downstream customers placing duplicate or multiple orders. In the short term, in terms of order inflow, delivery volume, price, and other aspects, NVIDIA's data center business is expected to remain strong and continue to show a quarter-over-quarter upward trend.
</li>
<li>
Following the Gartner Technology Hype Cycle, technologies typically go through a typical development cycle of rise, hype, decline, and resurgence. This current wave of generative AI is likely no exception, and market disturbances and fluctuations in stock prices during this period are difficult to avoid. However, in the medium to long term, NVIDIA's comprehensive layout and competitive strength in the field of AI are expected to continue benefiting from the global AI industry's development.
</li>
</ol>
In the short term, the tight supply-demand structure is expected to provide continued support to NVIDIA's performance in terms of orders, delivery volume, price, and other aspects. The pace of TSMC's CoWoS capacity release will be a key observation point. In the medium term, the development of the AI industry will gradually surpass its hype cycle, and market disturbances and fluctuations in stock prices during this period are difficult to avoid. However, NVIDIA's comprehensive layout and competitive strength in the field of AI are expected to continue to share the benefits of AI industry development.
</blockquote>
Looking ahead to NVIDIA's future stock performance, analysts point out that in the medium term, the stock price is expected to find support around a valuation of approximately 30x PE, which can be considered as a bottom support level for the stock price.
As a typical cyclical stock, NVIDIA's valuation has shown significant fluctuations over the past decade, which is consistent with the characteristics of general cyclical stocks. However, even at the trough of the cycle, its valuation level is significantly higher than the industry average of the semiconductor sector. Our statistics show that its dynamic valuation has generally remained above 30x PE. Since 2018, with the continuous increase in the proportion of revenue from its growing data center business, its valuation center has also improved significantly. We believe that its significant valuation premium compared to the Philadelphia Semiconductor Index mainly reflects the market's recognition of the certainty brought by a favorable competitive landscape and the high growth potential brought by AI.
Based on the analysis above, we expect that in the medium term, NVIDIA's stock price is expected to find support around a valuation of approximately 30x PE, which can be considered as a bottom support level for the stock price. Based on top-down analysis (mainly considering the capital expenditure plans of North American cloud vendors) and bottom-up aggregation (mainly analyzing customer orders), according to our model calculations, we predict that the company's FY2025 net profit (Non-GAAP) will be $40.88 billion, while the current Bloomberg consensus estimate is $40.66 billion. Therefore, in the short to medium term, we believe that NVIDIA's market value is expected to find support around $1.2 trillion (corresponding to approximately 30x PE for FY2025).

NVDA

CITIC Securities believes that in the medium and short term, generative AI will contribute to the explosive growth of short-term performance in the English market. However, NVIDIA remains a typical cyclical stock, and it is necessary to closely observe its upward momentum in short-term performance and the inflection point of its performance cycle.

CITIC Securities: NVIDIA's "monopoly" will last at least until Q2 next year, with a 30x forward P/E ratio supporting the stock price.